USGS - science for a changing world

Scientific Investigations Report 2007–5050

Scientific Investigations Report 2007–5050
Version 1.1, April 2010

Back to Table of Contents

Ground-Water Hydrology

Ground water moves from areas where it enters the ground, known as “recharge areas,” to areas where it leaves the ground, known as “discharge areas.” On a regional scale, recharge areas are typically high-elevation regions with large amounts of precipitation compared to surrounding areas. Ground water moves from recharge areas toward low-elevation areas in response to gravity. In low-elevation areas, ground water typically discharges to streams, lakes, or wetlands and then is returned to the atmosphere through evapotranspiration or leaves the basin as streamflow. Ground water can be removed anywhere along its flow path by wells.

Ground water moves in response to differences in hydraulic head, from areas of high head to areas of low head. In unconfined aquifers—those without overlying low-permeability strata—hydraulic head can be thought of as the elevation of the water table. In confined aquifers, hydraulic head can be thought of as the elevation of the aquifer plus the pressure of the confined ground water. Maps of hydraulic head are useful for identifying recharge and discharge areas, and for indicating the direction of ground-water flow. Although the regional scale movement of ground water largely follows topography, the actual flow paths that the ground water follows and the rate of ground-water movement is controlled by the permeability of the geologic materials through which it flows. The rate of ground-water movement is proportional to the hydraulic head gradient and the permeability of the geologic materials.

Ground-water systems are dynamic, with rates of recharge and discharge and hydraulic head varying in response to external stresses. The largest external influence on ground-water systems is climate. Drought cycles cause large fluctuations in recharge, ground-water levels, and discharge to springs and streams. Human-caused stresses, such as pumping and artifical recharge from canal leakage and deep percolation of irrigation water, also affect the ground-water system.

Hydrologic Budget

The hydrologic budget is the accounting of water moving into and out of a hydrologic system such as the upper Klamath Basin. A general hydrologic-budget equation is:

PRECIP + SWin + GWin = ET + SWout + GWout + ?S,      (1)



is precipitation,


is the inflow of surface water,


is the subsurface inflow of ground water,


is total evapotranspiration,


is the out flow of surface water,


is the subsurface outflow of ground water, and


is the change in water stored in the surface- and ground-water systems.


These individual terms may have multiple components. For example, ET includes the evapotranspiration from forests, wetlands, and agricultural crops. In the upper Klamath Basin, the largest terms are PRECIP, ET, and SWout. There is no evidence of subsurface ground-water flow into the basin (GWin). Ground-water flow out (GWout) toward the south is probable, but the amount is likely to be miniscule compared to other terms in the equation. The storage term (?S) includes both surface storage in reservoirs and subsurface storage of ground water in aquifers. When dealing with long-term, multiyear averages, changes in surface water storage are commonly negligible.

Long-term changes in ground-water storage are manifest as year-to-year changes in the water-table elevation. Long-term observation well data indicate that a slight, climate-related decline has taken place in water levels in wells in the Klamath Basin since the 1950s. The magnitude of the decline varies spatially, but ranges from zero to about 10 ft over 50 years. Larger declines have been measured near pumping centers, but are generally geographically restricted. The amount of water represented by the annual change in ground-water storage is small compared to the overall hydrologic budget. For example, assuming the change in storage occurred in the shallow, unconfined parts of the system and using a storage coefficient of 0.05 (a reasonable number for an unconfined volcanic aquifer), a decline of 5 ft in 50 years averaged over the entire 8,000 mi2 upper Klamath Basin equates to an annual change in storage of about 26,000 acre-ft.

Some components of a hydrologic budget, such as streamflow, can be measured directly. Other components, such as evapotranspiration, are impractical or impossible to measure directly at useful scales and must be estimated or inferred from other measurements. This section presents a very general discussion of the hydrologic budget of the entire upper Klamath Basin (summarized in table 3) to provide a context for a more detailed discussion of the hydrologic budget of the ground-water system. All figures presented in this section have associated uncertainty.

Data from the Oregon Climate Center PRISM Group (, accessed September 20, 2006) indicates that precipitation in the upper Klamath Basin averages about 10 million acre-ft/yr (1971–2000 average). Of that amount, only about 1.5 million acre-ft/yr flows out of the basin past Iron Gate Dam (1961–2000 average; 1971–2000 average is 1.6 million acre-ft/yr). Most of the remaining 8.5 million acre-ft/yr returns to the atmosphere through evapotranspiration at the location where the precipitation falls. Some of the 8.5 million acre-ft/yr, however, returns to the atmosphere elsewhere in the basin after it has moved through the hydrologic system. An example of the latter case would be water diverted from streams or pumped from ground water that returns to the atmosphere through evapotranspiration from irrigated fields (this type of loss is often termed “consumptive use”). A small amount of water is exported from the basin. La Marche (2001) estimated water exports to the Rogue River Basin to average 0.027 million acre-ft/yr between 1960 and 1996

Burt and Freeman (2003) estimated that evapotranspiration from agricultural fields in the Klamath Project in 1999 and 2000 averaged 0.48 million acre-ft/yr (the 2001 estimates are not included here because of the cut off of irrigation water that year). Estimates by Cooper (2004) suggest that average annual agricultural consumptive use in the principal agricultural areas outside of the Klamath Project in Oregon (including the Williamson, Sprague, and Wood River subbasins) totals about 0.2 million acre-ft/yr. Consumptive use by ground-water irrigated agriculture outside of the Klamath Project in California (including areas in the Tule Lake and Butte Valley subbasins) is estimated to be about 0.072 million acre-ft/yr on the basis of data from the CDWR 2000 land use survey. Areas irrigated with surface water in California outside of the Project are small in comparison to ground-water irrigated areas and are not included in this total. There is also significant evapotranspiration from wetlands and open water in the upper Klamath Basin. Hubbard (1970) estimated that evapotranspiration from open water and the fringe wetlands of Upper Klamath Lake averaged 0.29 million acre-ft/yr from 1965 to 1967. Risley and Gannett (2006) estimated that evapotranspiration from the Tule Lake and Lower Klamath Lake refuges totaled about 0.22 million acre-ft/yr from 2003 to 2005. Average evapotranspiration in the Klamath Marsh area was estimated to be about 0.17 million acre-ft/yr using the method of Priestly and Taylor (1972) (Tim Mayer, U.S. Fish and Wildlife Service, written commun., 2005). Burt and Freeman (2003) estimated evapotranspiration from other surfaces in the Project area, including open water outside of refuges, urban areas, and undeveloped land to average about 0.082 million acre-ft/yr in 1999 and 2000. The source documents listed above all discuss uncertainty of the evapotranspiration estimates in general terms. Bert and Freeman (2003), however, provide a quantitative uncertainty assessment, and assign confidence intervals of plus or minus 14 to 20 percent for the estimates discussed here. Although the above list of evapotranspiration losses, which totals 1.5 million acre-ft/yr, is not exhaustive, it includes the bulk of consumptive uses in the nonupland parts of the basin. When this number is added to the streamflow out of the basin, about 7 million acre-ft/yr of precipitation (about 70 percent of the total precipitation) still leaves the basin through other avenues, principally as evapotranspiration to the atmosphere in upland areas.

A substantial proportion of the roughly 3 million acre-ft/yr that leaves the basin through streamflow or evapotranspiration in nonupland areas moves through the regional ground-water system. Equation 1 describes flow into and out of the entire upper Klamath Basin. Most flow into and out of the regional ground-water system occurs entirely within the basin. The hydrologic budget of the regional ground-water system can be described by the equation:

RECH + GWin = GWdis + GWout + ?S,        (2)



is ground-water recharge, and


is ground-water discharge.


The largest terms in equation 2 are RECH and GWdis. Recharge (RECH) includes infiltration of precipitation, leakage from streams and canals, and deep percolation of irrigation water. Ground-water discharge (GWdis) includes natural discharge of ground water to springs and streams, water consumed by plants with their roots extending to the water table, and by pumping of wells. A schematic depiction of a ground-water flow system is shown in figure 6.

Ground-Water Recharge

Ground water originates as precipitation. Recharge is generally greatest in upland areas where the largest amount of precipitation occurs. The principal recharge areas in the upper Klamath Basin are the Cascade Range and uplands within and on the eastern margin of the basin. Only a fraction of the water that falls as precipitation, however, makes it to the ground-water system. Most either is returned to the atmosphere through evaporation from vegetative surfaces and transpiration by plants, or runs off. In areas where soils and underlying bedrock have low permeability, infiltrating precipitation tends to flow to streams. Such areas typically have well developed stream networks. In areas where soils and underlying bedrock are highly permeable, for example the young volcanic landscapes of the Cascade Range, water infiltrates more easily to the ground-water system. Young volcanic areas often have poorly developed stream networks as a result. Water that percolates through the soil to a depth beneath the root zone potentially can become ground water.

Although direct infiltration of precipitation is the principal source of recharge in the upper Klamath Basin, there are other sources. Stream leakage can be a source of ground-water recharge in areas where the elevation of the stream is higher than the water table and the streambed is permeable. For example, streams that enter the Klamath Marsh area from the Cascade Range in the northern part of the study area generally lose much or all of their flow into the highly permeable soil as they flow onto the pumice plain. No major streams in the upper Klamath Basin, however, are known to lose regionally significant water in this manner, and stream leakage probably is not a substantial source of recharge on a regional scale in the basin.

Irrigation activities also can result in artificial ground-water recharge. Irrigation canals typically lose some water to the shallow parts of the ground-water system. No measurements of canal leakage rates in the upper Klamath Basin were available or made during this study, but measurements exist for other areas. Canal leakage rates determined from ponding studies range from less than 1 to greater than 20 (ft3/s)/mi in the upper Deschutes Basin, directly to the north of the Klamath Basin (Gannett and others, 2001). The large rates are from unlined canals in fractured lava. Rates in areas underlain by sedimentary deposits commonly range from less than 1 to 3 (ft3/s)/mi. Canal leakage rates in the Methow Valley of Washington range from 1.0 to 10.7 (ft3/s)/mi, and average 1.8 (ft3/s)/mi (Konrad, 2003). Canal bed materials there include glaciofluvial deposits, colluvium, clay, and bedrock. In addition to canal leakage, water applied to fields can percolate beneath the root zone and into the shallow parts of the ground-water system. The amount of deep percolation of irrigation water depends on the irrigation method. Gannett and others (2001) estimated ground-water recharge from deep percolation of irrigation water in the upper Deschutes Basin to be about 49,000 acre-ft/yr, or about 11 percent of the estimated water deliveries. Studies in the Amargosa Desert in Nevada resulted in estimates of deep percolation ranging from 8 to 16 percent of applied water (Stonestrom and others, 2003). No data are available to determine the amount of ground-water recharge from canal leakage and deep percolation of irrigation water in the upper Klamath Basin. However, ground-water recharge from irrigation activities is indicated because the water table in the shallow aquifers in the Project area rises during the irrigation season, and 2001 measurements showed the shallow water table declined when irrigation was severely curtailed in the Project area. Moreover, some deep irrigation wells also respond when canals of the Klamath Project start flowing in the spring (Bill Ehorn, California Department of Water Resources, written commun., 2002), indicating some recharge takes place, at least locally, to the deeper parts of the ground-water system from irrigation-related activities. Most irrigation in the upper Klamath Basin occurs in alluvial stream valleys and lake basins, and these areas are commonly crisscrossed by drains. Much of the water recharged to the shallow parts of ground-water system by irrigation activities likely discharges to the drain system (or streams) after traveling underground only a short distance (probably less than thousands of feet). Data are insufficient to estimate net regional ground-water recharge from irrigation activities in the upper Klamath Basin; however, the low permeability of the lake sediments that underlie most of the Klamath Project area suggests that ground-water movement from the water table in the Project area to the deeper, regional ground-water system is somewhat restricted and that canal leakage and deep percolation of irrigation water probably are not a significant source of recharge to the regional ground-water system.

Ground-water recharge cannot be directly measured at a regional scale. Regional ground-water recharge can be estimated, however, by measuring ground-water discharge, which can be measured or estimated with reasonable accuracy. Equation 2 shows that ground-water recharge to a system is equal to the discharge plus or minus any changes in storage. The long-term change in ground-water storage in the upper Klamath Basin (as indicated by long-term water level data) is negligible compared to the annual ground-water budget, allowing recharge to be estimated by measuring or estimating components of discharge.

The principal avenues of ground-water discharge in the upper Klamath Basin are discharge to streams, evapotranspiration by plants with roots that penetrate to the water table (in a process known as “subirrigation”), and pumping. Ground-water discharge to streams is estimated to average about 1.8 million acre-ft/yr, or about 2,400 ft3/s.

Ground-water pumping in 2000, prior to the rapid increase starting in 2001, is estimated to have been about 0.15 acre-ft/yr. Ground-water discharge through subirrigation in areas where the water table is close to land surface is difficult to estimate because it often occurs in wetlands where water comes from both ground- and surface-water sources. Total evapotranspiration from Upper Klamath Lake and surrounding wetlands, the Tule Lake and Lower Klamath Lake Refuge wetlands, and Klamath Marsh is estimated to be roughly 0.7 million acre-ft/yr (table 3), a large amount of this, however, is supplied by surface water. Therefore, ground-water discharge through subirrigation is likely small compared to discharge to streams.

Given that regional-scale long-term changes in ground-water storage are small, average recharge to the regional ground-water system is assumed to be approximately equal to the estimated ground-water discharge to streams and wells of about 2 million acre-ft/yr (rounded up to reflect some evapotranspiration directly from the water table). This figure does not include recharge from irrigation activities in the Project or subsurface discharge to or recharge from adjacent basins.

