Scientific Investigations Report 2007–5007
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5007
Back to Table of Contents
Two models in the USGS’s Modular Modeling System (MMS; Leavesley and others, 1996) were used to estimate recharge for the entire aquifer system: Precipitation-Runoff Modeling System (PRMS; Leavesley and others, 1983) and DPM (Vaccaro, 2007). Both models are extensively documented and will not be described in detail here.
The models are driven by daily values for precipitation and for maximum and minimum air temperatures, and partition precipitation into either rain or snow. The models simulate snow accumulation and ablation, plant interception, evapotranspiration, surface runoff, infiltration, water storage in the root or soil zone, and recharge (deep percolation through the bottom of the root or soils zone). The models have similar required input parameters and calculate water balances on the basis of user-defined Hydrologic Response Units (HRUs) into which a watershed or an area is subdivided for model calculations.
PRMS was used to estimate recharge for four, generally wetter, forested upland areas where there are few human activities; these estimates were assumed to be the same for predevelopment and current LULC conditions. PRMS is a physically-based rainfall-runoff watershed model that is fully described and documented in Leavesley and others (1983). The PRMS models were previously constructed (Mastin and Vaccaro, 2002a) as part of the joint USGS and Reclamation Watershed and River System Management Program (U.S. Geological Survey, 1998), and modifications to PRMS for application to the Yakima River Basin are documented in Mastin and Vaccaro (2002b). The models simulate streamflow for estimated unregulated conditions for use in reservoir and river management by Reclamation.
DPM was used to estimate predevelopment and current recharge in 17 areas with extensive human activities, principally agricultural and urban areas. DPM was originally developed by Bauer and Vaccaro (1987) as a tool for estimating daily ground-water recharge over a broad array of landscapes and spatiotemporal scales for the purpose of providing an independent estimate of recharge for ground-water flow models. DPM was later modified by Bauer and Mastin (1997). As part of this study, DPM was modularized and incorporated into MMS (Vaccaro, 2007); the modularization included some modifications. Unlike PRMS, DPM simulates only the land-surface processes and does not simulate the movement of deep percolation below the root or soil zone.
For consistency in calculating recharge between PRMS and DPM, PRMS-recharge was defined as the excess water (deep percolation) leaving the root or soil zone after abstractions by surface runoff and evapotranspiration. The total amount of water that can be stored in the root or soil zone before recharge occurs in both PRMS and DPM is the total available water capacity (TAWC), which is the amount of water that can be stored in the soil column before gravity drainage occurs. TAWC approximates the total unsaturated storage capacity of the root or soil zone because it does not account for the volume of water stored below the wilting point.
The root or soil zone in PRMS is modeled as a single water-storage unit that has an upper part, where both evaporation and transpiration can occur, and a lower part, where only transpiration occurs. The depth of the root or soil zone is determined by the LULC for a HRU. Four LULCs are used in PRMS (bare soils, grasses, shrubs, and trees) and a fifth (water) was added by Mastin and Vaccaro (2002b). A water land cover does not have a root zone, and recharge is assumed to be zero for HRUs with that land cover; moisture additions to water HRUs are due to precipitation and abstractions are due to evaporation. For barren land, the root zone is the soil zone and is that upper part of the soil column where bare-soil evaporation occurs (Leavesley and others, 1983).
The root zone in DPM is the depth of the roots for the LULC of a HRU and for bare soils it is the soil zone, which is the depth of the mapped soil column. DPM currently has 31 LULCs that include a variety of crops, such as beans, grapes, orchards, corn, and hops. Recharge is assumed to be zero for land covers of water and impervious areas (barren rocks and built-up areas—urban areas and high-density commercial or residential). The root zone has a temporally constant depth for such covers as forests, orchards, and sagebrush, and expands with plant growth for such covers as beans or corn. The depth of the root zone is limited by the depth to bedrock for shallow soils, and thus may be less than a particular plant’s root depth. All plant types in DPM have a default value for the maximum root depth, maximum interception capacity, and maximum foliar cover. Interception capacity is the amount of water that a plant can store on its foliage and it varies greatly by plant type and for some plants, by growth stage. Foliar cover is similar to a leaf area index, and is the percent of shading the plant’s foliage provides to the ground; it is used in throughfall and soil evaporation calculations. Users can define these three plant parameters for any plant type contained in DPM because a parameter, such as root depth, can vary depending on a plant’s genetic stock, soils, and climatic setting. For some plant types, these three parameters are adjusted daily using a calculated daily plant growth stage.
