Seasonal Water Discharge Prediction: Wyoming River Basins

 

Christopher N. Jones
National Weather Service Office
Paducah, Kentucky

1. INTRODUCTION

The state of Wyoming is comprised of a complex topography of large basins, valleys, and high mountain ranges, creating a diverse climate due to numerous weather conditions. One important issue of Wyoming climatology that is of great importance to the Northern Great Plains and the Far West is water supply. The seasonal snowfall and spring precipitation that occur within the state are important in supplying the Far West and agricultural lands of the plains with water via the Snake, Missouri, and Colorado River Systems. Due to the demand for water throughout the Northern Great Plains and the Far West, it has become increasingly important to monitor the availability of this vital resource. Therefore, it is necessary to better understand the impact of water supply in Wyoming through the use of seasonal discharge values (total discharge for April-September), snow water equivalent (April 1 measurement), and snow course data.

Wyoming is supplied principally with moisture derived from the Gulf of Mexico via lower tropospheric south and southeast winds, and the Pacific Ocean via west and southwest winds. The former is especially important to those areas east of the Continental Divide, including the eastern Wyoming Plains, while the latter provides moisture to those areas mainly west of the Continental Divide. The Bighorn, Wind River, Absoroka, Teton, and Salt River Ranges are the state's dominant mountain features (Figure 1) and, because of their general north-south orientation, serve as barriers to the mean wind flow. The north-south orientations, coupled with the location of the mountainous terrain relative to the principal moisture flow, affect the intensity and duration of winter and spring snows, and summer thunderstorms (Druse 1990). In fact, the prevailing west winds which blow across the state during the winter force moist, low-level air masses to rise along the west slopes of these ranges. Consequently, condensation and precipitation occur, frequently in large amounts. During these winter months the polar jet is commonly found to be present to the south of the state, allowing arctic surges to occur frequently so that most of the precipitation falls as snow. The eastern two-thirds of the state, however, commonly receives much of its annual snowfall during the spring when Gulf moisture is present and upslope conditions occur.

Figure 1 Map of Wyoming River Basins and principal mountain ranges.

Much of the precipitation that occurs within the state during the summer months is the direct result of thunderstorms. Mild weather returns to Wyoming during the late spring and summer as the polar jet stream supplants itself to the north of the state. Once again, it is the warm moist air transported from the Gulf of Mexico that is responsible for thunderstorms being most prevalent over the eastern third of the state. Low-level moisture traversing the Central and Southern Plains, in conjunction with the warmer temperatures that occur over the lower elevations of eastern Wyoming throughout the summer, provide a favorable environment for thunderstorm development. However, it should be noted that these thunderstorms and other rain-producing events account for only 30 percent of Wyoming's water supply; 70 percent is accounted for by snow (Druse 1990). As a result of these topographical and meteorological conditions, the state has been subjected to severe drought and intense flooding throughout its history, especially at lower elevations.

A. Rationale and Purpose of Study

Due to the demand for water throughout the northern Great Plains and the Far West, it has become essential that a greater understanding of the hydrology of Wyoming be realized so that in cases of droughts or floods, effective water management may be implemented. It is logical to propose that if a method could be developed to reasonably predict the amount of water that might be rationed in cases of drought, or allow for appropriate actions concerning potential flood conditions each year, perhaps more effective water management could be applied. Techniques to predict streamflow are already used operationally in some River Forecast Centers (RFC). The RFC use snow course sites with good spatial resoultion (including elevation) for the purpose of statistical water supply forecasting (Garen 1992). That is exactly what this study is designed to achieve. The development of a method by which given data may be analyzed to give estimates of seasonal discharge amounts at numerous points throughout the various river basins of Wyoming. It is believed that such forecasts will provide a basis for water rationing in cases of severe drought, and will enhance the ability of government agencies to provide warnings concerning potential flood situations months in advance.

B. Regression and Correlation Analyses

It should be obvious that in the development of a predictive method, the less that is left to chance, the greater the ability to forecast more accurately. Thus, the methods that are chosen here for field investigations and analyses of the data are regression and correlation analyses.

