Selecting Streamflow Statistics

Fourteen statistics were chosen to evaluate how streamflow conditions may differ between defined historical and current time periods for both continuous- and partial-record streamgages. The first twelve statistics are the monthly 1-day low-flow median, which represents the median of minimum 1-day flows for each month. The second statistic is the September median, which represents the median of the daily mean flows for September. Lastly, the 7Q10 was calculated using the annual minimum 7-day average flow and fitted to a log-Pearson Type III distribution curve to determine the 10-year recurrence interval (Riggs, 1972). The September median and the 7Q10 statistics are of extra value because the New Jersey Geological and Water Survey Technical Memorandum 13–3 (Domber and others, 2013) uses LFM, defined as the difference between the 7Q10 low flow and the September median flow, to determine water availability in New Jersey. The LFM is defined in Technical Memorandum 13–3 in part using the September median flow because it is typically the lowest of the monthly flows for New Jersey (Domber and others, 2013). The LFM is a defined buffer intended to allow withdrawals while flows are in an established range to reduce the chances of creating unacceptable ecological effects. If a portion of the available flow was withdrawn, it was assumed ecological effects would be minimized.

Computing updated statistics for this group of streamgages for both historical and current periods provides insight into how conditions have potentially changed. Statistics computed at streamgages not identified as “Regulated” or “Impacted” provide information about current flow conditions and how those conditions may have statistically changed over time based primarily on climate variability, land use changes, and the cumulative effects of smaller withdrawal and discharge patterns in both surface water and groundwater. Along streams where the effects of municipal water withdrawals are known, evaluating the changes in updated statistics and reviewing the actual frequency curves can help water managers better understand the in-stream changes occurring. These changes may be a result of water use patterns and possibly a combination of climate and other factors affecting the stream. Thus, the identification of streamgages with known regulation or impacted flows provides data for understanFiguding the alterations. An example of the annual minimum 7-day average flows fitted to log-Pearson Type III distribution curves (7Q10) for the historical and current periods is presented in figure 2. This figure illustrates how changes in flow patterns can be reflected in distinctly different frequency curve shapes and indicates that conditions are different between these analysis periods.

After updated statistics were computed, they were assessed for normality. A nonparametric approach was selected because the two populations of interest are not normally distributed. The statistics were evaluated for change in distribution using a Paired Wilcoxon Signed-Rank Test. The Paired Wilcoxon Signed-Rank Test is a nonparametric statistical test that compares two paired groups (Helsel and others, 2020). The null hypothesis (probability [p-value] greater than 0.05) for this test assumes that the two populations have the same distribution with the same median. If the null hypothesis is rejected, it indicates that there is sufficient evidence to suggest that the medians and (or) distributions between the two populations differ. The annual minimum 7-day average flow and September median for continuous-record streamgages were also evaluated for trends using the Mann-Kendall (MK) Trend Test, separately, for each period. The MK Trend Test assesses the data to determine if there is a statistical monotonic trend upwards or downwards, that is, if the values are consistently increasing or decreasing over time (Helsel and others, 2020).

Continuous-Record Streamgages

Streamgages were required to have a complete streamflow record in each time period to be used in this analysis. A total of 30 continuous-record streamgages operated by the USGS New Jersey Water Science Center fit this criterion (fig. 1). Twenty-three of these streamgages are in the northern half of New Jersey, above the Fall Line, with varying Valley and Ridge, New England, and Piedmont geology (fig. 1). Only seven streamgages are located within the Coastal Plain province in the southern part of the state. This imbalance in selected streamgages was a direct result of applying the selection criteria. The 7Q10 statistical analysis was performed using approved daily streamflow data and the USGS software Surface Water Toolbox (Kiang and others, 2018). The remaining two statistics of interest, the monthly 1-day low-flow median, and the September median, were calculated. Then, the Paired Wilcoxon Signed-Rank Test was used to identify possible changes in the distribution between the two time periods for each of the fourteen statistics of interest. Percent change was calculated to understand the relative difference at each streamgage for each statistic between the historical and current time periods. Lastly, the MK Trend Test was performed on the annual minimum 7-day average flow and September median for each streamgage for both the historical and current periods. All analyses subsequent to the use of Surface Water Toolbox were performed using the R scripting language (R Core Team, 2022, R version 4.1.1).