Ground-water recharge from precipitation, therefore, is about 20 percent of the total precipitation basinwide. The exact percentage, however, varies spatially and temporally. Gannett and others (2001), working with a water-balance model developed by Boyd (1996), noted that ground-water recharge in the upper Deschutes Basin ranges from 5 to 70 percent depending on location. In the Cascade Range, where there is a large amount of precipitation, which far exceeds potential evapotranspiration, a large percentage enters the ground-water system. In contrast, only a small fraction of the precipitation recharges ground water in the very dry interior parts of the basin, where precipitation is a fraction of the potential evapotranspiration. Temporally, recharge varies seasonally and from year to year. Recharge from precipitation in mountainous areas, like the Cascade Range, occurs during spring snowmelt. Recharge from irrigation occurs during the irrigation season. The timing of recharge pulses from these sources can be seen in water level data from wells (discussed later). Recharge will vary from year to year depending on the annual precipitation. Estimated basinwide recharge in the upper Deschutes Basin ranged from less than 3 in/yr during the drought years of 1977 and 1994 to more than 20 in/yr in 1982 (Gannett and others, 2001).

Ground-Water Discharge to Streams

Water flows to streams through a variety of mechanisms. For convenience, streamflow is often broken into three components: surface runoff, interflow, and baseflow. The surface runoff component reaches the stream through overland flow or flow in the top of the soil profile. Such flow is typically rapid, and is responsible for the rapid rise in streamflow during and immediately after storms or snowmelt. Interflow, also termed subsurface runoff, reaches the stream through flow in unsaturated or temporarily saturated zones in the upper soil layers. Baseflow generally is considered to be fed by ground-water discharge. Baseflow can originate from a variety of scales of ground-water flow, ranging from short, local flow paths to long and deep regional flow paths. Baseflow generally is the source of water in streams in the late summer and fall, when little or no precipitation or remaining snow are available to provide surface runoff or interflow.

The flow in some streams in the upper Klamath Basin consists entirely of ground-water discharge. Such streams are characterized by consistent year-round flow with little seasonal variability. An example of a typical ground-water-fed stream is the Wood River at Fort Klamath (USGS stream-gaging station number 11504000), which during the period of record from 1913 to 1936 had a mean annual flow of 215 ft3/s and a standard deviation of daily mean flows of only 58 ft3/s. The mean September flow of the Wood River was 199 ft3/s, or about 93 percent of the mean annual flow. Other streams, in contrast, have a relatively small component of ground-water discharge and consist predominantly of surface runoff. Such streams have large seasonal variability, with high flows during and immediately after rainfall or snowmelt followed by low or no flow during the dry periods of the year. Stream gaging data from the Sycan River below Snake Creek near Beatty (station 11499100) provide an example of a stream with a large component of surface runoff. During the period of record from 1973 to 2003 the Sycan River at this location had a mean annual flow of 152 ft3/s and a standard deviation of daily mean flows of 282 ft3/s. The mean September flow of the Sycan River here is only 23 ft3/s, or about 15 percent of the mean annual flow. Most streams in the Klamath Basin exhibit all three components of discharge throughout the year. Identifying the amount of streamflow supplied by ground-water discharge is problematic during times of the year when there is substantial contribution from overland flow and interflow. However, in the late summer and fall, when there is scant precipitation and snow has melted, streamflow is composed largely of ground-water discharge. Exceptions to this generalization include streams receiving substantial irrigation return flow or water from reservoir releases. Where these exceptions do not occur, or can be accounted for, streamflow during the fall months (September–November) when precipitation, runoff, and interflow are nearly absent provides a good estimate of baseflow or ground-water discharge.

The location and quantity of ground-water discharge entering the stream network was estimated at numerous locations throughout the upper Klamath Basin (table 6, at back of report). Estimates are, for the most part, based on measurements of actual spring discharge or streamflow during late summer and fall. For some spring-dominated streams (Spring Creek, for example), streamflow over the entire year could have been used to determine ground-water discharge. However, to maintain consistency in the analysis, data from the fall months were used exclusively where possible. Some of the ground-water discharge estimates were based on the OWRD natural streamflow analysis for the Klamath Basin (Cooper, 2004). These estimates represent the median or typical flow for a particular month over a 30-year base period, from 1958 to 1997. For this study, the work was supplemented by additional analysis, streamflow measurements, and fieldwork performed from 1997 to 2005. When possible, the estimate represents the typical ground-water discharge for the fall over a base period from 1958 to 1987. Selection of this base period is detailed in Cooper (2002). However, sometimes the available data at a location was insufficient to generate an estimate that represented the base period. In those cases, the estimate may not reflect the long-term average conditions. The data sources and techniques used to estimate ground-water discharge are listed for each reach in table 6.

Estimates of ground-water discharge have inherent uncertainty. One source of uncertainty is the streamflow measurements on which they are based. For example, records from stream gages are rated “excellent” when 95 percent of the daily discharge values are within 5 percent of the true value, “good” when 95 percent of the daily discharge values are within 10 percent of the true value, and “fair” when 95 percent of the daily discharge values are within 15 percent of the true value. Some of the estimates in table 6 are based on regression models where estimates are derived by comparing streamgage data that span periods that are short or outside of the base period, or miscellaneous measurements with long-term flow data from streams determined to be hydrologically representative of the stream in question. Regression models are another source of uncertainty. A detailed description of the regression analyses is available in Cooper (2004). Some ground-water discharge estimates in table 6 are based on single measurements or averages of multiple measurements and are not shifted to the base period. Consequently, there is uncertainty as to the degree to which they represent long-term average conditions. An assessment of the level of certainty of each estimate is included in table 6, and most figures are rounded to two significant figures. Rates of discharge at the large spring complexes responsible for most of the ground-water discharge in the basin are generally well known and have the least associated uncertainty. Although there may be uncertainty in the estimates of long-term average ground water discharging at certain locations, the presence of discharge at the listed locations is well established, and the general distribution and magnitude of ground-water discharge in the upper Klamath Basin is well understood.

Geographic Distribution of Ground-Water Discharge to Streams

Ground-water discharge to major streams was estimated in five subregions. The subregions were based on the USGS 4th field Hydrologic Unit Code (HUC) with further analysis by smaller stream groupings in each subregion, based largely on 5th field HUCs. The five subregions are (1) Sprague River, (2) Williamson River, (3) Upper Klamath Lake, (4) Lost River, and (5) Klamath River.

The values in table 6 represent the estimated long-term average ground-water discharge (gains) to all major streams in the stream system, and therefore reflect most of the ground-water discharge to the streams. Average ground-water discharge into the stream network of the upper Klamath Basin (above Iron Gate Dam) totals about 2,400 ft3/s (1.8 million acre-ft/yr). This estimate includes gains to the Lost River, which are at least 195 ft3/s, but may be higher due to unknown gains to the river below Lost River Diversion Channel. Ground-water discharge directly to Lower Klamath Lake and the Tule Lake Sump was not estimated due to insufficient data.

Ground-water discharge varies from subbasin to subbasin, reflecting precipitation patterns as well as geologic controls on ground-water movement (fig. 7). The largest ground-water discharge areas are in the Lower Williamson, Wood River, Upper Klamath Lake, and Klamath River subbasins. Besides Upper Klamath Lake and parts of the Klamath River subbasins, the specific locations of ground-water discharge are largely known and the quality of the estimates is considered good. Estimates of ground-water discharge to marshes have larger uncertainty due to difficulties with mass balances.

Temporal Fluctuations in Ground-Water Discharge to Streams

Ground-water discharge to streams is not constant, but fluctuates with time in response to variations in recharge and, in some circumstances, ground-water pumping. In the upper Klamath Basin, varying recharge is the predominant cause of ground-water discharge fluctuations. Recharge varies in response to seasonal weather patterns (wet winters versus dry summers), as well as in response to decadal-scale drought cycles and longer-term climate trends. A graph of total annual precipitation at Crater Lake National Park (fig. 8) shows the year-to-year variations in precipitation and longer-term variations. A useful way to look at long-term climate cycles is a graph of the cumulative departure from average (fig. 8). Water-table fluctuations and variations in discharge often mimic this pattern. The precipitation at Crater Lake over the past several decades exhibits the pattern observed at other precipitation stations throughout the region and in streamflow and ground-water levels as well. Most notable is the dry period in the 1930s and early 1940s followed by a wet period in the late 1940s and 1950s. The pattern from the 1960s to the present is characterized by decadal scale drought cycles superimposed on an apparent drying trend. Generally dry periods include 1966 to 1968, 1976 to 1981, 1987 to 1994, and 2000 to 2005.

The timing and magnitude of ground-water discharge fluctuations can vary depending on the scale of the flow system involved. Small-scale systems with flow paths of less than a few miles and catchments of 1 to 10 mi2, for example the flow system feeding the springs at the head of Annie Creek near Crater Lake, fluctuate in response to present-year precipitation. Such features will have large discharge during wet years and small discharge during dry years. In contrast, large-scale flow systems with flow paths of tens of miles and catchments of hundreds of square miles, such as the low-elevation regional spring complexes feeding the Wood River or Spring Creek, respond more to longer-term climate signals. Instead of reflecting the year-to-year precipitation like small springs, large-scale systems tend to integrate precipitation over several years and follow a pattern similar to the cumulative departure from average precipitation (fig. 8).

Several techniques were used in this study to determine or evaluate variations in ground-water discharge. Ground-water discharge fluctuations were in some cases measured directly by gaging stations on streams that are solely spring fed. Such data are rare in the upper Klamath Basin. Ground-water discharge fluctuations were calculated in some areas where two or more gaging stations with overlapping records bracket a stream reach to which ground-water discharges. These types of data are available for several stream reaches in the upper Klamath Basin. The analyses in such situations, however, were complicated by ungaged diversions or tributary inflow. Information on ground-water discharge fluctuations also was provided by sets of miscellaneous streamflow measurements along reaches where ground water discharges. Many streams include surface runoff as well as a large component of ground-water discharge. Comparing late summer or fall flows (when streamflow is commonly composed largely of ground-water discharge) from year to year can provide useful information on temporal variations in ground-water discharge.

Ground-Water Discharge to Streams by Subbasin

Sprague River Subbasin—Geographic Distribution of Ground-Water Discharge

The Sprague River subbasin encompasses the entire drainage above its confluence with the Williamson River, including the Sycan River drainage basin (fig. 7). The subbasin includes many runoff-dominated streams in the volcanic upper watersheds as wells as isolated springs and ground-water-dominated streams in the sediment filled valleys in the lower reaches of tributaries and along the main stem. The Sprague River responds relatively quickly to precipitation and snowmelt events, with peaks commonly exceeding 2,500 ft3/s in April or May at the gaging station near Chiloquin (station 11501000). Mean monthly discharge at this location averages 1,300 ft3/s during April and 300 ft3/s during October. Major hydrologic features of the subbasin include the North and South Forks of the Sprague River, the Sycan River and Marsh, and Kamkaun Springs.

The Sprague River originates along the flanks of Gearheart Mountain and Coleman Rim in the highlands along the central-eastern edge of the upper Klamath Basin. From these highlands, the North and South Forks gain water from numerous tributaries as they flow down mountain canyons to the upper Sprague River Valley, above Beatty Gap. The hydrologic regimes of the North and South Forks have a pronounced runoff component and similar hydrographs near the uplands, with peaks occurring during snowmelt in the spring. However, above the Sprague River Valley, the North Fork gains significant ground water, whereas the South Fork does not.

From the confluence of the North and South Forks, the Sprague River meanders downstream through the narrowing upper Sprague River Valley, until it passes through Beatty Gap into the lower valley. Gains due to ground-water inflow occur in the upper valley, which contains both drained and un-drained wetlands. More ground-water discharge occurs to a spring complex (locally known as Medicine Springs) just downstream of Beatty Gap. From here, the Sprague River meanders through the lower Sprague River Valley for 75 mi, to its confluence with the Williamson River.

Aside from the runoff-driven Sycan River, tributaries north of the river and downstream of Beatty Gap are limited to a few unnamed ephemeral creeks draining the Knot Tableland and a few small springs near the mouth of the Sycan River. South of the river are four perennial and several ephemeral creeks. Three of the perennial creeks (Spring, Brown, and Whisky Creeks) are largely ground-water fed and lack a significant runoff component (fig. 7). (Note: The Spring Creek that is tributary to the Sprague River is too small to show at the scale of figures in this report. It enters the Sprague River just east of Brown Creek. The other Spring Creek mentioned in this report is tributary to the Williamson River.) An additional, but smaller amount of ground water discharges to Trout Creek as well as a few small springs near the mouth of Whisky Creek. Two large, isolated spring complexes, Kamkaun-McReady and Whitehorse, farther downstream, are the only other ground-water discharge areas in the lower valley.

Ground-water discharge to the North and South Forks of the Sprague River (above the valley) was estimated using data from gaging stations with short periods of record and miscellaneous measurements made from 1992 to 2002. The spatial distribution of gains is relatively well known in the reaches and main tributaries of the North and South Forks (fig 7). However, the locations of specific springs have not been identified in either subbasin. Continuous and miscellaneous streamflow measurements were analyzed using index regression techniques on the North and South Forks to improve understanding of the temporal variability of ground-water discharge. Gains to the South Fork above the valley are about 24 ft3/s, with most ground-water discharge occurring above the confluence with Brownsworth Creek (table 6). On the basis of regional regression, Deming and Fritz Creeks are estimated to contribute an additional 5 ft3/s of ground-water discharge to the system. Ground-water discharge to the North Fork is about 92 ft3/s, with one-third of the flow originating from Fivemile and Meryl Creeks. There are no direct inputs to the lower 10 mi of the North Fork.

The spatial distribution of ground-water discharge in the Upper Sprague River Valley (from the confluence of the North and South Forks to Beatty Gap) is more uncertain. Although there are no identified springs in the area, synoptic measurements show about 52 ft3/s of ground-water discharge along the 20-mi reach, including the main stem between Beatty Gap and the confluence of the North and South Forks, and the lower 11 mi of the South Fork (table 6).