First, appropriate databases with long-term records need to be utilized so that the full range of hydrological conditions is sampled and can be analyzed. By utilizing the Natural Resources Conservation Service's (NRCS) snow course network for the years 1961 through 1992, it is possible to obtain data concerning the depths of snow and the associated snow water equivalent at numerous sites around the state on given dates. Records of the various sites used in the study were more complete for this period of 31 years, hence it is used here. Typically the annual measurement of snow water equivalent (SWE) occurs on or around April 1. This snow course information is provided to the NRCS through manual observations at the measuring locations, as well as through telemetered data. This process involves the use of satellites which relay data transmitted by signal back to a base station which receives and disseminates the information. Also, the United States Geological Survey (USGS) provides data available from stream gaging stations across the state. Streamflow data for the years 1961 through 1992 are obtained in much the same way as from the NRCS's snow data. However, the stream gaging stations are generally more accessible and conducive to human measurements.

Regression analysis provides the ability to correlate two parameters in the hope that the prediction of one will be possible by a knowledge of the other (Fruend and Simon 1991). It is then possible to determine the strength and validity of the prediction by determining the correlation of the relationship between the two parameters. Thus, through regression analysis, it is possible to develop a method of prediction which can be validated by the coefficient of correlation. The coefficient of correlation, once determined through regression analysis, can then be squared in order to account for the variation in the relationship of the two parameters. Thus, if r is the coefficient of correlation as determined by regression analysis, then r2 is the variation of one parameter which is accounted for by the relationship with the other. Hence, a high value of r2 leads to a low probability of a relationship due to chance, while a low value of r2 supports a high probability for a relationship of chance. Equations 1, 2, 3, and 4 were used to compute the coefficient of correlation, r, (Fruend and Simon 1991) where x is the April 1 SWE, y is the seasonal discharge, and n is the number of observational pairs.

Fortunately, for purposes of this project, the painstakingly long calculations to determine r and r2 were computed using a Harvard Graphics (Software Publishing Corporation 1993) computer software program. The ability to utilize the Harvard Graphics software was made possible using R: Base 4.5++ database programming language written by Dietrich (1994). The program converted the snow course and stream gaging databases to ASCII delimited files which in turn could be used within the framework of the Harvard Graphics software. The software has the capability of not only plotting the graphs used in this project, but it can determine the values of r and r2 as well.

C. Data Gathering Procedures

Both the snow course sites and stream gaging stations supply several types of data. It was determined that the SWE and seasonal streamflow or discharge values were the two parameters most likely to show a high correlation. The filtering of data for the purposes of investigation and analysis are based on the following five steps:

  1. The data from the 150+ snow course sites were to be displayed. The database was printed to hard copy making it possible to look at several stations at once which had not been possible using the computer technology available.
  2. The next step required identification of the major river systems within Wyoming which had the highest degree of impact on the water resources of the state. The Snake, Wind, Upper Green, and Powder/Tongue River systems were chosen because of the increasing demands for water placed on them and their significant role in feeding the aforementioned larger river systems of the northern Great Plains and Far West.
  3. Snow course sites within the previously mentioned mountain ranges which provided a lengthy record of information were identified. Sites were also chosen because of differences in elevation and location, and proximity to known stream gaging stations. It will be seen later how each of these parameters ties in with the ultimate goal of determining seasonal discharge.
  4. Determination was made concerning possible stream gaging stations which could be correlated with the preliminary list of snow course sites. This was done using a prepared list of gaging stations for the entire state of Wyoming, which was available through the Hydrology Station Duty Manual (1992) at the NWSFO Cheyenne. After a preliminary list was formed, some stations were eliminated because of inadequate periods of record concerning discharge values.
  5. Finally, hard copy files were created for the final list of stream gaging stations to be utilized. Each station contained information concerning the peak stages and associated peak discharge along with the date of occurrence and month-by-month discharge values for April through September of each water year.

2. STREAMFLOW ANALYSES OF THE RIVER BASINS

Case studies of four river basins are investigated utilizing the snow course and streamflow data. It is expected that discharge values within basins of western Wyoming will be greatly influenced by snowmelt runoff, while those of eastern Wyoming will be influenced by a mixture of thunderstorm precipitation and snowmelt runoff. The four river basins are presented in the order of expected increasing dominance of snowmelt runoff.