Partial-Record Streamgages

The study limited the use of partial-record streamgages to those with at least 10 discrete measurements in both the historical and current periods, or 20 years or less of record. A total of 45 streamgages fit this criterion (fig. 1, table 1). Maintenance of Variance Extension, Type 1 (MOVE.1) or censored MOVE.1 regressions were implemented to calculate the fourteen statistics of interest (Hirsch, 1982). Maintenance of Variance Extension, Type 1 regression is widely used in the hydrologic community to extend historical low flows and low-flow statistics from continuous-record streamgaging stations to streamgaging stations that only have access to a partial record of flows. It establishes a correlation between base-flow measurements at partial-record streamgaging stations and daily mean streamflows measured at index streamgages that exhibit similar streamflow characteristics. Censored MOVE.1 regression was employed when there were many zero-valued flows at a partial-record streamgage. This approach was implemented to avoid low bias in estimating partial-record flows. The regressions were executed using the R-script documented in “Implementation of MOVE.1, censored MOVE.1, and piecewise MOVE.1 low-flow regressions with applications at partial-record streamgaging stations in New Jersey” by Colarullo and others (2018). Colarullo and others (2018) document details on the script, including base-flow screening, weighting scheme, and associated methodologies.

The selection of index streamgages in the MOVE.1 program directly affects the estimated flow statistics at partial-record streamgages. Incorporation of regression error and multiple index streamgages is crucial (Colarullo and others, 2018); however, additional factors such as knowledge of local hydrology and index streamgage conditions can also influence the streamgage selection process, extending beyond the criteria of little or no known regulation. Other criteria considered in the index streamgage selection process include (McHugh and others, 2024):

In order for the MOVE.1 regression to be valid, there must be at least nine measurements used in the regression for each index streamgage, with the average number of measurements used in the weighted regression being at least 9.5. For this study, expert knowledge of partial-record streamgages and index streamgages allowed for detailed verification of the selected index streamgages for the MOVE.1 regression. Based on the aforementioned criteria for selecting index streamgages, the actual number of available streamgages for this study was limited. Ten of the 11 index streamgages are also continuous-record streamgages analyzed in this study (fig. 1, table 1); Stony Brook at Princeton, N.J., (01401000) is not a continuous-record streamgage in this study. Because of a lack of available index streamgages in the unglaciated Piedmont region, the unglaciated New England and the unglaciated Piedmont provinces were combined as one grouping consisting of unglaciated index streamgages (Stanford and others, 1996). This left two more groupings of streamgages for running the MOVE.1 program: the glaciated region and the Coastal Plain (table 2).

The glaciated streamgages are located the furthest north and included the glaciated Valley and Ridge, New England, and Piedmont provinces (fig. 1). The Rockaway River at Main Street at Boonton, N.J., streamgage 01380450 was chosen to represent the glaciated New England province. The Flat Brook near Flatbrookville, N.J., streamgage 01440000 and the Paulins Kill at Blairstown, N.J., streamgage 01443500 represent the glaciated Valley and Ridge province. The Pequest River at Pequest, N.J., streamgage 01445500 represents both the glaciated Valley and Ridge and New England provinces, being located on a divide between these two provinces. Any partial-record streamgages located in the northern glaciated regions of the state were run through MOVE.1 using these index streamgages.

