WMOD00273 - Laserfiche WebLink

Home Browse Search

is highly variable, a conservative approach to exclude potential contamination periods is warranted. Since the wind direction during San Juan winter storms usually has some westerly component, all days were excluded during which generators CCB or LS (Fig. 1) were operated. These generators were west and therefore upwind of both control areas during westerly flow. There was also some concern about contamination of the Lone Cone control area by generators BMY, EGO, and DS under southwest flow. But given the greater intervening distances between these generators and the control area, the days when these generators operated were retained in the analysis. b. Snow Data Assimilation System The primary tool in this evaluation was the Snow Data Assimilation System @NODAS). The SNODAS is a spatially distributed snow energy and mass balance model that routinely assimilates all available snowpack observations. The SNODAS output variables include snow water equivalent (SWE), snowpack melt, and pack temperature, on a 1 km grid nationwide. Reclamation posts daily SNODAS graphics for Colorado on a web site. SNODAS was used for a recent estimate of water augmentation potential from winter seeding in the Colorado River Basin3 of one million acre-feet of additional snow water in an average year. Since this estimate was close to those from earlier studies of augmentation potential in the basin, there was added motivation to use SNODAS for the present analysis. For this study, SNODAS 24-hour SWE change from the seeding target area and control areas were compared to ascertain snow water augmentation from operational seeding. WWC has traditionally assessed seeding impacts with data from the Snow Telemetry (SNOTEL) system. SWE, depth, and snow precipitation are measured at these sites, which are located in mountainous areas. SNOTEL snowfall and SWE measurement resolution is limited and sites are sparse compared to SNODAS 1 km grid points (Fig. 2). Nevertheless, SNOTEL data are assimilated by SNODAS so the two are not independent of each other. One would expect that SNODAS SWE would compare well with that of SNOTEL at most SNOTEL locations in Colorado, and previous unpublished analyses by Hunter have verified this expectation. Furthermore, a study by the SNODAS developers shows good agreement of the system's SWE output with that of a well-accepted land surface model and physical snow pit measurements, at three separate locations in Colorado. We are restricted to using the 24-hour SWE change (dSWE) from 0600 to 0600 UTC, since there are no other SNODAS data periods or frequencies available. Therefore at a minimum, we chose the 0600 UTC (2300 MST) ending time after all seeding had ceased. The rationale for this decision was that if there was a difference between seeded and control areas, it would be evident at this time. If the seeding spanned multiple days, intermediate days were also included in the sample. Actual seeding durations varied with time and according to each generator site; of course these durations did not necessarily match with the 0600-0600 UTC days. The dSWE values were obtained from an areal average of all 1 km pixels in each area. Therefore they are normalized for area. Nevertheless, note from Fig. 1 that the San Juan target area is much larger than either control area. The control areas will therefore have much less 4