A. Powder/Tongue River Basin

The Powder and Tongue River Basin (Figure 2) is found to the east of the Bighorn Range in north-central Wyoming. The Powder and Tongue Rivers drain into the Missouri River System which is found to the north of the state. The snow course network in the Bighorn Range is extensive. However, many of the sites have records of insufficient length. Thus, only four correlations were conducted for the Powder/Tongue River Basin and these consisted of streamflow measurements made along Big Goose Creek and the Tongue River.

Figure 2. The Powder/Tongue River Basin of North-Central Wyoming.

Correlations among the snow course sites and the stream gaging stations were minimal, and in fact, provided little help in seasonal discharge forecasting. The correlation between Burgess Ranger Station and the Tongue River near Dayton, Wyoming (not shown) supplied the best data for the basin, and the correlation (r2 value) showed that only 28.9 percent of the variation from the regression line was accounted for by the SWE present at the snow course site. Thus, the remaining 71.1 percent of the correlation was due to other factors including spring rainfall and snowfall occurring after April 1. The relationship of the Geneva Pass snow course and Big Goose Creek near Sheridan, Wyoming (not shown) was weak as only 15.6 percent of the variation was accounted for. In addition, a relationship with a correlation of 9.2 percent was noted between Dome Lake and the Big Goose Creek gaging station (not shown). After viewing these data, it became evident that perhaps a better means of discovering a reliable forecasting method could be established if the SWE average of several sites was compared against one gaging station. However, in the case of the Powder and Tongue River Basin such a method seemed difficult to formulate. A correlation using the three previously mentioned snow course sites in the Bighorn Range and the Big Goose Creek gaging station (Figure 3) showed a relationship that was 89.3 percent due to chance. Using these data points, it becomes obvious that seasonal discharge within this basin was primarily related to factors other than SWE.

Figure 3. Burgess Ranger Station, Dome Lake, and Big Goose April 1 Snow Water Equivalent (SWE) average in inches (In) vs. Big Goose Creek Gaging Station near Sheridan, WY April through September Discharge/Runoff (Q) in Thousands of acre-feet (1000s AF). Plotted points are SWE vs. Q for the years 1961-1992. Solid black line is the regression line. Through regression analysis r is determined to be the variation of the relationship between SWE and Q. It is then determined that r2 is the variation of one parameter accounted for by the relationship with the other. To determine the expected discharge for a given water year follow the SWE value vertically until it intersects the regression line. The corresponding Q value is the expected discharge.

Perhaps the most important factor in the contribution of water to the streamflow within the Powder and Tongue River Basin is the high incidence of summer thunderstorms over the north-central and northeastern portions of Wyoming. Summer thunderstorms that form over the basin are capable of supplying copious amounts of water to the basin within a short period of time. Throughout the recorded weather history of Wyoming, this region has been notorious for receiving flash floods which occur as a result of intense thunderstorms (Druse 1990). Another factor is that of a lack of moisture accompanying winter weather systems. Mountain ranges found over the western portion of the state tend to utilize much of the available moisture that accompany weather systems which originate in the Pacific Ocean. Thus, snow depths at sites of approximately the same elevation are often greater at snow course sites in the western mountains of Wyoming. Also, Darrow (1995) showed that heavy snow events (20+ inches) in the Bighorn Range occur with two synoptic patterns: 1) Closed low in the central inter-mountain region and 2) meridional northerly flow (MNF). From December through March, MNF accounts for a majority of the heavy snow events. Lower precipitable water values are associated with the MNF pattern, meaning that SWE values in the Bighorns may be similarly lower due to the water content of the snow. Finally, it must be remembered that snowfall occurring after April 1 is not considered in the correlations. In light of the above reasons and the investigations and analyses shown here, predicting seasonal discharge within the Powder and Tongue River Basin would prove to be difficult based only on snow water equivalent.

B. Wind River Basin

The Wind River is located to the east of the Wind River Range in central Wyoming and flows into the Missouri River System via the Bighorn River. Correlations conducted for the Wind River Basin (Figure 4) suggest that a relationship may exist with sites located toward the northern end of the range.

Figure 4. The Wind River Basin of Central Wyoming.