The Lower Sprague Valley is defined as the area between the mouth of the river and Beatty Gap. The locations of gains in this valley are well known from numerous sets of synoptic measurements. Gains total about 150 ft3/s. Ground water discharges directly to the river from the springs below Beatty Gap, Kamkaun, McReady, and White Horse Springs, as well as through tributaries at Whisky, Spring, and Brown Creeks. Even though ground-water discharge occurs at discrete locations, the locations can be lumped into two areas in the lower valley: (1) the valley between Whisky Creek and Beatty Gap (75 ft3/s), and (2) the valley near Kamkaun, Whitehorse, and McReady Springs (73 ft3/s) (fig. 7 and table 6). The temporal and spatial distribution of ground-water discharge in this subbasin is well understood given the multiple synoptic measurements made when the streamflow recorded at Chiloquin (11501000) was near the long-term average flow.

The Sycan River is the other main tributary to the Sprague River, but it contributes relatively little baseflow (historically about 30 ft3/s) given its drainage area of 563 mi2. The Sycan River is a snowmelt-runoff dominated stream, with peak flows occurring during the spring freshet (March–June). Monthly mean flows at the gaging station near Beatty indicate spring runoff flows are significant (400 ft3/s), whereas fall baseflows are minimal. Most tributaries to the Sycan River are ephemeral, contributing flow only during snowmelt or precipitation events, with ground water being a minor contributor to streamflow.

The Sycan River originates in the forested uplands east of Sycan Marsh on the western side of Winter Ridge, and is the only perennial tributary to Sycan Marsh from the east. At this location, the river has the characteristics of a runoff-dominated stream, with peak flows occurring during spring to early summer and comparatively small baseflows in the fall. Long Creek is the main tributary west of Sycan Marsh and has lower peak flows, but a larger component of ground-water discharge than the Sycan River above the marsh, even though the watershed has about one-half the area. Ground-water discharge to the Sycan River and Long Creek above Sycan Marsh totals about 24 ft3/s.

Preliminary hydrologic analysis of Sycan Marsh indicates that it is predominantly a surface-water dominated wetland. However, some ground water discharges to the marsh from numerous springs associated with a fen at the northern part of the marsh. Other than the Sycan River and Long Creek, most tributaries to the marsh are ephemeral, contributing flow only during snowmelt or precipitation events. Nested piezometers show a downward head gradient in most of the marsh, indicating that water moves from the surface downward (Leslie Bach, The Nature Conservancy, oral commun., 2005), which suggests that the marsh is an area of ground-water recharge.

The relation between the ground-water system and Sycan Marsh was evaluated by means of a water balance. Surface inflows and outflows from gaging station data were adjusted to the base period and then combined with precipitation and marsh evapotranspiration estimates to derive the ground-water gains or losses. The water balance resulted in a slight loss (-10 to -20 ft3/s), suggesting that the marsh may be a ground-water recharge area. This is consistent with the downward head gradient seen in piezometer nests. This water balance has a large uncertainty because of the evapotranspiration and soil moisture terms.

The lower Sycan subbasin (area between the mouth and Sycan Marsh) has relatively little ground-water discharge (21 ft3/s), which occurs at two locations. The first is an isolated spring, Torrent Spring (12 ft3/s), 10 mi downstream of Sycan Marsh. The second is from a number of springs, seeps, and creeks along the lower 10 mi of the river (fig. 7 and table 6).

Sprague River Subbasin—Temporal Variations in Ground-Water Discharge

Quantifying the temporal variations in ground-water discharge in the Sprague River subbasin is difficult due to a lack of data. Ground water discharges to a variety of spring complexes and spring-fed streams in the basin. Present gaging stations in the Sprague River subbasin are not well suited to provide direct measurement of ground-water discharge variations because of the effects of diversion. However, some inferences can be made by evaluating late-season flows at gages on the main stem with long periods of record. Gaging stations at Beatty and near Chiloquin provide useful information. However, diversions, tributaries, and probable irrigation return flow affect measurements at these locations. At Beatty, September mean discharge, the best proxy available for baseflow above that location, varied from about 80 ft3/s to about 180 ft3/s during the period of record from 1954 to 1991. September mean discharge near Chiloquin (fig. 9) ranged from less than 150 ft3/s to greater than 350 ft3/s. Like other streams, the variations in September mean stream discharge generally follow climate cycles, with the highest flows following multiple wet years and the lowest flows following multiple dry years.

Williamson River Subbasin—Geographic Distribution of Ground-Water Discharge

The Williamson River originates from springs just east and south of Taylor Butte. From its source, the river flows almost due north through a wide, sediment-filled valley for 35 mi before flowing west for 5 mi, where it historically spread over a delta into Klamath Marsh. The natural channel at the entrance to the marsh no longer exists, however, because the river has been diked and redirected. Most tributaries to the upper Williamson River originate along the flanks of Yamsay Mountain and the ridge to the south and are ephemeral, with flows occurring during spring snowmelt. However, significant springs contribute water directly to the upper Williamson River, which, as a result, has robust baseflow in addition to a runoff signal in its hydrograph during spring (fig. 10). Data recorded below Sheep Creek (station 11491400) indicates that flows average about 90 ft3/s during spring and 57 ft3/s in fall (table 6).

The spatial distribution of ground-water discharge to the Upper Williamson River has been largely identified from synoptic measurements (fig. 7 and table 6). Ground water discharges directly into the Upper Williamson River at several large springs upstream from the gage below Sheep Creek (station 11491400) and averages 54 ft3/s, with Wickiup Spring (24 ft3/s) being the largest single contributor (table 6). An additional 26 ft3/s of gain occurs between Sheep Creek and the marsh. Total ground-water discharge in the area is about 80 ft3/? s. The knowledge of the temporal variations in ground-water discharge to the Upper Williamson River is good upstream from the Sheep Creek gaging station owing to data from the long-term records at that site. However, between Sheep Creek and Klamath Marsh, no continuous streamflow record exists.

The only other perennial tributary that reaches Klamath Marsh is the spring-fed Big Springs Creek. However, even this creek may go dry during successive drought years (Newcomb and Hart, 1958). Surprisingly, Big Springs Creek shows a relatively flashy response to snowmelt and rainfall events that is atypical for spring-fed streams. Presumably, this rapid response is due to the ability of local rainfall and snowmelt to move easily through the very permeable pumice soils. Water in most other perennial streams draining to the marsh from the eastern side of the Cascades infiltrates into the pumice plain before reaching the marsh. Water in Sand and Scott Creeks would reach the marsh, but it is diverted to irrigate pasture lands on the western edge of the marsh.

Ground-water discharge to Big Springs, Sand, and Scott Creeks, and other tributaries west of Klamath Marsh, totals about 78 ft3/s. About 12 ft3/s of the total discharge is to Miller and Sink Creeks, which lose their flow through infiltration into the pumice plain before reaching the marsh. Discharge to these streams was estimated from miscellaneous measurements and short-term gaging station data using index regression. All other tributaries are either ephemeral, or infiltrate into the pumice plain.

A mass balance indicates that average net annual ground-water discharge directly to Klamath Marsh was approximately 50 ft3/s between 1971 and 2000. This estimate is based on available gage data for tributary inputs, estimates for evapotranspiration, and direct precipitation on the marsh. The analysis assumes that the net change in water stored in the marsh during the 30-year period was negligible. The spatial distribution of ground-water discharge directly to the marsh is unknown.

The hydrograph of the Williamson River at the outlet of Klamath Marsh near Kirk (station 11493500, not shown) has a runoff signal, presumably from ephemeral tributaries and direct local runoff from the marsh. Flow of the Williamson River at the outlet of the marsh ceases during most summers due to the large amount of evapotranspiration in the marsh.

South of the gaging station near Kirk (altitude 4,483 ft), the Williamson River descends into a narrow, steep canyon as it drops in elevation. Small seeps and springs appear in the canyon walls near an altitude of 4,220 ft. As the river exits the canyon, three spring-fed streams contribute most of the baseflow to the Williamson River: Spring Creek, Larkin Creek, and Larkin Springs. Hydrographs of the Williamson River below these streams and above the Sprague River show a system with a large component of ground-water discharge that responds relatively slowly to precipitation and snowmelt events and that has gradual accession and recession curves. Peak flows commonly exceed 1,000 ft3/s and usually occur in March. Low flows consistently range near 300 to 350 ft3/s and occur during summer. Gains to the river due to ground-water discharge below Klamath Marsh occur at Spring Creek (300 ft3/s), Larkin Creek (10 ft3/s), Larkin Springs (10 ft3/s), and miscellaneous small springs (28 ft3/s) above Larkin Spring (table 6).

Williamson River Subbasin—Temporal Variations in Ground-Water Discharge

About 78 percent of the 67 ft3/s mean annual discharge of the uppermost Williamson River is composed of ground water. Information on fluctuations in ground-water discharge to the upper Williamson River comes largely from the gage downstream from Sheep Creek (station 11491400) operated since 1974. August–September flow of the upper Williamson River, which is mostly spring discharge, averages 52 ft3/s. Synoptic measurements in November 2002 showed 54 ft3/s ground-water discharge to the reach (table 6). A graph of monthly mean flows of the Williamson River below Sheep Creek (fig. 10) shows that the base flow, as represented by September mean discharge, varies by a factor of nearly 2, from 37 to 70 ft3/s. Comparing September mean flows with precipitation at Crater Lake (fig. 10) shows that this variation correlates with climate cycles. A plot of September flow of the Williamson River at Lenz and the cumulative departure from average precipitation at Crater Lake shows a positive linear relation with a correlation coefficient of 0.79 (fig. 11). Part of the observed variation in September mean flow could be due to variations in surface-water diversions, which also are correlated with climate, as irrigation demands are less during wet periods and greater during dry periods. Given the small irrigated area above the gage (about 3,000 acres), the probable climate-driven variation in September diversions is small compared to the observed variations in streamflow. This indicates that most of the observed variation in September mean flow can be attributed to fluctuations in ground-water discharge.

Newcomb and Hart (1958) showed that ground-water discharge to Big Springs Creek varies from zero to about 90 ft3/s in response to drought cycles. Meinzer (1927) shows the discharge of Big Springs Creek decreasing from 61 to 11.6 ft3/s between 1914 and 1925 in a more or less linear manner in response to a general drying climate trend. La Marche (2002) noted that Big Springs Creek also shows seasonal fluctuations in response to annual snow melt. This suggests that Big Springs Creek is fed by a local, possibly perched, flow system.

The area of the lower Williamson River, between the gage at Kirk and the confluence with the Sprague River, is one of the major ground-water discharge areas in the upper Klamath Basin. About 86 percent of the ground-water discharge in this area is to Spring Creek, a short tributary to the Williamson River that is fed entirely by springs. The remaining ground-water discharge is to Larkin Creek, Larkin Springs, and other nearby springs.

Spring Creek is particularly important because it provides much of the flow to the Williamson River, an important source of water to Upper Klamath Lake, during summer. Many measurements of instantaneous streamflow have been made along Spring Creek during the past 100 years by the USGS and OWRD. Spring Creek flow varies with time and correlates with climate (fig. 12). The correlation coefficient between Spring Creek flow and the cumulative departure from average precipitation at Crater Lake between 1932 and 2002 is 0.72. Spring Creek is unaffected by surface-water diversions, and ground-water pumpage in the area is not enough to cause the observed discharge variations.

A more continuous measure of the ground-water discharge variations in the area can be developed using data from streamflow gages on the Williamson River near Kirk (11493500), the Sprague River near Chiloquin (11501000), and the Williamson River below the Sprague River, near Chiloquin (11502500). If the streamflow at the former two gages is subtracted from the latter, the positive residual (indicating a gain in streamflow between the gages) is due primarily to ground-water discharge, most of which is from Spring Creek. The ground-water discharge in this area, on the basis of September mean flows, averages about 306 ft3/s, and ranges from about 250 to 400 ft3/s. The uncertainty of this estimate (on the basis of estimated gage error) is only about ±30 ft3/s. This analysis is complicated by the fact that there are ungaged diversions from the Sprague River below the gage at Chiloquin, most notably the Modoc Irrigation District canal. Diversion records for the Modoc Canal are available from 1915 to 1924, as are miscellaneous discharge measurements throughout the 1980s. Measurements of September flow average about 25 ft3/s. Accounting for this ungaged diversion increases the average ground-water discharge in this area based on gage data to 331 ft3/s. This figure compares favorably with the 350 ft3/s estimate based on synoptic and miscellaneous flow measurements. The temporal variations generally correspond to decadal precipitation cycles (fig. 12), and comparing the calculated September mean ground-water discharge and the cumulative departure from average precipitation at Crater Lake results in a correlation coefficient of about 0.68.

Upper Klamath Lake Subbasin—Geographic Distribution of Ground-Water Discharge

The Upper Klamath Lake subbasin encompasses 723 mi2 above the outlet of Upper Klamath Lake, excluding the Williamson and Sprague drainages. The subbasin includes Upper Klamath Lake, the broad, flat Wood River valley to the north and the adjacent uplands including the Cascade Range to the west, Mt. Mazama (the Crater Lake highlands) to the north, and multiple fault-block mountains and the Williamson River delta to the east. The uplands on the eastern side rise abruptly from the valley floor along north-south trending faults. The Wood River Valley is filled with Quaternary sediment, much of which is fine-grained and has low permeability. Major hydrologic features include the Wood River, Upper Klamath and Agency Lakes, and Sevenmile Creek.