PRMS and DPM require landscape characteristics as input parameters for each HRU. These characteristics are the average altitude, slope and aspect, the area, and the x- and y-locations of the centroid of the HRU. For the PRMS models, a digital elevation model (DEM) with a 208-ft grid-cell size of the basin was used by a GIS (Geographic Information System) interface, termed the GIS Weasel (Leavesley and others, 1997; Viger and others, 1997), to calculate HRU values for these characteristics (Mastin and Vaccaro, 2002a). The GIS Weasel is part of MMS and facilitates model development. For the DPM models, 10-m DEMs (U.S. Geological Survey, 2000) were mosaicked and the landscape characteristics for each HRU in each modeled area were calculated using a DPM ‘plug-in’ for the GIS Weasel (R. Viger, U.S. Geological Survey, written commun., 2005).
Daily values of precipitation and maximum and minimum air temperatures for 36 sites were previously compiled in the MMS-input format for WYs 1950–96 (Mastin and Vaccaro, 2002a). There were 17 National Weather Service sites, 12 Natural Resources and Conservation Service SNOTEL sites, and 7 Reclamation sites. Data from these sites were interpolated by PRMS to the HRUs using a distance-weighted scheme. Daily values from these sites for WYs 1997–98 were later compiled by Mastin (U.S. Geological Survey, written commun., 2002). To develop input for the DPM models, records from 13 sites (fig. 5) were either compiled or extended for WYs 1999-2003. These 13 sites did not include the wetter, high-elevation sites because of their distance from the areas being modeled using DPM.
Mean annual precipitation at the 36 sites ranges from about 7 to 128 in. and annual values ranged from about 3 in. to more than 140 in. The large differences in mean annual precipitation between weather sites are clearly indicated by the spatial distribution of mean annual precipitation (Daly and Taylor, 1998) (fig. 5). The daily precipitation has ranged from zero to more than 7 in. Mean annual minimum air temperatures were as low as about 20°F and mean annual maximum temperatures were as high as 70°F. Daily minimum air temperatures were as low as -30°F and daily maximum air temperatures were as high as 110°F.
Adjustments to daily precipitation interpolated from weather sites to a HRU use the ratios of the mean monthly precipitation of the HRU to those at the weather site. Mean monthly precipitation at the weather sites was calculated for the period of record. The mean monthly precipitation at a HRU was calculated using the GIS Weasel ‘plug-in’ from information of Daly and Taylor (1998), which is the mean monthly precipitation values for 4- by 4-km grid cells. The size of the cells results in large changes across cell boundaries, especially where there are large gradients in monthly precipitation. In turn, there are similar changes in the interpolated precipitation values at the HRUs that are near the grid cell boundaries.
Monthly minimum and maximum lapse rates (temperature change per 1,000 ft change in elevation) for both minimum and maximum air temperatures were calculated from the daily temperature data. These monthly values are based on lapse-rate calculations between all sites for each day for each month for the period 1950–98.
Three soil databases were used in this study. The STATSGO database (U. S. Department of Agriculture, 1994) was used to define the soil properties for the PRMS models and for three of the DPM models; these latter three models were for parts of the upper basin where data from the other databases were unavailable (table 1). For all but two of the remaining DPM models, the SSURGO database (Soil Survey Staff, undated) was used. The remaining two models (for the Yakama Nation irrigated lands) used a SSURGO-formatted database (S. Wangemann, Bureau of Indian Affairs, written commun., 2002; table 1). The SSURGO data include information from several survey areas because of the large size of the study area.
The PRMS soil parameters were calculated from the STATSGO database using the GIS Weasel. These parameters include such information as the TAWC, soil depth, the TAWC in the upper part of the root zone where soil evaporation can occur, and soil texture. For the DPM models, soil types were first determined for the three databases. A specified soil type has similar soil attributes throughout and carries these attributes as model input parameters. Thirty-three soil types were defined for the STATSGO database and 119 soil types were defined for the combined SSURGO databases. For each soil type, the soil depth, the number of 6-in. soil layers, and the depth-weighted average of the TAWC were calculated. On the basis of the calculated TAWC, the soil attributes of soil texture, saturated water capacity (the amount of available water storage in excess of TAWC to bring the soils to full saturation—the specific yield), and the lateral hydraulic conductivity were estimated.