The Togwotee Pass snow course, site located near the Continental Divide at the northern end of the Wind River Range (elevation of 9580 feet MSL), provided data that proved to be significant for the purpose of this study. The relationship of this site and the gaging station present on the Wind River near Dubois, Wyoming, was significant, with 79.8 percent of the variation accounted for by the SWE (Figure 5). The correlation between Big Warm Springs, a site at an elevation of 8370 feet MSL, and the aforementioned gaging station (not shown), produced a relationship with an r2 value of 66.1 percent. Also, an analysis of T-Cross Ranch (elevation 7900 feet MSL) and the same gaging station (not shown) produced an r2 value of 63.5 percent. Since the sites are close in proximity, it is believed that this difference in chance probabilities can be explained simply because of the difference in elevations of the three sites. Stations at higher elevations receive a greater amount of snow, and therefore may play a larger role in seasonal discharge prediction.

Figure 5. As in Figure 3, but for Togwotee Pass April 1 SE vs. Wind River near Dubois, Wyoming April-Sept. Q.

Switching the focus to the southern end of the range, it became apparent through analysis that the relationship of chance probabilities and elevation did not exist. The correlations of Blue Ridge (elevation 9620 feet MSL) and Hobbs Park (elevation 10100 feet MSL) versus the Little Wind River at Riverton, Wyoming (not shown), also provided little in the way of seasonal discharge prognostication. The analyses gave r2 values of 42.4 and 36.8 percent, respectively, leading to the conclusion that variations of seasonal discharge of the Little Wind River is not principally due to the SWE values of these two sites. Preliminary correlations, conducted with other sites located at elevations near 8500 feet MSL, produced r2 values of less than 10 percent. It is demonstrated here that perhaps the SWE of sites toward the northern end of the range had a greater effect on the resulting seasonal discharge values within the basin than those of sites near the southern end.

Noting that the southern end of the range has a less significant impact on seasonal discharges within the basin and excluding the values of correlation associated with the Togwotee Pass analysis, the intermediate values of r2 that are shown here suggest that a balance exists between SWE and spring/summer precipitation. As was seen with the Powder and Tongue analyses, the fact that the basin is to the east of the Wind River Range leads to a "shadow" effect on winter precipitation amounts. Water precipitates as snow on the west slope and the crest of the range. Thus, snow amounts are less on the east slopes within the "shadow" of the range (Barry and Chorley 1992). During the summer, the transport of moisture to this central basin for purposes of thunderstorm development, is hindered by the mountains located to the south and east. These mountains provide a barrier to moisture from the Gulf of Mexico. Hence, the effect of seasonal discharge values due in part to spring/summer precipitation is not nearly as prominent as that witnessed in the Powder and Tongue River Basin.

It is appropriate to discuss here the correlations of the Togwotee Pass site and the Wind River. Upon investigation of the initial Togwotee Pass correlation, several sites were combined in an attempt to formulate a better seasonal discharge forecasting technique. Using the Togwotee Pass, Big Warm Springs, T-Cross Ranch, and Little Warm snow course sites, a correlation with the Wind River near Dubois was created (not shown). The findings showed that 79.9 percent of the variation in the seasonal discharge data is accounted for by the SWE average of these four sites. This value was almost the exact amount (79.8 percent) of that which was demonstrated by the correlation of Togwotee Pass as an individual site versus the Wind River gaging site near Dubois. Thus, the SWE of the snow course site located at Togwotee Pass appeared to be a tell-tale value in the prediction of seasonal discharge within the Wind River basin.

C. Upper Green River Basin

On the west side of the Wind River Range lies the Upper Green River Basin (Figure 6), which is bordered on the west by the east slopes of the Salt River Range. The Upper Green is a contributor to the Colorado River System, an increasingly vital water source to the arid lands of the Far West.

Figure 6. The Upper Green River Basin of West Wyoming.

First, completed correlations for the Upper Green Basin demonstrated a remarkable similarity regardless of elevation or location within the basin relative to the surrounding mountain ranges. It is believed that this was due to the lack of summertime thunderstorm occurrences within the basin and the diminished barrier effects of mountain ranges to the west. Since the Salt River Range does not exhibit the characteristically high peaks of the Bighorn and Wind River Ranges, variations in seasonal discharge values were accounted for by the SWE at an average of 68 to 74 percent. It should be noted that the records of the snow course sites and the two gaging stations that were used in the investigation of the Upper Green Basin were perhaps the most complete that were applied for analysis throughout this project.

