Laserfiche WebLink
<br />,,-' <br /> <br />... <br /> <br />divide in west-central Colorado. Higher E~levations within the watershed <br />exceed 4 km. On the other hand, the Arizona study area, which consists <br />of the Mogollon Rim and the surrounding ,area, is marked by a generally <br />flat summit and relatively low elevations near 2200 m. <br /> <br />Input Data <br /> <br />Regularly scheduled 0000 and 1200 U'l'C soundings taken at GJT, SLC, <br />and INW, from January 1982 through March 1989, were used as input data <br />in the Gunnison study. The MM4 runs contained soundings generated every <br />six hours for the period 1982 through 1983 and 1988 :through March 1989. <br />The MM4 was initialized at 12-hr intervalf~ from the large-scale analysis <br />generated by the European Center for ~Iedium Range Weather Forecast <br />model. Matthews et ale (1992a, 1992b;1 discuss the use of MM4 in <br />application to the Gunnison watershed. In the case of Arizona, <br />regularly scheduled soundings from ORA (Desert Rock), INW, and TUS <br />(Tucson), for the period November 1988 through April 1988, were used as <br />input in the application of the LSPM. <br /> <br />Gauge measurements from the higher elevations in or near the <br />Gunnison watershed were used for comparison with model estimates <br />developed for those locations. The 14 gauges used in the study are <br />given in figure 2. The mean of the nonmis:sing gauge values for each day <br />became the actual basis of comparison. Sums of the daily means were <br />used for monthly time unit comparisons. It was felt that the mean of <br />the high-elevation gauges would better indicate model performance over <br />the watershed. Comparisons with indbridual gauges confirmed this <br />assumption as results varied considerably, likely because of gauge <br />location effects, measurement errors, and model inaccuracy. A similar <br />approach was used in the Arizona study in using 11 gauges located on the <br />Mogollon Rim. <br /> <br />RESULTS AND DISCUSSION <br /> <br />Model runs with the LSPM were made fQr the Gunnison watershed at S- <br />and 10-km resolutions using local soundings taken 1982-1989, inclusive. <br />Table 1 contains results from the compsLrison of precipitation gauge <br />means and corresponding model means for daily and monthly totals. <br />Correlations are given for each resolution., considering the full period, <br />winters, and summers. Precipitation amounts and sample sizes are <br />included. The table also contains some r~~sults from modeling the Salt- <br />Verde watersheds at 10-km resolution for winters of 1978-88. <br /> <br />Gunnison daily 10-km results for full-periOd, winters, and sununers <br />yielded correlations of 0.59 (R2 = 0.35), 0.64 (R2 = 0.41), and 0.53 (R2 <br />= 0.28; all three cases with 0-0 pairs included), respectively. The <br />corresponding monthly results were correlations of 0.82 (R2 = 0.68), 0.84 <br />(R2 = 0.71), and 0.76 (R2 = 0.57), indicative of the LSPM's usefulness. <br />Clearly, the model performs best for the orographic conditions that most <br />frequently occur during winter. At 5-km resolution, correlations were <br />essentially the same as those obtained in the 10-km runs. Gauge and <br />simulated precipitation differences were less than 10 percent of the <br />gauge totals (actually 6.1 percent). <br /> <br />The results that included the 0-0 pairs are of most interest <br />because the model must simulate all types of weather cases, including <br />