A USGS national database (Loveland and others, 1991) was used to estimate a dominant LULC for the HRUs in the previously developed PRMS models. The PRMS models include most of the forested lands in the study area, and land covers of shrub or grass occur mainly in two of the modeled areas. The LULC for a HRU was assumed to be applicable for both predevelopment and current conditions in the four areas modeled with PRMS.
Several LULC databases were used to estimate the dominant current LULC condition for the HRUs in the DPM models, which differentiate among different crop types. Reclamation (E. Young, Bureau of Reclamation, written commun., 2003) provided (1) a GIS database for the basin that identified irrigated lands, but not crop type; and (2) databases for crop types for several smaller areas. The Kittitas Conservation District provided a crop-type distribution for most of Kittitas County (Kittitas Conservation District, written commun., 2003) and the South Yakima Conservation District provided information for Roza and Sunnyside Valley Irrigation Districts (fig. 6) (South Yakima Conservation District, written commun., 2004). A LULC for the basin was obtained from the USGS national database (Homer and others, 2004), and this database also provided information on the location of urban-to-low-density residential areas. Detailed crop-type coverage for a small subbasin was developed by the USGS as part of their National Water Quality Assessment Program. A field survey was conducted for the Wapato Irrigation Project (fig. 6) to identify fields planted in orchards, vineyards, and hops. Additional field surveys were conducted in some areas that were identified as irrigated lands from either the Reclamation or the USGS data, but the crop types were not known. Lastly, a 2004 geodatabase was obtained from Washington State Department of Agriculture (DOA), Pesticide Management Division (T. Maxwell, Washington State Department of Agriculture, written commun., 2005). This geodatabase contained information on size of irrigated fields and crop types, aggregated to the section level; thus, the field location within the section could not be determined. This database provided information on the dominant crop type in areas where the LULC was identified as irrigated croplands but the crop type was not known. Most of the crop types in these areas were small grains, hay, pasture, and row crops. Excluding the DOA database, the LULC information was combined into a single spatial database (starting with the least detailed data and substituting in the more detailed data sets) for developing LULC identification numbers for the DPM models.
The database represents a multi-year (approximately 1995–2004) composite LULC that was assumed to be constant for the period of recharge calculations. Information on temporal changes in crop distributions and types and in the distribution of built-up areas was generally not available, especially for years prior to the 1990s, and it would not be computationally feasible to include such information. The LULC in the models, however, allows potential future use of the models to be representative of existing conditions. The potential error in recharge due to using the composite LULC is described in a subsequent section.
An estimated distribution of the natural (predevelopment) vegetative cover was needed to calculate predevelopment recharge using DPM. Most of the areas with human activities within the DPM models are in the semiarid to arid parts of the basin, where a typical native-plant community consists of intermixed sagebrush and grasses; thus the HRUs with human influences were set to a predevelopment cover type of sagebrush.
In DPM, the HRUs with irrigated croplands require as an input the annual application rate of water. Application rates were calculated using two different methods, depending on whether the crops were irrigated with surface-water (principally located within an irrigation district) or with ground water. For surface-water irrigated crops, it was first determined which district a HRU was in. The boundaries of the irrigation districts (fig. 6) in the study area were obtained from Reclamation (E. Young, Bureau of Reclamation, written commun., 2002) and are based on 1974 maps that show irrigation districts in Washington State. The maps included information on the average on-farm deliveries of water (in/yr) and the effective canal losses (in/yr) for a district as a whole. Together, these values were assumed to be the application rate, and provided a consistent and uniform method to estimate rates. For the irrigation districts, the total application rates varied from 32.4 to 86.4 in/yr, with 22 districts having rates of 36 in/yr and higher, and 16 districts having rates of 42 in/yr and higher. Although the actual water applied may be less than the allowable/estimated rate, Washington State’s water law has a ‘use-it or lose-it’ (relinquishment) provision. Thus, irrigation water above the amount needed for a particular crop type may be diverted for on-farm use but may not be applied to a field, and is instead discharged to a farm drain or wasteway. It was beyond the scope of this study to estimate how much water is actually applied to the more than 500,000 acres of surface-water irrigated croplands; a simplified method to indirectly account for potentially unused (not applied) water is described in the following section.