The Kendall Ranger Station, Elkhart Guard Station, and Gros Ventre snow course sites were all correlated individually with the Green River gaging site located at Warren Bridge near Daniel, Wyoming (not shown). The r2 values determined were 68.1, 68.9, and 72.2 percent, respectively. In addition, a correlation was conducted between the SWE average of the Elkhart Park and Kendall Ranger Station sites, in conjunction with the Snider Basin Ranger Station site and the Green River near LaBarge gaging station (not shown). The resulting r2 value of the correlation was 72.0 percent, which is consistent with the values determined for the Upper Green to this point. Also, the average SWE of the Gros Ventre, Loomis Park, and Kendall sites were correlated against the gaging sites at both Warren Bridge (Figure 7) and LaBarge (not shown). The Warren Bridge correlation provided the best analysis within the basin for prediction purposes. However, it had only a slightly greater r2 value of 73.8 percent. The correlation concerning the LaBarge gaging station did not fare much worse as the r2 value was 71.8 percent. However, in the case of the Warren Bridge correlation, data from 1986 were eliminated because of severe flooding that took place in early June of that year as a result of intense snowmelt runoff. On June 6, 1986, this gaging site recorded its peak discharge of 5490 acre-feet (AF), and the discharge for the month preceding the flood was 863,300 AF (Druse 1990). This was more than 13 times the average amount of discharge for the month of May during the preceding 25 years of record! If these data were included in the correlation, the variation of seasonal discharge due to SWE would drop to 37.9 percent. A graph of this correlation with 1986 data is given in Figure 8. Again, since only the April 1 SWE values were used, it is likely that snowfall occurring after April 1 contributed to the high seasonal discharge amount.

Figure 7. As in Figure 3, but for Gros Ventre, Loomis Park, and Kendall R.S. April 1 SWE (Average) vs. Green River at Warren Bridge near Daniel, Wyoming April-Sept. Q.

Figure 8. As in Figure 3, but for Gros Ventre, Loomis Park, and Kendall R.S. April 1 SWE (Average including 1986) vs. Green River at Warren Bridge near Daniel, Wyoming, April-Sept. Q

Since r2 values varied only slightly across the basin, almost any of the correlations discussed thus far could be utilized. Although values hovered near the 70 percent mark leaving some room to chance, the fact that the values were quite similar, supports the view that future forecasts would be accurate.

D. Snake River Basin

The Snake River Basin (Figure 9) of western Wyoming provided the best possibility for an accurate formulation of a seasonal discharge forecasting technique. Thunderstorm activity occurring over the basin is minimal during the summer, and the mountain barrier effect on Pacific weather systems is negligible. The relatively unobstructed transport of Pacific moisture provides for high snowfall amounts for the mountain ranges surrounding the basin. Thus, correlations between snow course sites and gaging stations of the Snake River Basin are among the strongest investigated within the framework of this project.

Figure 9. The Snake River Basin of West Wyoming.

Although the Snake River Basin supplies sites and gaging stations with lengthy records, suitable correlations were difficult to conduct due to the location of reservoirs which regulate water discharge and the location of snow course sites in relation to small tributaries feeding the Snake. Thus, many of the sites located to the north of Jackson were difficult to use for purposes of individual correlations because of the short length of tributaries in that area. However, one site that did show promise as a potential forecasting site was Base Camp, located in the northeast portion of the basin at an elevation of 7030 feet MSL and in close proximity to Pacific Creek. The correlation between Base Camp and the Pacific Creek gaging station at Moran (not shown) yielded an r2 value of 72.9 percent, providing an adequate comparison for forecasting. A snow course site of approximately the same elevation (6750 feet MSL) near Moran showed a similar r2 value of 65.5 percent when correlated against the Snake River above the reservoir near Alpine, Wyoming (not shown).