The general hydrology of the subbasin is dominated by ground-water discharge from spring complexes coincident with fault scarps at the western and eastern edges of the Wood River Valley and Upper Klamath Lake (fig. 7). About one-half of the ground-water discharge in the subbasin occurs to the Wood River and its tributaries (table 6). Approximately one-third of the discharge occurs directly into Upper Klamath Lake at known and unknown locations. The remaining ground-water discharge occurs in the tributaries draining the eastern flank of the Cascade Range.

The Wood River receives the largest amount of ground water in the Upper Klamath Lake subbasin (490 ft3/s), with most of the discharge occurring at discrete spring complexes along the fault scarp on the eastern boundary of the valley (table 6). Two tributaries originating on the flanks of the Crater Lake highland, Annie and Sun Creeks, contribute roughly 14 percent of ground water discharged into the river. The estimates were derived from miscellaneous measurements taken in the Wood River subbasin, with subsequent regressions to index gages.

Ground water discharges from the Cascade Range to tributaries along the western margin of the subbasin at a rate of about 120 ft3/s (table 6). The majority of that flow originates from tributary springs in the valley along the western fault scarp of the region. The creek with the largest watershed in this subarea, Fourmile Creek south of Pelican Butte, contributes only 2 ft3/s of baseflow to the region. Estimates were derived from short-term gaging station records and regression to index stations elsewhere in the basin.

Ground-water inflow to Upper Klamath Lake was estimated using a monthly water balance for the 1965 through 1967 water years. Hubbard (1970) measured or calculated all tributary inflows and outflows from the lake, including streamflow, diversions, precipitation, evapotranspiration, and agricultural return flows. Hubbard’s monthly estimates of ground-water inflow to the lake of averaged about 350 ft3/s from 1965 to 1967 (the median value is about 330 ft3/s). This is about 15 percent of the 2,330 ft3/s average total inflow to the lake during that period. Hubbard’s estimated ground-water discharge to the lake compares favorably with estimates of the difference between Upper Klamath Lake inflows and outflows (such as, ground-water inflow) by others such as Cooper (2004) that cover a much longer base period (30 years compared to 3 years). Hubbard’s estimate of average ground-water inflow was revised downward to about 320 ft3/s by the Bureau of Reclamation (2005) using updated stage-capacity curves for the lake. Although many springs have been mapped around the margins of Upper Klamath Lake, their combined discharge is much less than the estimated ground-water inflow. Consequently, the spatial distribution of much of the ground-water inflow directly to the lake is unknown.

Upper Klamath Lake Subbasin—Temporal Variations in Ground-Water Discharge

Temporal variations in ground-water discharge to Annie Spring can be evaluated using data from the gaging station (11503000) that has been operated on Annie Creek just below the spring since 1977 (fig. 13). The discharge from Annie Spring is small, averaging about 3 ft3/s. It is included here to illustrate the behavior of smaller flow systems. Annie Spring shows temporal variations that are different from those of the large-scale systems discussed previously. The lowest flows of large-scale spring systems are typically August through September. The lowest flows of Annie Creek, in contrast, are January through March. The likely cause is that Annie Spring is fed by ground water recently recharged and following very short flow paths, and consequently much of the water feeding the springs is frozen as snow during the winter months. The annual low flows of large-scale systems typically increase each year during periods of successive wetter-than-average years. This is less pronounced with Annie Creek. A graph of monthly and January to March mean flows of Annie Spring (fig. 13) shows that it peaks before the cumulative departure from average precipitation curve. This is because of the lack of storage effects in the small flow system. Annie Spring and similar small-scale flow systems in the upper Klamath Basin have the characteristics of runoff-dominated streams.

Gaging stations have been operated intermittently on the Wood River since 1913. However, data are not easily compared because the stations have been operated at different locations that are affected differently by tributary inflow, return flow, and diversion, and the periods of record are short, ranging from roughly 1 to 14 years. Although the gaging station data do not provide a continuous long-term record of ground-water discharge, they do provide useful information on the magnitude and timing of ground-water discharge fluctuations. A USGS gaging station at Fort Klamath (11504000) operated intermittently from 1913 to 1936 shows a probable drought-related decrease in annual mean flow from approximately 310 to 140 ft3/s during its period of operation (fig. 14). Another USGS gaging station operated 4 mi south of Fort Klamath (11504100) from 1965 to 1967 shows a climate-related decrease in annual mean flow from 350 to 290 ft3/s during that period. A gaging station was operated in the early 1990s, and recently near the headwaters springs of the Wood River (about 1 mile downstream at Dixon Road) by Graham Matthews and Associates (GMA). Variations in ground-water discharge to the Wood River headwater springs can be evaluated using the GMA data (provided by Graham Matthews, written commun., May 13, 2003) along with a multitude of miscellaneous instantaneous discharge measurements made over several decades by USGS and OWRD (fig. 14). The measurements near the headwaters springs are largely unaffected by tributary inflow and diversions, and are not noticeably influenced by the presently small amount of ground-water pumping in the area. Variations in ground-water discharge to the Wood River headwater springs correlate well with the cumulative departure from average precipitation at Crater Lake (r = 0.75). Measurements show the discharge of Wood River near the headwaters springs increasing from 180 to 320 ft3/s during wet conditions in the early 1980s, and then decreasing from 320 to 160 ft3/s owing to drought in the late 1980s and early 1990s. This reduction in ground-water discharge to this single spring complex applied over 1 year equates to 114,000 acre-ft of water.

A synthetic hydrograph of the Wood River near the headwaters springs can be created by using the relation between Wood River discharge measurements and concurrent daily mean flows from the gaging station on Fall River (14057500), a similar-scale spring-fed stream about 70 mi north in the Deschutes Basin (fig. 14). The relation can be modeled using a second order polynomial with an R2 of 0.80. This synthetic hydrograph provides a reasonable depiction of the continuous temporal variations in the discharge of the Wood River headwaters springs.

Temporal variations in ground-water discharge directly to Upper Klamath Lake have not been measured. Given the magnitude of the inflow (320 to 350 ft3/s), temporal variations can be inferred from other springs in the area with comparable discharge rates and flow-path lengths (such as Wood River Springs).

Lost River Subbasin—Geographic Distribution of Ground-Water Discharge

The Lost River subbasin occupies about 1,650 mi2 southeast of Upper Klamath Lake. In its natural state, the Lost River subbasin had no outlet and it drained internally. Water occasionally flowed to the subbasin, however, from the Klamath River during floods through a slough connecting the two drainages south of Klamath Falls (La Rue, 1922). With the development of Reclamation’s Klamath Project and the construction of the Lost River diversion dam and channel (fig. 3), controlled flow between the Klamath and Lost Rivers in both directions now occurs. Major hydrologic features in this subbasin include Clear Lake and Gerber Reservoirs in the uplands, the Lost River, and the Tule Lake Sump.

The Lost River proper originates at the outlet of Clear Lake Reservoir in the southeastern part of the basin. From there the river flows northwest, dropping from the plateau containing Clear Lake into the Langell Valley where the river flows to the town of Bonanza. From there it flows west into Poe Valley and subsequently through Olene Gap into the Klamath Valley before turning southeast and terminating in the Tule Lake Sump in California.

The hydrology of the Lost River subbasin is runoff dominated above Clear Lake and Gerber Reservoirs. The drainage area of Clear Lake consists of a broad, low relief, volcanic plateau with a mean altitude of roughly 5,000 ft covering more than 750 mi2 south and east of the lake. Only one perennial stream, Willow Creek, exists in the plateau. Values for monthly mean inflows to Clear Lake (calculated from a mass balance) show that high flows occur in March (500 ft3/s), whereas low flows occur in late summer (30 ft3/s). The lands draining to Gerber Reservoir are more mountainous than those draining to Clear Lake, but are geologically similar. Mean monthly inflows to Gerber Reservoir are highest in March (280 ft3/s) and lowest during the late summer (4–5 ft3/s). Water from Gerber Reservoir flows to the Lost River via Miller Creek.

Several springs contribute flow to the Lost River subbasin in the sediment filled Langell, Yonna, and Poe valleys. Bonanza Spring, near the town of Bonanza, is a major contributor of baseflow to the river as are a series of springs adjacent to the river near Olene Gap.

Little historical data are available with which to estimate ground-water discharge in most of the Lost River subbasin. For Clear Lake and Gerber Reservoirs, USGS streamgaging data collected prior to the construction of the reservoirs were used along with inflows reported by Reclamation (Bureau of Reclamation, 1954, Appendix B) to estimate ground-water discharge during fall. The average discharge during the period of record was 40 ft3/s and 10 ft3/s, respectively for the two reservoirs. Limited data for the remaining area are available from gages operated intermittently in the early 1900s and late 1990s on the Lost River. Sets of synoptic measurements (Leonard and Harris, 1974; Grondin 2004) were sufficient to determine the location of gains to the river below the reservoirs. These measurements demonstrate that most ground-water discharge into the Lost River proper occurs at two locations: Bonanza Springs and the area just upstream of Olene Gap. The overall gain to the river between Olene and the two reservoirs is about 140 ft3/s, largely on the basis of synoptic measurements. The temporal variability of ground-water inflow is poorly known owing to the short periods of record. Below Olene Gap, data were insufficient to estimate ground-water discharge to the river. Likewise, data were insufficient to estimate direct ground-water discharge to the Tule Lake Sump.

Lost River Subbasin—Temporal Variations in Ground-Water Discharge

Bonanza Springs is the only location in the Lost River subbasin where data are sufficient to evaluate temporal variations in ground-water discharge. These springs discharge from basalt to the Lost River. Twenty-one discharge measurements of the springs, made by USGS, OWRD, or Reclamation, are in the published and unpublished literature (fig 15). Discharge of the springs is in all cases determined by comparing the difference in streamflow of the Lost River upstream and downstream from the town of Bonanza. Many upstream measurements were made between 2 and 3 mi from Bonanza, and some were made at a bridge about 5 mi upstream. Downstream measurements all have been made about 3 mi downstream at Harpold Dam. Some determinations of spring flow account for all tributary inflows (including agricultural drains) and diversions between the upstream and downstream measurement sites. Other than Bonanza Springs, these gains and losses are minor outside of the irrigation season. Many determinations of spring flow include only the upstream and downstream measurements and measurement of the single major tributary, Buck Creek. Determinations of spring discharge made by comparing flows only at Keller Bridge, Buck Creek, and Harpold Dam outside of the irrigation season are considered reasonable because the spring discharge is much larger than the other stream gains and losses.

Discharge measurements of Bonanza Springs show considerable temporal variation (fig. 15). The largest measurement, 118 ft3/s in October 1958, occurred after a 15-year period of wetter-than-average weather. The smallest measurement, 38 ft3/s in January 1992, occurred late in a drought that started in the mid-1980s. Overall, the pattern of spring discharge follows the general pattern of precipitation, reflecting drought cycles and a general drying trend since the late 1950s. Unfortunately, no measurements are available from the very dry period in the early 1940s. Bonanza Springs discharge is affected by climate, ground-water pumping, and artificial manipulation of the stage of the Lost River (Grondin, 2004). Discharge from the main spring can cease entirely during the irrigation season in dry years. Most measurements after 1960 (fig. 15) were made well after the irrigation season (December to April), so the system should have mostly recovered from the seasonal effects of pumping and diversion. Present information is insufficient, however, to determine precisely how much of the variation in spring discharge is natural and how much is related to pumping.

Klamath River Subbasin—Geographic Distribution of Ground-Water Discharge

The Klamath River subbasin encompasses the area between the outlet of Upper Klamath Lake and Iron Gate Dam. The main hydrologic features are John C. Boyle, Copco, and Iron Gate Reservoirs, Lake Ewauna, Lower Klamath Lake, and the Klamath River. The Klamath River begins at the outlet of Upper Klamath Lake, where, for the first mile or so, it is known as the Link River. From the dam, the Link River flows about 1 mi through a narrow gorge into a broad, flat valley containing Lake Ewauna, the head of Klamath River proper. Lake Ewauna, impounded by Keno Dam, is a long, narrow reservoir that traverses the northern part of the Lower Klamath Lake subbasin.

Prior to development of the region, the Klamath River occasionally spilled across the low, nearly level divide into the Lost River subbasin during floods. La Rue (1922) hypothesized that water also may have flowed from the Lost River to the Klamath River subbasin during floods. Recent analysis of topographic mapping from the early 1900s, however, suggests that the Lost River was incised to the degree that flow from the Lost River system to the Klamath River subbasin was highly unlikely (Jon Hicks, Bureau of Reclamation, oral commun., 2006). Water also moved between the Klamath River and Lower Klamath Lake subbasin prior to development. Water flowed from the Klamath River into Lower Klamath Lake during periods of high flow, usually in winter or spring. After high flows, water would flow out of the lake through the Klamath Strait back into the Klamath River. There is some uncertainty as to timing and duration of flow from the lake to the river, and it probably varied from year to year with hydrologic conditions (Weddell, 2000; Bureau of Reclamation, 2005). Flows into and out of the Lost River and Lower Klamath Lake subbasins are now controlled.

Downstream from Lake Ewauna and Keno Dam, the Klamath River enters a canyon and flows into John C. Boyle Reservoir (operated by PacifiCorp), near the confluence with Spencer Creek. Below John C. Boyle Dam, the river drops into another steep canyon. About 1 mile below the dam, a large spring complex contributes significant flow to the river. Numerous perennial streams originating from the High Cascades and older Western Cascades also add flow to the river between Keno and Iron Gate Dams. These tributaries are predominately runoff dominated; however several (for example, Spencer and Fall Creeks) have large components of ground-water discharge. Flows in the Klamath River above Iron Gate Dam are largely regulated by Reclamation and PacifiCorp impoundments, including Link River Dam at the outlet of Upper Klamath Lake. Gaging station data from below Iron Gate Dam (11516530) show a high mean monthly flow of 3,600 ft3/s in March and a low mean monthly flow of 770 ft3/s in August.