The application rates input to the model for ground-water irrigated crops were estimated on the basis of the dominant crop type for a HRU (a model input parameter) and the estimated long-term average potential water use of that crop. The average potential water use for each crop type was estimated by operating DPM for WYs 1950–2003 for each crop type and averaging the daily values of potential plant-water use (soil moisture non-limiting). These average values were originally calculated by Vaccaro and Sumioka (2006) as part of estimating ground-water pumpage from the aquifer system. The estimated application rates for the ground-water irrigated crops ranged from 15.0 to 38.9 in/yr.
Built-up areas were assumed to be represented by an impervious LULC; accounting for lawn and landscape irrigation in these areas and subsequent recharge of excess applied water was beyond the scope of this study.
A source of recharge that was not simulated by the models is that from septic-system drainfields. Vaccaro and Sumioka (2006) estimated ground-water pumpage from the aquifer system for the Public Water Supply (PWS) systems and for self-supplied domestic households, which was aggregated by the 2000 census blocks of the U.S. Census Bureau (2004). These estimates were annual values for pumpage by the PWS systems, beginning in 1960, or later if the system was established after 1960, and the values were for 5-year increments, beginning in 1960 for the domestic pumpage. Vaccaro and Sumioka (2006) also estimated the monthly distribution of the PWS and domestic pumpage based on a percentage of the annual value; the percentages were calculated using all available data. The months with the lowest percentage of annual pumpage were November through February, and the values ranged from 4.8 to 5.4 percent. The percentage for March, when outdoor use of water starts, was 5.9. Based on the percentages for November through February and assuming that there is a base-level percentage of 5.5 for the months of March through October, about 63 percent of the annual pumpage is for indoor use. About 90 percent (Solley and others, 1988, 1993) of this value was assumed to be the base-level, non-consumptive use of water that was returned to the ground through septic systems; thus about 57 percent of the annual pumpage becomes septic-system recharge. Multiplication of the annual pumpage data by this value gave a distribution of septic-system recharge from 1960 to 2000 by 5-year increments. Values, including PWS, between the 5-year increments, such as 1961–64, were assumed to be constant from the start of the interval, in this case 1960, until the beginning of the next increment, in this case 1965. Census blocks with 1 acre-ft or less of annual pumpage or that were in the PRMS modeled areas were not included in the calculations. It was assumed that from 1950 to 1960 there was no change in the amount of septic-system recharge. During this period, more that 85 percent of the population growth in the basin occurred in municipalities that do not use septic systems. The basin-wide septic-system recharge from self-supplied domestic pumpage and PWS (non-sewered) in 1960 was only about 7,000 acre-ft and 2,700 acre-ft, respectively, and a small change from 1950 to 1960 is negligible compared to changes in the other components of recharge. The estimates of septic-system recharge for 2000 were a reasonable approximation for 2001 through 2003. The resulting mean annual distribution of septic-system recharge was added to the mean annual recharge values calculated by DPM.
The three-county (Yakima, Kittitas, and Benton) area that makes up most of the study area contains about 9,500 irrigated fields with different crop types and irrigation methods (T. Maxwell, written commun., 2005). The amount of water actually applied to the fields in surface-water irrigated areas is not known, especially on an annual basis during WYs 1950–2003. In addition, the relinquishment provision in State law and the implementation of Best Management Practices in the basin suggests that the water application rate input to the models may be too large. To reduce the total amount of water that can become recharge, a simplified method was employed in which the application of water to the surface-water irrigated croplands was defined in DPM as being supplied to the crops from above their foliage (sprinkler, center pivot, or wheel line). This allows for more evaporation of the applied water from the intercepted water on the plant foliage, and, in effect, reduces the application rate and ultimately the amount of recharge. Depending on crop type, evaporation can be as much as 10 in/yr. In addition, the application rate was identified in DPM to be constant over the irrigation season and was not based on crop growth, which also reduced calculated recharge.
Irrigation of croplands and other human activities occur in some of the areas modeled using PRMS. For two areas with the most human influences (Cle Elum and Wenas, table 1), boundaries were digitized to define model areas, and DPM models were constructed to estimate recharge for current conditions. The recharge simulated in these models was substituted in place of the recharge calculated by PRMS to estimate the distribution of current-condition recharge.
Back to Table of Contents