Unglaciated index streamgages incorporated both unglaciated New England and unglaciated Piedmont provinces and are located somewhat centrally in the state or generally south of the most northern glaciated region. The Whippany River at Morristown, N.J., streamgage 01381500 represents the unglaciated New England province. The Stony Brook at Princeton, N.J., streamgage 01401000 represents the unglaciated Piedmont, and the Pequest River at Pequest, N.J., streamgage 01445500 was also used to represent the region between the unglaciated New England and Valley and Ridge provinces. Partial-record streamgages located either centrally or in unglaciated northern regions were run using MOVE.1 through these three index streamgages.

The final grouping of index streamgages is the Coastal Plain, which represent the southern half of the state and both the Inner and Outer Coastal Plain. The Manasquan River at Squankum, N.J., streamgage 01408000, the Great Egg Harbor River at Folsom, N.J., streamgage 01411000, the Maurice River at Norma, N.J., streamgage 01411500, and the North Branch Rancocas Creek at Pemberton N.J., streamgage 01467000 were the index streamgages used to run MOVE.1 for any partial-record streamgages located in these southern regions of the state.

The Paired Wilcoxon Signed-Rank Test was then implemented to test for a change in the distribution between the two time periods. Percent change was calculated to understand the relative difference at each streamgage for each statistic between the historical and current time periods.

Statewide Land Use Conditions for Historical and Current Periods

Land use was evaluated using a snapshot of each study period. Geographic Information Retrieval and Analysis (GIRAS) 1973 (Mitchell and others, 1977) and National Land Cover Dataset (NLCD) 2016 (Dewitz and U.S. Geological Survey, 2021) datasets were obtained, and categories were generalized to match land use types between the two time periods (fig. 4). Because of the limited availability of historical land-use datasets, each period was assessed using data from a time frame of less than 10 years before the end of the period. Land use resolution in the GIRAS dataset is considerably lower than that of NLCD, documented by some discrepancies such as those shown in Figure 5. Streamgage basin land-use characteristics were summarized using the “Tabulate Area” tool in ArcGIS Pro 3.0 (Environmental Systems Research Institute, Inc., 2022).

Land Use Conditions for Selected Streamgage Basins During Historical and Current Periods

Land use was evaluated in each of the 30 continuous-record and 45 partial-record streamgage basins using snapshots from the historical and current study periods (fig. 3). Land use categories were generalized to Water, Urban, Barren, Forest, Agriculture, Wetlands, and Other due to the low spatial resolution of the GIRAS 1973 dataset (fig. 4). Although NLCD 1992 (USGS, 2000) was not used to understand the defined historical or current periods selected by the study, it was used as an intermediate snapshot to assess confidence in GIRAS. Summaries at multiple streamgages indicated similar percentages between NLCD 1992 and GIRAS, providing additional confidence in the land use changes between the historical and current periods.

A number of considerable land use changes were observed. The Haycock Brook at Pompton Lakes N.J., streamgage 01387980 indicated an increase in Water of nearly 20 percent between 1973 and 2016. The Ringwood Creek near Wanaque N.J., streamgage 01384500 indicated an increase in Forest of nearly 77 percent, the Nichomus Run near Woodstown N.J., streamgage 01482510 increased by 64.2 percent in Agriculture, and the Goshen Creek at Goshen N.J., streamgage 01411418 increased by 40.2 percent in Wetlands. Other considerable changes included a 46 percent increase at the Green Egg Harbor River tributary at Sicklerville N.J., streamgage 01410787 and a 60.7 percent decrease at the Rahway River at Rahway, N.J., streamgage 01395000 in Urban between 1973 and 2016. Figure 5 shows the difference in data resolution for this basin, where the GIRAS dataset excludes the detailed water and forested areas captured by NLCD, causing an artificial “increased” urban area for the earlier period.

There are substantial caveats to using these values based on the difference in categorization methods and image resolution. National Land Cover Dataset categories were consolidated to reflect the same organization as the generalized GIRAS categories. In multiple instances, wetland and forest sub-categories were consolidated for the historical period (fig. 4). Based on the inconsistencies between datasets, land use change should be scrutinized. The use of satellite and aerial imagery may be an effective method to supplement this dataset.