Most ground-water discharge in this subbasin occurs along the Klamath River and principal tributaries. A small amount of ground water also discharges to springs southwest of Lower Klamath Lake. Ground-water discharge in the Klamath River subbasin was calculated directly from long-term streamflow data and corrected for reservoir storage in reaches of the Klamath River from the gaging station at Keno (11509500) to the gaging station below the John C. Boyle Power Plant (11510700) (about 5 mi below the dam), and from that gage to the gage below Iron Gate Dam (11516530). Some short-term streamflow records available for tributaries were useful for discriminating ground-water discharge directly to the river from discharge to tributaries. Gains and losses between Link River Dam and Keno were not estimated owing to large uncertainties in the data in that reach. Discharge to springs southwest of Lower Klamath Lake was not measured for this study, but measurements are available from Wood (1960) and Reclamation records.

The largest source of ground-water discharge between the gage at Keno and the gage below the John C. Boyle Power Plant is a series of springs about a mile below the dam (fig. 7 and table 6). Although early references to these springs are scarce, Newcomb and Hart (1958) note that springs contribute “considerable inflow” to the river in this area. Their observations and those of local residents cited in their report predate construction of John C. Boyle Dam, indicating that these springs do not merely represent reservoir seepage. Records show that average gain from the springs is about 190 ft3/s (table 6). The temporal variation in the discharge of these springs is well characterized. The baseflow of Spencer Creek, tributary to the Klamath River in this reach, is about 27 ft3/s.

Between the gage below the John C. Boyle Power Plant and that below Iron Gate Dam, gains averaged 140 ft3/s from 1967 to 2000 (table 6). The spatial location of the ground-water discharge in this reach is not well known; however, there is evidence of inflow to the main stem of the Klamath River between river miles 207 and 213, roughly between Shovel and Rock Creeks. Thermal infrared remote sensing shows that the river cools in this reach, an indication of ground-water discharge (Watershed Sciences, 2002). This reach corresponds with the boundary between the High Cascade and Western Cascade subprovinces and is therefore an area of expected ground-water discharge. In addition, the reach traverses a large landslide complex with numerous mapped springs. Some inflow in this reach is due to tributary streams. Fall Creek, which drains an area dominated by rocks of the High Cascade subprovince, is the largest known contributor, with a baseflow of about 36 ft3/s, whereas the much larger Jenny Creek watershed, which is underlain largely by low-permeability older volcanic rocks, contributes only about 9 ft3/s. Flow data are sparse for the remaining tributaries in Oregon, but regional regression techniques show that these contributions probably amount to slightly greater than 1 ft3/s. The component of baseflow to the reach from California tributaries is unknown.

Ground water also discharges to a number of spring complexes southwest of Lower Klamath Lake (fig. 7 and table 6). Reclamation engineer Louis Hall made a reconnaissance of the Lower Klamath Lake subbasin in September 1908 during which he inventoried springs along the margin of the lake and estimated their discharge. His estimates of spring discharge to the lake total 104 ft3/s, although his estimating methods are not known (Tom Perry, Bureau of Reclamation, written commun., 2006). Wood (1960) made several measurements of discharge at three of the principal spring complexes during water year 1955. He also reports CDWR observations that year for a fourth discharge measurement. Wood’s measurements are not directly comparable to Hall’s earlier estimates because of differences in locations; however, in cases where general comparisons can be made, Wood’s measured flows appear to be about one-half of Hall’s estimates. Averages of Wood’s measurements total about 35 ft3/s. Some small seasonal variability is evident in the measurements, but the data are too sparse to define a pattern and identify the source of the variability. Measurements made in 1955 may reasonably represent the average flow during the early 1950s. Precipitation during water year 1955 was less than average, but 1953 and 1954 were close to the long-term average. Discharge of these springs probably is now less than it was in the mid-1950s, owing to dryer conditions in recent decades and increased ground-water development in the area.

Klamath River Subbasin—Temporal Variations in Ground-Water Discharge

The principal sources of data used to evaluate ground-water discharge variations in the Klamath River subbasin are stream-gaging stations at Keno (11509500), below the John C. Boyle power plant (11510700), and below Iron Gate Dam (11516530). These data are augmented with short-term gaging station records from Spencer and Fall Creeks (11510000 and 11512000, respectively). Understanding ground-water discharge variations along the Klamath River, however, is complicated by inflow from ungaged tributaries, unmeasured diversions, and changes in reservoir storage.

Ground-water discharges to the Klamath River between the Keno gage and the gage below the John C. Boyle power plant. Nearly all of the discharge is from a spring complex near river mile 224 about 1 mi below the John C. Boyle Dam and about 3.5 mi above the power plant. Thermal infrared remote sensing on July 15, 2002, showed that this spring complex cooled the river about 10°F (Watershed Sciences, 2002). Gage data indicate the flow in the river just below the springs was about 370 ft3/s on that date. Temporal variations in net ground-water discharge to this reach can be evaluated by comparing August mean flows at the two gages and accounting for changes in storage of John C. Boyle Reservoir (reservoir data from Rob Allerman, PacifiCorp, written commun, 2002). August means were used to evaluate temporal variations here instead of the September–November means to minimize the effects of fall storms. This can be done because diversions are insignificant and the mean August inflow (240 ft3/s) is reasonably close to the mean September–November inflow (230 ft3/s).

A graph showing the August mean net inflow to the reach between Keno and the gage below the John C. Boyle power plant (fig. 16) shows that ground-water discharge to this reach varied from less than 200 ft3/s to greater than 300 ft3/s during the period of record. The average error in the individual inflow estimates is approximately ± 50 ft3/s on the basis of measurement error of the stream gages. The general pattern of ground-water discharge loosely follows the decadal cycles seen in precipitation, with the lowest inflows corresponding to extended periods of drought, but that correlation is low (r = 0.30).

Ground-water discharge also occurs to the Klamath River between the gaging station below the John C. Boyle power plant and the gaging station below Iron Gate Dam. Evaluating temporal variations in ground-water discharge to this reach is made difficult by ungaged tributary inflow and probable diversions. Moreover, calculated inflow values include cumulative errors in data from the two gaging stations and in storage measurements of two large reservoirs. As with the upstream reach, temporal variations in ground-water discharge can be evaluated by calculating the differences between flows at the two gages and accounting for changes in reservoir storage (fig. 16). September to November mean inflows are used here to minimize the effects of diversions.

Between 1967 and 2002, the September to November mean inflow between the gaging stations below the John C. Boyle power plant and Iron Gate Dam ranged from 30 to 330 ft3/s, averaging 140 ft3/s (fig. 17). About 45 ft3/s if this inflow is from Fall and Jenny Creeks (table 6). The uncertainty of the inflow estimates due to measurement error is ± 110 ft3/s. The correlation between ground-water discharge and climate is not as apparent here as elsewhere in the basin owing to the small amount of net ground-water inflow relative to the streamflow measurement error.

Summary of Geographic and Temporal Variations of Ground-Water Discharge to Streams

Analysis of streamflow data indicates that many streams in the upper Klamath Basin have a large component of ground water. Most streams throughout the world rely on ground-water discharge to support flows during the dry season. The upper Klamath Basin and other basins on the eastern flank of the Cascade Range are unique in that ground water discharge composes a large proportion of the total streamflow. This is attributable to the substantial regional ground-water system that exists in the permeable volcanic terrane. Some major streams in the basin, such as the Wood River and Spring Creek, are virtually entirely ground water fed. It has long been recognized that much of the water flowing into Upper Klamath Lake originates as ground-water discharge (Bureau of Reclamation, 1954, p. 150). Of the 2,200–2,300 ft3/s average total inflow to the lake (from Hubbard, 1970, and Reclamation records), at least 60 percent can be attributed to ground-water discharge in the Wood River subbasin and springs in the lower Sprague River drainage and the Williamson River drainage below Kirk. This quantity does not include ground-water discharge to upper parts of the Williamson and Sprague River systems, which would make the figure even larger. The large component of ground water in streamflow influences the hydrologic response of the basin to climate cycles, and has implications for flow forecasting (Risley and others, 2005).

Discharge from all major ground-water discharge areas in the basin fluctuates over time. Ground-water discharge fluctuations are primarily climate driven, and, therefore, discharge from the various sources tends to vary in unison. Owing to the effects of ground-water storage, regional-scale discharge areas integrate climate conditions over multiple years. Consequently, ground-water discharge fluctuations tend to follow a pattern similar to the cumulative departure from average precipitation. A practical implication of this observation is that ground-water discharge from storage may support robust streamflow during a dry year following a series of wet years. Conversely, it may take multiple years of average or above average conditions following a protracted drought to replace ground-water storage and return spring discharge (and hence, streamflow) to predrought conditions.

Ground-water discharge variations can represent substantial volumes of water on an annual basis. The combined ground-water discharge to the lower Williamson, Sprague, and Wood Rivers just upstream of Upper Klamath Lake can vary by at least 450 ft3/s in response to climate cycles. This equates to an annual volume of 326,000 acre-ft. The actual variation in ground-water discharge to Upper Klamath Lake and its tributaries is larger, because the probable variations in ground-water discharge to Fort Creek, Crooked Creek, and springs discharging directly to the lake have not been included. Gaging station data show that net ground-water inflow to the Klamath River (and ground-water fed tributaries) between Keno and Iron Gate Dam probably varies at least 150 ft3/s in response to climate.

Ground-Water Discharge to Wells

Ground water pumped from wells in the upper Klamath Basin is used primarily for public supply and agriculture. For public supply, ground water is provided by public or privately owned utilities for drinking, municipal, industrial, and commercial purposes. Most ground water pumped for irrigation is used for agriculture; however, some is used for irrigation of cemeteries, parks, and golf courses. The following discussion is limited to ground-water withdrawals for public-supply and agricultural irrigation. Industrial and domestic withdrawals from individual private wells are not discussed because their proportion of the total ground-water use in the basin is small.


Ground-water pumping for public-supply and irrigation was estimated using different methods. Since 1985, OWRD has required public suppliers to report ground-water pumpage annually. Public suppliers in Oregon report monthly pumpage totals by individual well for each water year. The totals typically are based on direct flowmeter measurements or calculated from pumping rates and duration of pumping. The totals for this report were based on reported pumpage in 2000. No pumpage data were available for the small California communities of Dorris, Macdoel, and Tulelake. Ground-water withdrawals for these suppliers were estimated using population data.

Neither Oregon nor California requires well owners to report ground-water withdrawals for irrigation. Some irrigators in Oregon report their yearly pumpage to OWRD either voluntarily or as required by conditions in their water-right permit, but such reporting is rare. The lack of any comprehensive reporting system by both States means that indirect methods based on water right information, satellite imagery, and land and water surveys must be used to estimate the rate and distribution of ground-water withdrawal.

Ground-Water Pumpage for Public-Supply Use

Ground water is the source of water supplied by eight public systems and one quasi-public system in the study area. Public suppliers of ground water include the communities of Klamath Falls, Bly, Chiloquin, Merrill, and Malin. One resort community northwest of Klamath Falls is considered a quasi-municipal system on the basis of the variety of water uses. Ground-water withdrawals for the communities of Dorris, Macdoel, and Tulelake were estimated using recent population totals and a per capita use of 150 gallons per day (according to the methods of Broad and Collins, 1996). Public-supply systems that served more than 25 people or had at least 15 connections pumped approximately 9.3 Mgal/d (million gallons per day) or about 14.4 ft3/s in 2000. By comparison, in the 5-year period from 1996 through 2000, public supply withdrawals in the basin averaged an estimated 8.2 Mgal/d (12.7 ft3/s). The City of Klamath Falls, with a population of 19,400, accounted for 84 percent of the 2000 total, reporting withdrawals of 7.8 Mgal/d (12.1 ft3/s) from 9 city wells. Per capita use in cities such as Klamath Falls is larger than in rural areas and small towns due to the larger relative amount of commercial, industrial, and irrigation included.

Ground-Water Pumping for Irrigation

In Oregon, ground-water pumpage for irrigation was estimated by matching maps of primary and supplemental irrigation ground-water rights to areas that were determined to be irrigated by using a land-cover data set created from 30-meter resolution Landsat satellite imagery taken during the 2000 irrigation season. The Landsat image data were analyzed to help identify vegetation types and conditions (such as the stage of growth) in a process known as “classification.” The ground-water rights data sets consisted of geographic information system (GIS) layers showing places of use (POUs) and points of appropriation (POAs), and were provided by the OWRD along with data from their Water Rights Information System. The POUs represent the fields where water is applied under the terms of the water right. That right might be a primary right for ground-water irrigation, or a supplemental right under which ground water is used to supplement a primary surface-water right on the same land. A water-right can cover a single tract or several tracts not necessarily adjacent to one another. POAs correspond to specific wells at particular locations on ground-water rights. A single ground-water right may include more than one well. Pumpage was estimated only for active primary and supplemental ground-water rights for irrigation outside of irrigation district boundaries. It was assumed that no primary ground-water rights are in areas included in irrigation districts within the Klamath Project, and that supplemental ground-water irrigation within irrigation districts was negligible in 2000.

The POU boundaries were overlaid with the lands determined to be irrigated using the classified Landsat imagery (fig. 18). The purpose of the overlay was to match specific water rights to irrigated areas shown on the land-cover map developed from the imagery. The results reveal which fields with ground-water rights were actually irrigated during the 2000 irrigation season. Partial overlay matches, where either the irrigated lands from the imagery did not completely fill a POU field boundary or the imagery showed many small blocks of irrigated areas within a field boundary, were evaluated individually and the acres included or dropped (considered irrigated or not irrigated) on the basis of specific criteria. Any irrigated fields smaller than 3 acres were eliminated. Ground-water pumpage used in irrigated areas was assigned to a specific well based on water right information. Where the land-cover data showed an obvious irrigated field (for example, from a center pivot) but no matching water right in the GIS data set, a withdrawal location based on the centroid of the field was used. In total, 19,250 acres of cropland in the upper Klamath Basin in Oregon was estimated to have been irrigated with ground water in 2000, or about 30 percent of the 64,000 acres with Oregon primary ground-water rights evaluated in this analysis.

