WMOD00275 - Laserfiche WebLink

Home Browse Search

;;,..Il lC.,.q ;- .~- 707 AUGUST 2006 DRESSLER ET AL. 300 Ei' c Q) ::J C" ~ Q) 0.5 > :;::> tV :; E ::J (,) -..- Colorado Basin - - snow-covered area .......... snowcourse - SNOTEL sites o o 2000 3000 elevation, m 4000 1000 FIG. 2. Cumulative frequency of recording SNOTEL and snow course site elevation locations relative to Colorado basin topog- raphy and as masked by the snow-covered portion of the basin. This study examines SWE measurements within and adjacent to the Colorado River basin using 240 SNOTEL and 500 snow course sites, 93 of which are collocated (Fig. 1). The period of 1990-99 was exam- ined because of an increase in the number of SNOTEL sites reporting after 1990 and as a complement to the earlier study of Serreze et al. (1999), which compared collocated SWE values for 1980-87 within the same domain on 1 April, a date generally corresponding to peak SWE. Within the 1990-99 period 3 yr are analyzed in more detail-those representing above-, near-, and below-average accumulation years. While there were differences across the basin, 1993 was typically an above-average year, whereas 1998 and 1999 were near and below average, respectively. A comparison date of 30 March was chosen to both approximate peak SWE and to coincide with the date of the maximum number of stations recording snow. Comparing the elevation distribution of the 240 SNOTEL and 500 snow course sites with that for the snow-covered part of the Colorado River basin, it is apparent that SNOTEL sites are clustered at middle to higher elevations and snow courses are more represen- tative of the basin elevation and, therefore, snow area, as a whole (Fig. 2). Snow courses, however, under- sample elevations above 3500 m (only 16% of the area in the Upper Colorado but with a lot of the snow). Data quality control procedures followed those de- veloped by Serreze et al. (1999) for SNOTEL data, which masked outliers and eliminated negative SWE values. Stations with missing values for the first 15 days of the water year were assumed to have had delayed servicing and data for that year were not used. Daily ~ c rn C> c :e 200 o o ~ rn c o ~ 100 0 ..... o oDD I.. Q) .0 E :J C 00000 AAAAAAA~AAAAAAAAAAAAAA ODD ODD o . SNOTEL C snowcourse CCDCD o 12/29 3/9 4/13 date 5/18 2/2 FIG. 3. The 1993 weekly SNOTEL and snow course reporting stations (after quality control), as used for hypsometric interpo- lation. The number of snow courses on this graphic was calculated from a linear interpolation between snow course measurement days (linear fill). Additional mixed datasets (data not shown) in- cluded SNOTEL plus collocated snow courses on measurement day only and SNOTEL plus snow courses for all days, with linear interpolation of snow course values between measurement days (linear fill). SWE increments greater than 25 cm, or consecutive days with increases and subsequent decreases each greater than 6 cm, were deemed erroneous and were not used. Monthly SWE decreases more than five stan- dard deviations from the mean were deemed to be er- roneous, and monthly SWE increases more than five standard deviations from the mean without a compa- rable extreme value for precipitation or a correspond- ing precipitation increment of more than three standard deviations were deemed to be erroneous. When erro- neous data were identified at a site, subsequent SWE measurements for that water year were not used in or- der to eliminate the contaminating effect of an indi- vidual erroneous value. Point values were interpolated using hypsometry (SWE regressed with elevation) and elevation-de- trended residuals, as in Fassnacht et al. (2003). The SWE was regressed based on station elevation and ap- plied to the 1-km grid domain using a 1-km digital el- evation model. A new regression was calculated for each observation day to account for any changes that meteorological factors had on the snowpack. Only ac- tive recording stations were considered in the regres- sion calculation for each day (Fig. 3), because some