Statewide Average Monthly and Annual Precipitation Totals for Current and Historical Periods

Monthly and annual precipitation data were downloaded in raster format at a 4-kilometer resolution from the Parameter-Elevation Regressions on Independent Slopes Model dataset (PRISM Climate Group, 2022) for both the historical and current periods. These 4-kilometer gridded data were clipped to the New Jersey state boundary (U.S. Census Bureau, 2022) and multi-state basin boundaries, and were summarized for each basin using the “Zonal Statistics” tool in ArcGIS Pro 3.0 by Esri (Environmental Systems Research Institute, Inc., 2022).

Average Monthly and Annual Precipitation Totals for Selected Streamgages Basins During Current and Historical Periods

Mean precipitation values from PRISM were calculated monthly and annually for each period, resulting in 26 values for each basin (PRISM Climate Group, 2022): one mean value per month historically and currently (24 values), and one mean value summarizing yearly precipitation for the entirety of the historical and current periods (2 values). The 30-year historical mean annual values ranged from 1,034.3 millimeters (mm) or 40.7 inches (in.) at South Branch Big Timber Creek at Blackwood, N.J., streamgage 01467330 to 1,310.8 mm or 51.6 in. at Belcher Creek at West Milford, N.J., streamgage 01382890. Current period values ranged from 1,123.2 mm or 44.2 in. at Dias Creek near Cape May Court House, N.J., streamgage 01411408 to 1,352.8 mm or 53.3 in. at Belcher Creek at West Milford, N.J., streamgage 01382890. Every basin experienced an increase in mean annual precipitation, with some of the largest increases occurring in central New Jersey (fig. 6).

Of the twelve months, nine months showed an increase in precipitation between the two study periods, whereas February, August, and November showed decreases in precipitation (fig. 7). June and October had the greatest changes, with increases of approximately 24 and 27 percent, respectively. Multiple basins indicated an average increase in both of these months of nearly 30 mm or 1.18 in. between the historical and current periods.

Researchers widely recognize through observation and modeling that a changing climate could likely continue to cause changes in precipitation and extreme weather events (Giri and others, 2020; Georgescu and others, 2021; Rush and others, 2021). These climate variables, coupled with land use change, may affect streamflow characteristics throughout New Jersey through storm-related geomorphic changes, potential drought conditions, variability of water use values, and other changes to topography across the state.

Statewide Water Use Data for Current and Historical Periods

Water use datasets were extracted for both current and historical periods. The current period dataset includes data from 1990 to 2019 taken from the New Jersey Water-Transfer Data System (New Jersey Department of Environmental Protection Division of Water Supply and Geoscience, 2022). The historical period dataset was extracted from an internal USGS database that was compiled as part of the Coastal Plain aquifer project (Zapecza and others, 1987). This dataset was clipped to only include intakes within the continuous- and partial-record basins of interest, and water use data were attached to these basins. Water use per basin was summarized by average volume (millions of gallons) by month and year for each period. Groundwater sites with water use data in both the historical and current periods were extracted to compare water use more directly, and wastewater effluent discharge data were summarized yearly by basin for the current period.

Average Monthly and Annual Water Use Totals for Basins of Selected Streamgages During Current and Historical Periods

Water use data were summarized monthly and annually between the historical and current periods. The historical period water use records included 180 groundwater withdrawal sites within the basins of interest compared to 3,860 groundwater withdrawal and surface water diversion sites in the current period. This overarching comparison includes some caveats. The water use data available before 2003 do not include the same degree of spatial or temporal information as the more recent data. In 1999, the State of New Jersey established the New Jersey Water-Transfer Data System database, which maintains records of groundwater withdrawals, surface water diversions, and wastewater discharges (New Jersey Department of Environmental Protection Division of Water Supply and Geoscience, 2022). In 2003, the New Jersey Environmental Management System was established to provide a more consistent way for tracking water use data at sites with unique identification numbers (New Jersey Department of Environmental Protection, Office of Information Resources Management, Bureau of Geographic Information Systems, 2023).