The estimation method produced a conservative number of irrigated acres and points of appropriation. The apparent low percentage of ground-water rights exercised in 2000 may be attributable to counting only the acres appearing to be irrigated on Landsat images rather than the total acreage carried on the water right, and editing criteria that were more likely to eliminate than include fields where evidence of irrigation was questionable on the satellite image. Any irrigation in areas not included in the OWRD digital water right POU maps is not included in the estimate. Estimates of pumpage within irrigation district boundaries under a pilot water bank and similar programs starting in 2001 are discussed in a separate section

The crop types in areas irrigated with ground water in Oregon were determined from the classified Landsat satellite imagery using the methodology described in appendix A. Specific crop types could not be reliably identified using the Landsat imagery. Five vegetative classes of crop types were identified, however, on the basis of spectral signatures and potential water requirements. The five classes are (1) alfalfa and irrigated grasses, (2) small grains, (3) onions and garlic, (4) potatoes and corn, and (5) strawberries.

About 40,000 acres were irrigated with ground water in the upper Klamath Basin in California during 2000. California water law does not require a permit or approval to withdraw ground water from wells (California State Water Resources Control Board, 1990). The estimation of ground-water withdrawals in this report relied primarily on data compiled from a comprehensive land and water survey by the CDWR’s Northern District in summer 2000 for the State water plan update (Todd Hillaire, CWDR, written commun., April, 28, 2003). For each tract of land surveyed, CDWR determined the source of water, crop type and in most instances, the type of irrigation system used. The estimate in this report includes only those lands assessed through the survey as entirely or partially irrigated with ground water.

Irrigation water use on individual parcels was estimated based on the crop type, acreage, crop-water requirements, and irrigation method used. Alfalfa, pasture, grains, potatoes, onions, and garlic represented the predominant crop types, but mint, sunflowers, strawberries, and sugar beets also were grown. Crop-water requirements were derived using published data (Cuenca and others, 1992), evapotranspiration (ET) totals recorded at Bureau of Reclamation Agrimet sites in Klamath Falls and Worden, and information provided by State and county agricultural agencies. The irrigation seasons for certain crop types were determined from Agrimet data, interviews with county and State agricultural agencies, and from published crop reports by the U.S. Department of Agriculture (2004). Sprinkler application efficiency rates were not field measured but obtained from published sources, including King and others (1978) and from communication with State agricultural agents. Efficiencies assigned ranged from 45 percent for gravity systems, to 75 percent for most center pivots, to 90 percent for drip systems. Most ground-water irrigation was done using sprinkler systems. The average application efficiency was estimated to be about 72 percent. Withdrawals were computed by dividing the crop-water requirements by the irrigation method efficiency.

On lands with primary surface-water rights (outside of irrigation districts) and supplemental ground-water rights, the wells were assumed to have been used and were assigned 50 percent of the computed total irrigation requirement. About 620 acres with supplemental ground-water rights outside of irrigation districts were estimated to have been irrigated in 2000. This estimate may be conservative for reasons previously listed.

During the 2000 irrigation season, an estimated 150,000 acre-ft of ground water was pumped to irrigate about 59,600 acres (table 4 and fig. 18). In Oregon, 19,200 acres were irrigated with ground water (about 32 percent of the total), and 40,400 acres (68 percent of the total) were irrigated with ground water in California. Withdrawals in the Butte Valley were about 75,600 acre-ft, or 50 percent of the total in the upper Klamath Basin. All withdrawals for this subbasin were in California. Pumpage in the upper Lost River subbasin, which includes Swan Lake, Langell, Yonna, and Poe Valleys in Oregon, totaled about 28,800 acre-ft, or 19 percent of the total withdrawals. All irrigated acres for this subbasin were in Oregon. In the Sprague River Basin, approximately 11,600 acre-ft of water was pumped, about 8 percent of the total pumpage in the study area. Pumpage in the lower Lost River and Lower Klamath Lake subbasin, which span both States, totaled about 28,600 acre-ft, or about 19 percent of the total pumpage. Ground-water pumpage for irrigation in the Wood River subbasin totaled only about 1,100 acre-ft.

Recent Increases in the Rate of Ground-Water Pumping

The number of acres irrigated with ground water in the Oregon part of the upper Klamath Basin increased by about 1,800 acres per year between 1950 and 2000. Figure 19 shows that the total acres in the Klamath Basin with primary ground-water rights in Oregon increased slowly with time, with periodic larger increases associated with droughts in the late 1960s, late 1970s, and early 1990s. Historical information on ground-water use in California was not readily available. Ground-water withdrawal in the entire upper Klamath Basin during water year 2000 is estimated to have been about 150,000 acre-ft (table 4). The amounts of historical supplemental ground-water pumping are not known, but generally are assumed to be a fraction of the total amount pumped. The number of acres with supplemental ground-water rights is less than one-half the number with primary rights, and supplemental ground-water use presumably occurs only during dry years or parts of irrigation seasons when surface water is not available.

Since 2001, there has been a marked increase in ground-water pumping in the upper Klamath Basin in response to changes in surface-water management and to a series of consecutive dryer-than-average years. The increase is largely due to government programs such as Reclamation’s ground-water acquisition program in 2001 and pilot water bank in 2003, 2004, and 2005 designed to augment surface-water supplies. The pilot water bank was mandated by the 2002 NOAA Fisheries Biological Opinion (National Marine Fisheries Service, 2002) regarding operation of the Klamath Project and its effects on Klamath River coho salmon. Pumping for Reclamation programs, which accounts for most the new use, is metered and therefore reasonably well quantified. The amounts of pumping related to Reclamation programs are shown in table 5. Flow-meter data provided to Reclamation by well owners indicates that the total amounts of ground water pumped for the water bank in 2003 and 2004 were approximately 55,700 and 75,800 acre-ft respectively. The reported pumpage for the water bank in 2003 represents a 41-percent increase over the estimated pumping during 2000 in the upper Klamath Basin. The 2004 pumpage represents a 56-percent increase. The spatial distribution of 2003 and 2004 water-bank pumping is shown in figure 20. Most of this increased pumping (about 61,000 acre-ft) was in the lower Lost River and Lower Klamath Lake subbasins. Ground-water pumping in this area prior to 2001 is estimated to be about 28,600 acre-ft. Therefore, the additional 61,000 acre-ft of water-bank pumping during 2004 represents an approximate 3-fold increase in the total ground-water use in that area. The response to this increased pumping is discussed in the section on hydraulic head fluctuations.

Hydraulic Head Distribution and Ground-Water Flow Directions

Hydraulic head provides the driving force for ground-water flow. Ground water flows from areas of high head toward areas of low head. The change of head with distance is referred to as the “head gradient.” In the uppermost parts of aquifer systems, the head generally follows topography and is highest in upland areas, where recharge typically occurs, and lowest in lowland areas, where ground water typically discharges to streams. Hydraulic head often changes with depth as well as horizontally, so ground-water movement typically includes a vertical component of flow. Vertical head gradients are downward in recharge areas; as a result, the elevation of the static (nonpumping) water levels in wells in recharge areas decreases (becomes deeper) with increasing well depth. Head gradients are typically upward in discharge areas, causing the elevation of the static water levels in wells to increase (become shallower) with increasing well depth. If an upward vertical gradient is sufficient, wells of a certain depth may flow at the surface. Flowing artesian wells are common in parts of the Wood and Sprague River subbasins.

Knowing the distribution of hydraulic head is critical to understanding the directions of ground-water flow. Figure 21 shows the generalized hydraulic head distribution in the upper Klamath Basin. Information used to map hydraulic head is obtained from water wells, springs, and streams. The static water level in a well represents the hydraulic head in the aquifer at the depth of the open interval of the well. Water levels in wells open to more than one aquifer, or to large vertical thicknesses in a single aquifer, represent an average of the heads in the open interval. Springs also provide information on hydraulic head distribution, as springs represent places where the water table intersects land surface. Large-volume springs provide useful information on the head in the regional ground-water flow system. Stream reaches that gain flow from ground-water discharge also provide information on the hydraulic head distribution of the ground-water system. Streams that gain large volumes of water due to ground-water inflow are at or below the elevation of the head in the adjacent aquifer system. The mapping of the hydraulic head distribution in the upper Klamath Basin relied upon data from all of these sources. Water level measurements from approximately 1,000 field-located water wells provided most of the detailed information in populated parts of the basin. In sparsely populated and unpopulated parts of the basin, where wells are scarce, springs provided much of the information. Spring elevations generally were obtained from 1:24,000-scale topographic maps. Gaining stream reaches were used to constrain head elevations in the Williamson River drainage, Wood River Valley, and in the Klamath River canyon below John C. Boyle Dam. Heads shown on figure 21 in the Lost River subbasin from the outlet of Clear Lake Reservoir (Malone Dam) to Olene Gap were modified from Grondin (2004), and contours in the Butte Valley area were modified from Wood (1960).

The head distribution depicted in figure 21 is a generalization. Limited available data prevented mapping all the complexities of the true head distribution. The map depicts the top of the saturated zone as closely as possible, and generally represents the water-table surface. For low-lying areas, the map is based on static water levels in wells and may not reflect water levels in temporarily saturated soil horizons in irrigated areas. For the Wood River subbasin, the map was drawn using wells penetrating an artesian aquifer, where the heads are locally above land surface. Contours are most detailed and have the smallest intervals in areas where data are plentiful, and more generalized with large intervals where data area sparse.

The highest water-level elevations in the upper Klamath Basin occur in the principal recharge areas. These include the Cascade Range, the highland around Medicine Lake Volcano, and uplands along the eastern margin of the basin, including Yamsay Mountain, Winter Ridge, Gearhart Mountain, and Coleman and Barns Rims. Ground water flows from the Cascade Range eastward toward the lower elevations of the basin. From Crater Lake, head gradients are toward Klamath Marsh and southeastward toward the Wood River valley. South of Crater Lake, ground water flows eastward toward the Wood River Valley and Upper Klamath Lake. Where the Klamath River cuts through the Cascade Range, ground-water flow is generally parallel to the axis of the range and toward the river. From the Medicine Lake highlands, ground water flows generally northward toward Butte Valley and the Lower Klamath and Tule Lake subbasins. Head gradients along the eastern margin of the basin are generally westward. From Yamsay Mountain, ground water flows westward toward the upper Williamson River, and southeastward toward Sycan Marsh. Ground water flows from the Gearhart Mountain area generally southwestward toward the Sycan and Sprague River drainages. From the Barns Rim area, ground water flows generally toward Gerber Reservoir and the upper Lost River. In the Modoc Plateau area east and south of Clear Lake, head gradients slope westward toward the Tule Lake subbasin.

Sprague River Subbasin

Hydraulic head data in the Sprague River subbasin are largely limited to valley bottoms and are sparse in upland areas. Head data from wells and springs indicate regional ground water flows generally westward from upland areas around Gearhart Mountain. Between Bly and Beatty, ground water flows toward the valley from local recharge areas around Yainax Butte. Ground water also flows toward this part of the valley from the north. These head gradients are consistent with the observed ground-water discharge to the river between Bly and Beatty. Head data also show ground water flowing toward the Whisky Creek area from Yainax Butte and Bly Mountain. Ground water discharges to streams in the Whisky Creek area. The general pattern of ground-water flow toward the valley occurs over most of the length of the Sprague River, although discharge occurs only at specific locations. An upward vertical head gradient occurs locally in the Sprague River Valley. Numerous flowing artesian wells have been mapped in the area between Beatty and the town of Sprague River (Leonard and Harris, 1974). The localized artesian aquifer consists of volcanic, fluvial, and volcaniclastic deposits confined by fine-grained lacustrine deposits.

Wells are sparse in the Sycan River subbasin, so ground-water flow directions are inferred mostly from spring altitudes and topography. Ground water appears to flow from Winter Ridge and uplands south of Sycan Marsh generally westward. Ground water flows southeastward toward the Sycan River subbasin from Yamsay Mountain. North of Spodue Mountain, the head gradient is generally northward toward the Sycan River. This is consistent with the ground-water discharge at Torrent Spring. West and south of Spodue Mountain, ground water flows generally southward toward the Sprague River subbasin. Piezometer data from Sycan Marsh show a downward head gradient over most of the area (Leslie Bach, The Nature Conservancy, oral commun., 2005). At the northern end of Sycan Marsh an area on the valley floor contains numerous springs, indicating that upward head gradients occur locally.

Williamson River Subbasin

The Williamson River originates near Taylor Butte and flows generally northward between Yamsay Mountain and the faulted volcanic upland to the west. Two local recharge areas exist in the upper Williamson River drainage. Head data from wells and springs, along with precipitation data, indicate that Yamsay Mountain is a significant local recharge area and that ground water flows westward from Yamsay Mountain toward the upper Williamson River. Head data, mostly from springs, indicates that the faulted upland west of the uppermost Williamson River also is an area of local recharge, and that ground water flows from that area eastward toward the Williamson River. Head gradients sloping toward the Williamson River from both of these recharge areas causes the river to gain flow due to ground-water discharge in its uppermost reaches.

Data from wells and springs show that the head gradient slopes toward Klamath Marsh from recharge areas in the Cascade Range. Out on the broad plain east of the Cascade Range, the water table is relatively flat, sloping gently toward the marsh. Ground-water flow directions between Klamath Marsh and Kirk are poorly understood due to the lack of data. South of Kirk and west of Solomon Butte, a substantial southward head gradient exists, indicating that ground water flows from this area toward major discharge (spring) areas along the lower Williamson River and tributaries. Ground water also appears to flow south of Solomon Butte toward discharge areas along the lower Sprague River.