Upon review of these correlations, it became evident that perhaps a better relationship exists between sites of higher elevations and the gaging stations. This proved to be true with the Togwotee Pass snow course site located on the east side of the Snake River Basin near the Continental Divide. The elevation of this site is 9580 feet MSL. It should be noted that this same snow course site was utilized for purposes of analyzing seasonal discharge within the Wind River Basin. Certainly, Togwotee Pass can be used for investigation of both river basins since data obtained from the site can postulate snow amounts at the immediate region near the site. First, a correlation between Togwotee Pass and a small tributary of the Snake (not shown) showed a reduction in the r2 value (58.0 percent), leading one to suggest that perhaps this is not a tell-tale site for prediction within the Snake River Basin. A correlation of Togwotee Pass and the Snake River gaging station above the reservoir near Alpine (not shown) that suggested otherwise. The r2 value for this particular analysis was 81.5 percent and provided a basis for further investigation of the sites at the upper end of the basin.

Using the SWE averages of the snow course sites of Base Camp, Togwotee Pass, Grassy Lake, and Lewis Lake Divide, a correlation with the previously mentioned Snake River station was performed (Figure 10). It should be stated that individual analyses of the Grassy Lake and Lewis Lake Divide snow course sites were made difficult because of the regulated discharge of Jackson Lake. Although the elevation of the Grassy Lake and Lewis Lake Divide sites are only 7265 feet MSL and 7850 feet MSL, respectively, it was found that the correlation of the SWE average and the Snake River was the highest found during this study. The corresponding r2 value of 82.1 percent suggested that the location of the sites near the upper end of the basin, in conjunction with the effect of increasing elevations, contributed strongly to the strength of the relationship. The SWE average determined in the correlation provided a cross-section of high and low elevation sites which gave representative values in the basin.

Figure 10. As in Figure 3, but for Base Camp, Grassy Lake, Lewis Lake Divide, and Togwotee Pass April 1 SWE (Average) vs. Snake River above Reservoir near Alpine, Wyoming April-Sept. Q.

The correlations of the four sites with the Snake River above the reservoir near Alpine, and that of Togwotee Pass with the same gaging station, are excellent predictors of Snake River seasonal discharge. Utilizing the knowledge attained from the previous correlations examined, it appears that site elevation is an important factor in the prediction of seasonal discharge for the Snake River.

3. RESULTS

The four river basins are presented in order of decreasing correlation between April 1 SWE and seasonal discharge. The graphical results of the four case studies can be used to determine future streamflow. The April 1 SWE (In) value on the x-axis can be followed vertically until it intersects the regression line (solid black). The corresponding discharge value (1000s AF) gives the expected April-September discharge at the gaging station.

A. Snake River Basin

Due to the wintertime transport of Pacific moisture to the Snake River Basin and minimal summer thunderstorm activity, correlation values between SWE and runoff within the basin were the highest determined during the study. Correlations between high elevation sites and the seasonal discharge values were exceptional. A SWE average of several stations rendered an r2 value of 82.1 percent, and a subsequent correlation of Togwotee Pass showed a value of 81.5 percent. Relationships between low elevation snow course sites and gaging stations demonstrated values ranging from 58.0 to 72.9 percent.

The low elevation sites were compared with neighboring creeks, and it is likely that the minimal discharge levels associated with them lead to the lower r2 values. Conversely, the correlation utilizing the higher elevation sites for purposes of averaging SWE data was an excellent prognosticator of seasonal discharge of the Snake River.

B. Upper Green River Basin

The Upper Green River Basin demonstrated remarkable uniformity between correlations conducted for sites of varying location and elevation. Like the Snake River Basin, the Upper Green did not seem to exhibit the barrier effects of the Salt River Range to the west, nor were the effects of summer thunderstorms of considerable concern. Thus, the variation of seasonal discharge due to SWE measurements of the Upper Green yielded r2 values of near 70 percent. Since the Upper Green had been subjected to severe flooding in the past, most notably 1986, the relatively strong relationship that exists between the SWE and the seasonal discharges may prove to be important in flood prediction for the basin. Noting that the values remain similar for correlations made throughout the basin, prediction of seasonal discharge may be quite accurate.