Additionally, the historical data referenced from Zapecza and others (1987) includes the disclaimer that surface water diversions were not available or included in the dataset and were therefore not included in this analysis. Although some water use data summaries are presented here, these data were included to illustrate the change in magnitude of water withdrawals between periods and not to compute a statistical change in water use volume. When comparing all available New Jersey Water-Transfer Data System data, including both surface water (current period) and groundwater (historical and current periods), water use across the entirety of the study basins increased considerably between the historical and current study periods. As an example of the overall magnitude of increased water use, the July monthly average usage across all basins can be estimated at about 1,383 million gallons during the historical period, which can be compared to 37,405 million gallons in the current period.

However, as previously mentioned, the values presented above are not direct comparisons, as the current period includes both surface water diversions and groundwater well withdrawals, whereas the historical period does not include surface water diversions. Although this substantial difference may be a result of limitations in historical data collection practices, it could also reflect the changes in the types of industries present and an increase in population of more than 1.5 million throughout New Jersey between 1979 and 2019 (U.S. Census Bureau, 2023).

To make a more direct comparison, locations with water use data in both study periods were identified. Of the current period groundwater locations, 73 wells were also documented historically. Only well sites within surficial aquifers (NJDEP, 1998) were incorporated in water use assessments for this study. Basin summaries showed a decrease in mean annual water use at 17 out of 22 basins (fig. 8, table 3). These changes may be attributed to water conservation efforts, efficiencies in technology, and (or) the addition of other water intakes (in other words, new sites) within the basin. Percent change in water use at the selected locations between the study periods ranged from −90.9 percent at Beden Brook near Hopewell, N.J., streamgage 01401520 to 158.5 percent at North Branch Rancocas Creek at Pemberton, N.J., streamgage 01467000.

Trends in the September median values for water use at these 73 wells were analyzed using an MK Trend Test. This test was selected for this study because it is a nonparametric test that does not require the data to be normally distributed and is appropriate for hydrologic data. This test is not subject to influence from outliers or rare events, which can be common in hydrologic data. Additionally, this test assumes that the data are independent, meaning there is no autocorrelation present in the dataset. A trend is considered to be statistically significant when p-value is less than 0.05 (Helsel and others, 2020). No significance was found when the MK Trend Test was applied to individual sites, nor when applied to the sites grouped by basin.

Ultimately, comparing the water use data between the historical and current periods presented in this study proved ineffective in identifying trends over time as a result of the inconsistencies in the two datasets between the historical and current study periods.

Wastewater data were assessed annually by basin during the current period to understand potential effects on water use or flow statistics (New Jersey Department of Environmental Protection Division of Water Supply and Geoscience, 2022). Although wastewater data exist for 49 basins, these data are assumed to have limitations due to missing and incomplete records. No wastewater data are available during the historical period to include in a comparison. When comparing basin-wide wastewater effluent values between 1990 and 2019, 12 basins indicated a decrease in effluent discharge, 16 basins had values of zero (reflecting no data in either year), and 21 basins indicated an increase in effluent discharge ranging from approximately 4 to 4,600 million gallons. Of the 22 basins with groundwater water use data for both periods examined in the study, 20 also had wastewater data. Changes in wastewater discharge data between 1990 and 2019 ranged from −100 percent (Great Egg Harbor River at Folsom, N.J., streamgage 01411000, with a difference of −19.22 million gallons) to 9,400 percent (Rockaway River at Pine Brook, N.J., streamgage 01381200, with a difference of 3,700 million gallons) (table 3). However, owing to the assumed limitations with the wastewater data, their use in this work proved to be inconclusive and limited.