Upward vertical gradients are apparent locally in spring areas near the headwaters of the Williamson River and west of Klamath Marsh, where numerous flowing artesian wells have been mapped. Downward head gradients are apparent in well data along the Williamson River near Sheep Creek and in uplands north of the marsh. Well 28S/09E-20BAB northwest of the marsh had a water-level altitude of about 4,510 ft in July 2006, below the altitude of nearby springs near and at roughly the same altitude as the marsh. The water level altitude in well 28S/10E-27DBD was about 4,503 ft in April 2006, several feet below the water level in the marsh. This raises the possibility of subsurface drainage of the northern part of Klamath Marsh, at least during dry climate cycles.

The lower Williamson River (below Kirk) is one of the most significant ground-water discharge areas in the upper Klamath Basin. Between its descent from Kirk and the point where it emerges onto its delta, the Williamson River is largely confined to a relatively narrow valley. Head gradients slope toward the Williamson River in this area from the east and west

Upper Klamath Lake Subbasin

Well and spring data show that the head gradient slopes eastward from the Cascade Range toward the Wood River Valley north of Upper Klamath Lake. Consequently, many streams emerging from the Cascade Range have large baseflow, and several large springs discharge at the western edge of the valley. A steep gradient toward the basin from the east indicates the potential for ground-water flow across the bounding escarpments from uplands immediately to the east. This gradient from the east appears to extend southward at least to Modoc Point.

A southward hydraulic head gradient from Crater Lake extends to the northern edge of Agency Lake. Between Crater Lake and the northern end of the Wood River Valley, the gradient is 100 to 300 ft/mi. The gradient decreases to about 40 ft/mi from the northern end of the valley to the area of Fort Klamath, where the gradient continues south at less than 5 ft/mi. The head altitudes shown in the Wood River Valley on figure 21 represent water bearing strata (generally sand and pumice) in the upper 100–300 ft of the basin-fill deposits. These strata are overlain and confined by clay layers, resulting in artesian conditions, with heads above land surface over much of the area. This is the principal developed aquifer in the Wood River Valley.

Little head data exist for the ridge east of Upper Klamath Lake; however, springs along the margin of Upper Klamath Lake indicate a head gradient toward the lake, consistent with the area to the north. Head data in the Rocky Point area indicate a gradient toward the lake ranging from more than 100 ft/mi in the Cascade Range to about 15 ft/mi on the valley floor. The gradients toward Upper Klamath Lake from the north, east, and west suggest that ground water is discharging to the lake. South of Upper Klamath Lake, however, the head gradient slopes southeastward toward the Klamath River Valley and the Lower Klamath Lake subbasin.

Upper Lost River Subbasin

Much of head data and analysis for the upper Lost River subbasin are from Grondin (2004). The Lost River originates at Clear Lake Reservoir and winds through a complex of interconnected structural basins north to Bonanza, west to Olene Gap, and then generally southwestward to Tule Lake. Head gradients indicate that ground water flows generally toward the upper Lost River Valley from regional recharge areas to the east and from local recharge associated with Bryant Mountain and other uplands surrounding the subbasin. Although data are scarce, ground water appears also to flow from recharge areas associated with uplands around Yainax Butte and Bly Mountain. The hydraulic head gradient is exceedingly flat in the alluvial valleys of the upper Lost River subbasin; however, the water-table surface likely is variably influenced by pumping. Although the head gradient in the valley is small, it does indicate that ground water moves generally down valley.

Ground-water flow directions in structural valleys adjacent to the upper Lost River subbasin, such as Swan Lake and Yonna Valleys, appear to be generally toward the Lost River. One exception is the southernmost Poe Valley, where the head gradient slopes steeply southward toward the Tule Lake subbasin. Another exception is the southern part of Swan Lake Valley, where the gradient appears to slope steeply toward the southwest. The path that ground water follows out of the southwestern margin of the Swan Lake Valley is unclear, but most likely it is southward toward Olene Gap.

Vertical head gradients are variable in the upper Lost River subbasin. Grondin (2004) evaluated well log information, geophysical data, and water-level measurements, and found downward vertical gradients between basin-fill sediments and the underlying basalt to be common along valley margins and upward gradients in valley centers in the eastern Lost River subbasin. Vertical gradients within the sedimentary section or the basalt were not common. Downward vertical gradients were observed in the Swan Lake Valley, but heads in that area generally were above the elevation of the Lost River.

Klamath Valley

The Klamath Valley area comprises the Lost River drainage from Olene Gap to about Merrill and includes the area immediately south of Klamath Falls and northeast of the Klamath Hills. Ground water flows into the Klamath Valley from the area of Klamath Falls and the uplands to the southwest. Ground water also flows into the Klamath Valley from uplands to the northeast and the Olene Gap area. Local recharge creates a gradient toward the valley from the Klamath Hills as well. Within the Klamath Valley, ground water flows southeastward toward the Tule Lake subbasin. Head gradients on the valley floor are low, averaging 2–3 ft/mi. Southeast of Merrill and north of Sheepy Ridge, the head gradient steepens to about 20 ft/mi. This steepening coincides with a possible subsurface extension of Sheepy Ridge and the margin of the Tule Lake structural basin.

Butte Valley/Red Rock Valley Area

Butte Valley is an internally drained structural basin. The head distribution in the Butte Valley area shown on figure 21 is modified from Wood (1960) and augmented in adjacent uplands using additional well and spring data. Since the creation of Wood’s (1960) water-table map, water levels have declined in parts of Butte Valley as much as 15–25 ft in response to pumping and climate. Declines occurred primarily during 1975–1990 and since 2000. The declines appear to be localized, as water levels have been stable in other parts of Butte Valley. Localized declines since the 1950s have changed the configuration of the water-table surface in Butte Valley, but have not changed the overall regional ground-water flow directions.

Head data show a steep gradient toward Butte Valley from volcanic uplands to the south and west. Gradients from the uplands to the south and around Medicine Lake Volcano range from 100 to 300 ft/mi. In the lower elevation parts of the Butte Valley/Red Rock Valley area, gradients flatten markedly to less than a few feet per mile. Water-table altitudes are between 4,200 and 4,230 ft over much of this area. The water-table surface is not smooth, but is affected by local geology, surface-water hydrology, and pumping. There is a regional northeastward slope to the gradient, indicating ground-water flow from the Butte Valley/Red Rock Valley area toward the Lower Klamath Lake subbasin across the intervening uplands. The gradient steepens beneath the intervening uplands to roughly 50 ft/mi.

Lower Klamath Lake Subbasin

Lower Klamath Lake occupies a structural basin bounded roughly by the Klamath River, the Klamath Hills, uplands to the southwest, and Sheepy Ridge. Hydraulic head data (fig. 21) indicate ground-water flow toward Lower Klamath Lake from uplands just north of the Klamath River. There also is flow from the Butte Valley/Red Rock Valley area southwest of the basin. Heads in the Lower Klamath Lake subbasin and Klamath Valley are similar; however, in the Klamath Hills, which separate the two subbasins, heads are slightly higher, indicating local recharge or regional discharge. The higher heads cause a small gradient toward the Lower Klamath Lake subbasin from the Klamath Hills.

The hydraulic head gradient is very small on the floor of the Lower Klamath Lake subbasin, sloping gently southeastward at 1 to 2 ft/mi over most of the area. East of the Lower Klamath Lake, near Sheepy Ridge, the gradient steepens to approximately 16 ft/mi toward the Tule Lake subbasin, indicating ground-water flow in that direction. This is consistent with isotopic data from deep wells in the Tule Lake subbasin, which indicate the deep aquifer there contains a fraction of water from the Lower Klamath Lake subbasin (Palmer and others, 2007).

Tule Lake Subbasin

Hydraulic head data from wells shows that ground water flows toward the Tule Lake subbasin from the north, east, and west. Ground water flows from the north from the Klamath Valley and southernmost Poe Valley. Ground water flows eastward from the Lower Klamath Lake subbasin into the Tule Lake subbasin as described in the preceding section. A relatively steep gradient of up to 100 ft/mi toward the Tule Lake subbasin occurs in the Clear Lake area and on the Modoc Plateau to the south. The ground-water flow directions inferred from hydraulic head gradients are consistent with isotopic data that suggest that water produced by deep wells in the Tule Lake subbasin originated as recharge in the interior and eastern parts of the upper Klamath Basin with a probable component of water from the Lower Klamath Lake subbasin (Palmer and others, 2007). Head gradients are small on the valley floor, being less than 1 ft/mi in much of the area. Analysis by CDWR (Eaves and others, 2002) indicates that the head gradients in lavas underlying basin-filling sediments slope gently toward the south at 2.5 ft/mi or less. Gradients are much smaller in the shallow water-bearing zones in the sedimentary section, and are generally toward the Tule Lake Sump. Head data from wells indicate ground-water flow southward out of the Tule Lake subbasin, east of Medicine Lake Volcano, toward the Pit River Basin south of the study area.

Subsurface drainage from the Tule Lake subbasin is consistent with historical observations in the area, many of which were summarized by La Rue (1922). Much of this evidence was manifest prior to the draining of Tule Lake, which covered an area of about 150 mi2 prior to draining (according to a 1905 Reclamation survey map). La Rue cited Native American accounts of a whirlpool in the lake, and stated that at high stage water discharged into lava flows along the southern margin (a phenomenon common in the lavas of central Oregon). Early efforts to drain the lake included construction of pits in the lava designed to act as drains. La Rue noted that silt deposits in the lava slope southward, away from the lake. He also considered the fact that “waters of Tule Lake are fresh and the lake bed comparatively free of salts” as proof that the lake “in the past had an outlet.”

Klamath Canyon Area

The Klamath Canyon area includes the reach of the Klamath River between John C. Boyle and Iron Gate Dams. Along most of this reach, the river occupies a steep-walled canyon as it cuts through the Cascade Range. The canyon widens between the upper end of Copco Lake reservoir and Iron Gate Dam, becoming narrow again downstream. Data from sparse wells and numerous springs indicate that ground water flows toward the river throughout the Klamath Canyon area.

Fluctuations in Hydraulic Head

Hydraulic head fluctuates with time in response to external stresses, the most important of which are variations in natural recharge from precipitation, pumping, lake stage, and recharge from canal leakage. These fluctuations are manifest as variations in the water levels in wells.

Ground-water-level fluctuation data are collected by taking multiple water-level measurements in the same well over a period of time. Multiple water-level measurements are available for 257 wells in the upper Klamath Basin. Observation wells are monitored periodically by the USGS, OWRD, and CDWR. Wells have been monitored for periods ranging from less than 1 year to more than 50 years, and measurements have been made at intervals ranging from once every 2 hours (using automated recording devices) to a few times a year. The short-interval measurements effectively create a continuous record of water-level fluctuations.

Ninety-one wells in the basin have been monitored by OWRD, some for periods greater than 50 years. Twenty wells with relatively long-term (10–50 years) records currently are being measured by OWRD (fig. 22). Measurements in those wells generally are made one to four times a year. Sixty-two wells were measured quarterly during this study by the USGS for periods ranging from 1 to 6 years (fig. 22). Nineteen wells were instrumented with continuous recorders, devices that measure and record the water-level elevation every 2 hours. Graphs of water-level fluctuations in all of the wells monitored by the USGS are available on the USGS web site ( Not all wells monitored in the upper Klamath Basin are shown in figure 22. Scores of additional wells have been or currently are monitored by the CDWR and the OWRD for specific purposes. Data from the wells shown in figure 22, which includes all wells monitored by the USGS, provide a comprehensive picture of the dynamic nature of the regional ground-water system.

Climate-Induced Fluctuations

Water levels in most wells fluctuate in response to natural, climate-induced changes in recharge. The greatest response to climate-induced water-level fluctuations in the upper Klamath Basin occurs in the Cascade Range. The response to diminished precipitation (and hence recharge) in the Cascade Range during the current drought cycle is exemplified by the hydrograph of well 30S/07E-06AAA on the lower eastern flank of the Crater Lake highlands (fig. 23). The water level in that well has declined approximately 12 ft since 2000 because of climate-related decreased recharge. On the eastern side of the basin, a similar post-2000 trend exists in well 36S/14E-25BCB (fig. 24), but the magnitude of the recent decline is less. A comparison of these water-level fluctuations with precipitation at Crater Lake in the Cascade Range (fig. 24) shows that periods of rising ground-water levels generally correspond to periods of increasing precipitation, and falling water-levels correspond to periods of decreasing precipitation. Figure 24 also shows that the decadal drought cycles are responsible for the largest water-level fluctuations. During periods of abundant precipitation, the rate of ground-water recharge exceeds, at least temporarily, the rate of discharge. When ground-water recharge exceeds discharge, the amount of ground water in storage must increase, causing the water table to rise. During dry periods, in contrast, the rate of discharge exceeds the rate of recharge, and ground-water levels decline as a result.

Water table fluctuations in response to variations in recharge are most prominent in the Cascade Range, the primary recharge area. Climate-related fluctuations may be difficult to discern in some interior parts of the basin, for two reasons. First, precipitation and, hence, recharge are comparatively small in the interior parts of the basin, so climate-induced water-level fluctuations are correspondingly small. Second, water levels in these areas are affected by ground-water pumping, canal operation, and irrigation, the effects of which can mask the climate signal.