C. Wind River Basin

The Wind River Basin displayed a trend toward supporting the relationship between sites at the northern end of the Wind River Range and seasonal discharge. Togwotee Pass, utilized for the Wind River Basin analyses because of its representation of snow amounts at the northern end, provided the highest individual correlation in the basin. The r2 value of the correlation of Togwotee Pass and the Wind River near Dubois was 79.8 percent, which was closely related to the value of 79.9 percent which was obtained when an average of SWE of northern end sites was conducted. It would appear that this snow course site is crucial in accurately forecasting seasonal discharge of the Wind River. It was also noticed that as elevation decreased at the northern end of the range so did the corresponding r2 values. The correlations involving the snow course sites of Togwotee Pass (elevation 9580 feet MSL), Big Warm Springs (elevation 8370 feet MSL), and T-Cross Ranch (elevation 7900 feet MSL), yielded values of 79.8, 66.1, and 63.5 percent, respectively. These findings were consistent with those found during a study of Colorado runoff completed by Jarrett (1987). That study found that above approximately 7500 feet MSL snowmelt runoff dominates streamflow. Thus, the effect of elevation in decreasing the correlation between seasonal discharge and SWE is supportable.

D. Powder/Tongue River Basin

Investigations of correlations involving seasonal discharge and SWE within the Powder/Tongue River Basin provide little chance of accurate forecasting of seasonal discharge amounts. Contributing to the low values of correlation is the effect of the north-south mountain ranges located to the west of the basin. Moisture available for precipitating over the mountains of the west is lacking in Pacific weather systems which reach the Bighorn Range, and heavy snows associated with the MNF pattern have low water content. Also, summertime thunderstorm occurrence is high over this portion of the state and is a likely culprit in the poor correlations as well. A correlation of the Burgess Ranger Station site and the Tongue River gaging station near Dayton provided the best r2 value which topped out at only 28.9 percent. Values of individual correlations involving the remaining snow course sites were less than 20 percent, while the SWE averaging technique that worked well in the other basins yielded a return of 10.7 percent.

At the southern end of the Wind River Range, relationships between SWE and gaging stations were substantially less. It should be noted that site elevations at the southern end of the range are similar to site elevations at the northern end. Thus, the lower correlation values could be due to the mountain barrier effect of the Wind River Range which robs Pacific weather systems of their moisture.

4. SUMMARY

This study provided detailed analyses of the correlation of seasonal discharge values of four primary river basins of Wyoming: Powder/Tongue, Wind, Upper Green, and Snake. The purpose of this study was to develop seasonal discharge prediction methods which provide insight as to the probabilities of drought and flood occurrences to employ better water management. However, it should be noted that in the case of inevitable, catastrophic thunderstorm events, timely implementation may be of little or no help.

The study was based on analyses of correlations of snow water equivalent (SWE) and discharge values within the following four river basins:

The results of this study demonstrated the degree of accuracy in predicting seasonal discharge values using methods developed for each of the four river basins. In the Snake River Basin snowmelt is the dominant factor in the predictive method, while at the other end of the spectrum the dominant factor in the Powder/Tongue River Basin is likely spring rainfall. On the other hand, the predictive methods for the Upper Green and Wind Rivers showed results of moderate success with the Upper Green River Basin better correlated than the Wind River Basin.

5. ADDITIONAL FINDINGS AND RECOMMENDATIONS FOR FUTURE STUDIES

It became apparent in the examination of numerous snow course sites and gaging stations that certain winters of record demonstrated drought or flood characteristics throughout the state. For many of the sites that were analyzed for this study, the years 1971 and 1972 provided the highest snow amounts. In addition, the Upper Green River Basin noted extraordinary snowfall amounts during the winter of 1986 which consequently led to the severe flooding that was previously mentioned. Also, the years 1973 and 1977 each exhibited a minimum amount of precipitation for many of the site records. In fact, the drought of 1977 affected much of the state, with the Wind River and Upper Green River Basins two of the most severely affected areas (Druse 1990).