Continuous-Record Streamgages

The 14 statistics of interest selected for this study, monthly 1-day low flow median, September median flow, and 7Q10, were assessed using the Paired Wilcoxon Signed-Rank Test to determine if there was a statistical difference in the data distribution between the historical and current time periods at continuous-record streamgages. Results of this test showed that the January, February, June, September, and December 1-day monthly low-flow medians and the September median had a statistically significant difference in distribution between the two time periods using a significance level of 0.05 (table 4). The September median results showed a decrease in the actual median flow and a considerable change in the reduced range of flow in the current period (fig. 9). The median value for all streamgages was reduced from 46 to 34 cubic feet per second (ft³/s), but the maximum range of September medians was reduced from 127 to 100.8 ft³/s (fig. 9). Figure 9 includes the overall results for the 7Q10 statistic, and also shows that the median and maximum magnitude of the 7Q10 flow decreased between study periods, although the result was not statistically significant. Although the overall comparison of 7Q10 statistics between study periods does not indicate significant changes, the statistics computed at many of the selected streamgages did exhibit considerable changes. These changes are discussed later in this section.

Several of the 1-day monthly low-flow medians were also identified as exhibiting a statistically significant change from the historical to the current period. The overall change for the monthly statistics does not indicate a simple and clear pattern across all months. The medians for January, February, April, November, and December increased slightly, whereas the remaining months showed either no change (March) or a decrease in the median value for the current period (table 4). Also, the changes in January, February, and December were statistically significant as increased flow, whereas the June and September 1-day medians were statistically significant as decreasing flows. Figure 10 shows the boxplots for the 1-day monthly low-flow medians, color-coded by month.

Percent change between the two time periods of interest was also assessed for each of the 30 continuous-record streamgages. A streamgage was identified as having a considerable change if it was greater than 100 percent in the positive direction or 20 percent in the negative direction. Seven streamgages were classified as having a considerable percent change in the positive direction and 10 streamgages indicated a considerable percent change in the negative direction (fig. 11). Nearly all streamgages with a substantial increase in flow value were north of the Fall Line (fig. 11). Several streamgages were identified as monitoring streamflows affected by reservoir operations, or impacted by water use withdrawals and (or) permitted discharge effluent. Human alteration activities like withdrawals from shallow groundwater can affect base-flow conditions while effluent discharges immediately upstream of a streamgage increase the flow passing the streamgage both of which will result in statistics that differ from those computed from unaltered flow conditions (Charles and Nicholson, 2012; Dudley and others, 2020). These statistics are useful for understanding the current period conditions at the streamgages, but the change in flow conditions between periods may not be representative of natural or climate-influenced changes. Many of the streamgages that indicated considerable percent change in the positive (increasing) direction were noted as regulated or impacted.

The MK Trend Test was performed on annual minimum 7-day average flow and September median for each continuous-record streamgage for the historical and current periods to test for monotonic upward or downward trends (table 5). Ten streamgages had a significant p-value for at least one of the time periods for annual minimum 7-day average flow and (or) September median. Every streamgage had a positive trend except for Swimming River near Red Bank N.J., streamgage 01407500 for the historical period for both statistics of interest. Computed low flows at the Swimming River near Red Bank N.J. streamgage are considered unreliable because of a combination of complex conditions related to spillway sensitivity, discharge measurement conditions, leakage, bypass flows, and diversions, which may increase the range of uncertainty and add “noise” to the final record. The Kendall’s Tau correlation coefficient measures the monotony of the slope. It can vary between -1 and 1, where it is negative when the trend is decreasing and positive if the trend is increasing (Helsel and others, 2020).

Table 6 lists the 22 continuous-record streamgages with at least one variable with a considerable positive or negative percent change and the computed change in selected statistics. Examining a subset of the streamgages that are generally unregulated and unimpacted, including several streamgages where regulation is mostly present only during higher flow conditions, reveals a change pattern in 7Q10 discharge values as consistently decreasing. Most changes ranged from about –39 percent to 10 percent between the study periods. Decreasing changes were also noted for many of the other statistics computed at these streamgages. Table 6 includes a column that identifies streamgages where known regulation or impacted streamflow conditions exist, helping users better understand the statistics.