Pumping-Induced Fluctuations

When a well is pumped, the water table near the well declines due to the removal of ground water from storage. A conical depression centered on the well develops on the water table (or potentiometric surface in the case of a confined aquifer) and expands until it captures sufficient discharge and (or) induces enough new recharge to equal the pumping rate. After pumping ceases, the water table recovers as the aquifer returns to pre-pumping conditions. Key factors that determine the magnitude of water-table fluctuations caused by pumping are the aquifer characteristics, the rate and duration of pumping, the presence of aquifer boundaries, and the number of wells affecting the water table in a given area. In aquifers that have low permeability, pumping-induced water-table fluctuations can be large and even interfere with the operation of nearby wells. If the long-term average pumping rate exceeds the rate at which the aquifer can supply water, water levels will not recover fully and long-term water-level declines will occur.

Seasonal pumping affects many wells throughout the upper Klamath Basin. Water-level fluctuations from pumping generally range from a few feet to 20 ft. Pumping effects can be seen in the hydrographs for a well 35S/12E-26DCD near Beatty (fig. 25) and well 40S/12E-32CDB near Malin (fig. 26). Hydrographs for both wells have a steep drawdown curve during the summer followed by a broad recovery curve that rises throughout the winter and spring. Hydrographs for the wells in figures 25 and 26 also show slight year-to-year declines, probably due to a combination of pumping and climate.

Response to Canal Leakage

Water-level fluctuations due to irrigation-canal leakage occur in many wells throughout the irrigated areas in the central part of the study area, with water levels rising during the irrigation season when canals are flowing, and falling when canals are dry. The magnitude of these annual fluctuations varies with the proximity of the well to the canal, the depth of the well, and the local geology. Annual fluctuations due to canal leakage of more than 10 ft have been documented (fig. 27), although fluctuations in the range of 4 to 5 ft are more common.

The water-level response in well 40S/09E-28ADB (fig. 27) is an example of canal and drain influences on wells open to sedimentary materials. This well is constructed into late Tertiary sediment on the northwest flank of the Klamath Hills, about 900 ft from the North Canal. The North Canal is diverted directly from the Klamath River and operates almost continuously. Although the water-level in the well responds to the canal operation nearly year-round, the response is most prevalent during the summer irrigation season. Note that the ground-water response to canal leakage was almost nonexistent in 2001, when no water flowed through the Klamath Project canals for most of the irrigation season (fig. 27).

Ground-water levels can respond rapidly to canal leakage, even at considerable depths, particularly in areas where fractured lava is the predominant rock type. Well 39S/12E-35ABB was constructed to allow separate water-level measurements in two distinct water-bearing intervals. The upper interval, which responds primarily to the canal operation, is open from 92 to 516 ft below land surface. The main water-bearing zone is in pyroclastic material between 305 and 360 ft below land surface, and is overlain by lava flows. The water level in the upper interval of this well responds in a matter of days to the operation of the Langell Valley Irrigation District canal system (fig. 28). The water level starts to rise shortly after the canals start flowing, peaking late in the irrigation season, and dropping soon after canals are shut off for the season. The rapid response of the water table to canal leakage at such depth likely is due to rapid downward movement of water through interconnected vertical fractures in the lava flows. The lower water-bearing interval in the observation well is open from 950 to 1,005 ft below land surface, and is not influenced by canal operation. Both the upper and lower water-bearing zones in this well respond to the pumping effects of nearby wells (fig. 29). Individual wells can respond to both canal operation and pumping.

Response to Lake Stage

As previously discussed, ground water discharges to Upper Klamath Lake. The lake, therefore, represents a local boundary to the regional ground-water system. As a result, water levels in most wells near the lake track variations in lake stage. The water-level in well 35S/06E-10ACC (fig. 30), drilled on the lower northeast flank of Pelican Butte, closely follows the stage in Upper Klamath Lake. The well, about 3,500 ft from the shoreline of the lake, is constructed into layered lava flows that are saturated below a depth of about 470 ft (altitude 4,140). Well 38S/09E-17CBC, located near the Oregon Institute of Technology campus in Klamath Falls, also fluctuates with the stage of Upper Klamath Lake (fig. 31). The well is about 5,200 ft from the lake, and is constructed into interbedded sediment and lava to a total depth of 425 ft.

Long-Term Water-Table Fluctuations

Water levels in wells in the upper Klamath Basin that have been monitored for several decades show fluctuations in response to many of the stresses just discussed. In addition, measurements in most of the wells also reflect decadal scale, wet-dry climate cycles, with some showing the effects of multiyear pumping stresses. Water level trends observed near Bly (well 36S/14E-25BCB), Bonanza (well 39S/11E-20AAD), and the southern Langell Valley (well 41S/14E-08CAA) exemplify areas where ground-water levels are responding mostly to variations in recharge (climate) (fig. 32), showing decadal scale fluctuations of 4–5 ft.

The ground-water flow system appears to be responding to prolonged pumping stresses in several other areas in the upper Klamath Basin, including the area between the communities of Sprague River and Beatty, parts of Butte Valley, south Poe Valley and the area of the Shasta View Irrigation District just north of Malin, parts of west Langell Valley, an area east of Lorella, and the Klamath Valley.

Water levels in two observation wells near the town of Sprague River, 36S/10E-14ACC and 36S/11E-20DCA (fig. 33), have declined over 30 ft since monitoring began in the early 1960s. Leonard and Harris (1974) hypothesized that the relatively steady, and somewhat localized, long-term decline represented a loss of hydraulic head caused by discharge from free-flowing wells. The OWRD later attributed the decline, at least in part, to some wells in the area constructed in a manner that allowed ground water from a the confined basalt aquifer in the lower parts of the wells to flow uphole and into sedimentary units (with lower head) in shallower parts of the well bores (Oregon Water Resources Department, 1987). Borehole geophysical logging has confirmed the interaquifer leakage; however the geologic units receiving the leakage appear to be rhyolitic lavas mapped in the area by Sherrod and Pickthorn (1992) (Mark Norton, Oregon Water Resources Department, unpub. data). For example, geophysical logs for one nearby well show uphole interaquifer flow of about 200 gal/min into a unit with a relatively high natural gamma signature indicative of silicic material. The driller described the material as “broken lava rock.” Water levels in both wells have been more stable since the mid-1990s (fig. 33), likely owing to the ground-water-flow system beginning to reach a new equilibrium.

Water levels in long-term observation well 46N/01E-06N01 in Butte Valley were stable from the 1950s to the mid-1970s (fig. 34). Since 1975, the water level has declined about 20 ft. The trend likely reflects increased pumping stresses during times when precipitation is low, with intermittent times of partial recovery during wet years in the mid-1980s and mid-1990s.

An observation well in southern Poe Valley (40S/11E-11BAD) shows a series of responses to development from which it has never fully recovered (fig. 35). Most notable is the water-level decline of about 20 ft between 1985 and 1995, a recovery of less than 5 ft between 1995 and 1998, and a decline of about 12 ft since 1998. Another observation well (40S/12E-30DCB) about 3 mi south-southeast of the Poe Valley well, in the area of the Shasta View Irrigation District, has a similar water-level trend where the records overlap from 1994 to present (fig. 35). The total decline in well 40S/12E-30DCB since 1998, however, is slightly greater at about 19 ft. The similarity of the head fluctuations suggests that the effects of pumping stresses in both areas may migrate across the subbasin boundary.

Long-term water-level trends in northwest Langell Valley and east of Lorella indicate that pumping stresses in those areas are periodically greater than average (for the area), probably resulting from the occasional use of supplemental ground-water rights. The water level in the observation well in northwest Langell Valley (39S/11E-26ABD) apparently responded to increased pumping during the 1970s, again in the early 1990s, and again starting in 2001 (fig. 36). The ground-water flow system nearly recovered after each of the earlier cycles. In 2005, the water level was about 13 ft below the level measured during 1998, the most recent wet period. A trend similar to that observed in the Langell Valley well, but with a smaller amplitude of fluctuation, is seen in the long-term record at well 39S/12E-35ADD, just east of Lorella (fig. 36). The 2005 water level in the Lorella well was only about 4 ft below the 1998 measurement. Recharge from canal leakage in the area may also be influencing the water-level trend in the Lorella well (Grondin, 2004).

Recent pumping stresses in the Klamath Valley area are reflected in water levels measured in observation well 41S/09E-12AAB (fig. 37). With the exception of a sharp decline in 1970, the water-level trend in the well appears to have been in dynamic equilibrium until 2001. The increased pumping stresses combined with drought contributed to a water-level decline of about 12 ft between 2001 and 2004.

Fluctuations in Hydraulic Head in Response to Recent Increases in Ground-Water Pumping

Prior to 2001, the ground-water system in most of the upper Klamath Basin was in a state of dynamic equilibrium, under which water levels rose and fell in response to climate cycles and seasonal pumping, but generally without chronic long-term declines. (Water levels in some wells near the town of Sprague River [fig. 33] are an exception.) Historically, water levels declined for several years during droughts, but, with local exceptions, water levels eventually rose to (or nearly to) predrought levels during subsequent multiyear wet periods. Wells generally have been drilled deep enough and pumps set low enough to accommodate these historical water-level fluctuations. Pumping in the upper Klamath Basin increased an estimated 50 percent starting in 2001 in response to changes in water management and a prolonged drought. The ground-water system has responded to the increased pumping with water levels showing acute, seasonal, and long-term effects.

Acute effects occur close to pumping wells, generally within hundreds to thousands of feet. These effects typically are the result of the cone of depression of the pumping well spreading to neighboring wells, resulting in a decline in the static water levels, sometimes referred to as “well interference.” These effects typically have a rapid onset and dissipate relatively soon after pumping ends.

Seasonal effects reflect the general lowering of the water table over a broad area (several square miles to tens of square miles) in response to the combined seasonal pumping of multiple wells and, in some places, seasonal variations in recharge. These effects typically build up over the irrigation season and largely recover over the following winter. Figure 38 shows seasonal water-level declines between spring and fall 2004 caused by increased pumping in the basin. Water levels declined more than 10 ft over more than 130 mi2 and more than 20 ft over about 20 mi2 during the 2004 irrigation season. Declines of 10–20 ft are apparent in an area extending from north of the Klamath Hills, through the Klamath Valley, into the northern and eastern parts of the Tulelake subbasin. Smaller areas in the Klamath Valley and the southeastern part of the Tulelake subbasin show seasonal water-level declines exceeding 20 ft in some wells. Seasonal water-level declines of 1–3 ft were measured in most wells distant from pumping centers. These widespread declines are due to natural seasonal fluctuation, possibly amplified by dispersed pumping and ongoing drought. Although a general decline in water levels was measured during this period, levels in some wells that are hydraulically connected to the shallow aquifer system in the basin-fill sediments rose between spring and fall, ranging from a fraction of a foot to as much as 3 ft. This is an annual occurrence entirely due to artificial recharge to the shallow system by canal leakage and deep percolation of irrigation water.

Long-term pumping effects refer to the lowering of the water table for more than a season, often years. Long-term effects can be caused by both climate and pumping stresses. Long-term water level declines typically occur over broad regions, such as an entire subbasin. Long-term decline generally is measured by comparing the spring high water levels each year. Such lowering of the water table has been observed over most of the upper Klamath Basin since about 2000 because of ongoing drought. The only exception is in shallow aquifers in the Klamath Project area, where water levels are maintained by recharge from canal leakage and deep percolation of irrigation water. Long-term declines due to pumping have occurred locally in addition to this drought-related decline. Distinguishing pumping related declines from drought related declines in the basin is difficult because of the scarcity of data from previous drought cycles. However, near the town of Tulelake, where long-term water-level data exist, the rate of the year-to-year decline observed in the present drought cycle in well 48N/04E-35L02 appears to be about twice that observed in the most recent previous drought, from the late 1980s through mid-1990s (fig. 39).

The year-to-year water-level declines can be evaluated by observing the changes in water levels between spring 2001 and spring 2004 (fig. 40). Although data are sparse in the northern part of the area, measurements show that over the 3-year period, water levels declined more than 10 ft in deep water-bearing zones (primarily basalt underlying basin-filling sediments) over more than 135 mi2 of the Klamath Valley and northern Tule Lake subbasin. Declines exceed 15 ft over an area of about 37 mi2 encompassing the State line in the Tule Lake subbasin and extending southward to the town of Tulelake and northward to Malin. Declines of 10–15 ft during this period are common north of Malin. Declines of 5 to 10 ft are common in the southeast part of the Tule Lake subbasin. Levels in three wells on the Modoc Plateau southeast of the Tule Lake subbasin declined 5 to 10 ft (fig. 40). The decline in this southernmost area, where no new pumping has occurred, is somewhat enigmatic, but may indicate that effects are propagating southeastward from pumping centers in the Tule Lake subbasin.

If the post-2000 pumping rates continue in the future, the regional ground-water system possibly will eventually achieve a new state of dynamic equilibrium. This will occur when the depression in the water table is large enough to redirect sufficient regional ground-water flow into the area to offset the increased pumping. At equilibrium, however, the increased discharge in the area of pumping must be offset by decreased discharge elsewhere, likely manifesting itself as a combination of decreased discharge to adjacent basins and decreased discharge to streams, lakes, and wetlands.

Certain details are readily apparent from the recently collected data and existing knowledge of the area. Ground-water pumping is accompanied by declines in water levels that occur at a variety of temporal and spatial scales. The amount of ground water that can be pumped in a period of time will be determined in part by how much drawdown water users and regulatory agencies will tolerate, and in part by how much interference with streams and lakes will be considered acceptable. The drawdown can be easily measured. Where drawdowns acutely affect individual springs, the effects on discharge may be easy to measure. However, where the effects are to larger streams or lakes and represent a small part of the overall flow, they usually are difficult to discriminate from other fluctuations by measurement. Such effects, however, can be calculated using computer models or analytical methods.

Back to Table of Contents

AccessibilityFOIAPrivacyPolicies and Notices

Take Pride in America logoU.S. Department of the Interior | U.S. Geological Survey
Persistent URL:
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:42:09 EST