The results of the study indicate that although some reliable methods to forecast seasonal discharge have been attained, errors within the methods of the study need to be recognized and additional correlations completed. Although some of these errors have been observed, testing of the developed forecasting methods needs to be implemented. One such error that was noticed is that snowfall occurring after April 1 is not taken into account in the SWE measurements. Thus, correlations of some sites could be greatly influenced by the absence of such data. Preliminary results from the Wyoming flood episode of 1995 indicated that the April 1 SWE values would likely underestimate the seasonal (April-September) discharge amounts. In early May 1995, a copious snowfall was recorded in the Wind River Range, a snowfall not represented in the April 1 SWE value. The SWE at the Togwotee Pass site for April 1 and May 15 differed by approximately 32 percent. Perhaps SWE measurements conducted between May 1 and June 1 could give better estimates of seasonal discharge amounts. Consideration must also be given to the utilization of higher elevation sites later in the forecast season (Garen 1992).

Also, contributing to poor correlations is the effect of droughts and floods on stream gaging data. The example of the 1986 Upper Green River Basin snowmelt runoff is one such anomaly. Case studies of smaller tributaries within the four analyzed river basins need to be more fully examined to determine the extent of their contributions to seasonal discharge. In addition, other river basins of Wyoming such as the Lower Green, Bighorn, North Platte, and Upper Yellowstone need to be investigated so that prediction of seasonal discharge amounts can be made. Obviously, as more correlations are completed, the chance for effective prediction of seasonal discharge increases.

Additional correlations can be conducted concerning snow depth of sites located in the various mountain ranges of the state which could lead to a better understanding of snowfall patterns of the Wyoming ranges. Correlations of east and west facing slopes could be investigated, as well as analyses of north versus south facing sites. It has also been demonstrated how thunderstorm precipitation might influence seasonal discharge values of gaging stations, particularly those located over eastern portions of the state. Thus, a comprehensive study concerned with SWE and spring/summer precipitation could be conducted for purposes of formulating additional methods of prediction. The spring/summer precipitation totals for varying locations around the state could possibly be obtained through government agencies such as the USGS or through the National Weather Service's cooperative observer program. Findings of this type of study could support the contention that thunderstorm contamination exists and may be a significant factor. Investigations relating to snowpack management and subsequent water diversions from the resulting snowmelt should be continued in order to maximize effective water management in the areas of agricultural, industrial, and municipal uses. Future studies of other factors related to seasonal discharge should also be investigated. These could include regional geological analyses related to groundwater, soil moisture content, and the effects of reservoir storage.

6. ACKNOWLEDGMENTS

The author wishes to express his sincere thanks to Mr. Tom Dietrich, Service Hydrologist at the National Weather Service Forecast Office, Cheyenne, Wyoming, for his advice and help concerning research for this paper. His support and efforts made the gathering, assemblage, and completion of this study possible. Thanks are also in order to Dr. William Hoyt of the Earth Science Department of the University of Northern Colorado for his suggestions and comments concerning this study. Also, an appreciation to Dr. L. Glen Cobb, Dr. D. Andre Erasmus, Dr. Doug Wesley, Ed Berry, Noreen Schwein, Pat Spoden, and Christie Malnati for their useful suggestions that improved the manuscript.

7. REFERENCES

Barry, R.G., and R.J. Chorley, 1992: Atmosphere, Weather, and Climate. Routledge, 392 pp.

Darrow, M.A., 1995: Flow Patterns Conducive for Heavy Snow in the Northern Big Horn Mountains of North-Central Wyoming. National Weather Digest, 19, 2-9.

Dietrich, T., 1992: Hydrologic Station Duty Manual. National Weather Service Forecast Office, Cheyenne, WY, 95 pp.

____________, 1994: R:Base 4.5++. Version 1.0 Computer Software. IBM Version.

Druse, S.A., 1990: Wyoming Floods and Droughts. National Weather Summary 1988-89, USGS Water Supply Paper 2375, 575-582.

Fruend, J.E., and G.A. Simon, 1991: Statistics: A First Course. Prentice-Hall, 584 pp.,

Garen, D.C., 1992: Improved Techniques in Regression-Based Streamflow Volume Forecasting. Journal of Water Resources Planning and Management, 118, 654-670.

Jarrett, R.D., 1987: Flood Hydrology of Foothill and Mountain Streams in Colorado. Colorado State University, 200 pp.

Software Publishing Corp., 1993: Harvard Graphics. Version 2.0 Computer Software. Software Publishing Corp. IBM Version.

 


USA.gov is the U.S. government's official web portal to all federal, state and local government web resources and services.