Additionally, two of the statistics computed in this study can be used to estimate the at-site LFM for the continuous-record streamgages. Although this would not be a direct comparison to the methods applied in the NJDEP Technical Memorandum 2013–3 (Domber and others, 2013), it might provide some insight into how low-flow water availability may vary between study periods. A thorough analysis and updated computation of the LFM at selected streamgages was not done as part of this study. However, a simple review of the computed statistics indicates that 10 of the 13 nonregulated, unimpacted continuous-record streamgages exhibited a decreasing change and nine of the 12 continuous-record impacted streamgages also exhibited a decreasing change. The average decreasing change for the 10 nonregulated, unimpacted streamgages was around 22 percent, whereas the average decreasing change for the nine impacted streamgages was around 32 percent.

Five streamgages with considerable percent changes among the 14 computed statistics between study periods were selected and assessed in further detail based on changes in the landscape, precipitation, and water use. These details are discussed in the “Case Studies of Changes at Continuous- and Partial-Record Streamgages” section of this report.

Partial-Record Streamgages

The Paired Wilcoxon Signed-Rank Test was performed for the September median, 7Q10, and monthly 1-day low-flow medians at partial-record streamgages. None of these calculations resulted in a statistically significant difference in the data distribution between the historical and current time periods (fig. 12). As indicated earlier in this report, a MOVE.1 regression, or censored regression, was used to estimate statistics at the partial-record streamgages by correlating discrete discharge measurements with computed daily streamflow at index streamgages. One potential limitation of this approach is the availability of a long-term continuous-record streamgage with data in both periods and similar basin characteristics as the partial-record gages. Although statistically significant changes were not identified for the grouped comparison of study periods at partial-record streamgages, there was a small decrease in discharge for all low-flow statistics except for the 1-day October and December medians and 7Q10 statistics (table 7). The 1-day June median statistic and the overall 7Q10 each indicated an overall increase of 0.05 ft³/s. However, similar to the pattern observed for continuous-record streamgages, the statistics computed for the individual partial-record streamgages varied between periods. The change in 1-day monthly low-flow medians ranged from −100 to 600 percent between the historical and current periods. Changes in discharge between historical and current 7Q10 ranged from −100 to 300 percent. Increases in 7Q10 discharge values were identified at 16 streamgages, whereas 14 streamgages indicated a decrease; all other streamgages had no change or a change from zero. There were also seven partial-record streamgages that had no change in 7Q10 discharge between the historical and current periods. Lastly, there were 10 streamgages that had a change from zero, meaning the results of the 7Q10 statistic during the historical period was 0 ft³/s and changed to a value greater than 0 ft³/s in the current period. The change in September median ranged from −67.7 percent to 450 percent. Twenty-one partial-record streamgages exhibited an increase in the September median, whereas 17 streamgages indicated a decrease. There were also five streamgages that did not have a change between the two time periods, and there were two streamgages that had a change from zero. The partial-record streamgages that had a considerable percent change between the historical and current periods in the negative and positive direction are listed in table 8.

None of the data distributions of the statistics of interest had statistically significant changes at partial-record streamgages. However, five streamgages with considerable differences were selected and assessed in further detail based on changes in the landscape, precipitation, and water use. These details are discussed as case studies at the “Case Studies of Changes at Continuous- and Partial-Record Streamgages” in the following section of this report.

Case Studies of Changes at Continuous- and Partial-Record Streamgages

Five streamgages for both continuous- and partial-record streamgages were selected based on considerable percent change values for any of the calculated statistics. These streamgages were evaluated further to better understand how landscape, environmental, and possible management changes may have affected the flow regime. Supplemental streamgage information is based on observational data and water-year summaries from the National Water Information System water-year summaries for each streamgage (U.S. Geological Survey, 2022a).