WMOD00381 - Laserfiche WebLink

Home Browse Search

r 1" ~'I~ j , - were evidenced by the special R/S observations. . While these errors are, in the mean, relatively small, large errors in a case by case situation can exist, suggesting the dangers of using i,nterpolated values as opposed to , .\ I :1 1 l observed data in operational forecasting deci.sions. The changes in performance of the orographic precipitation model can be summarized as follows: (I) the use of serial 3 hourly Durango soundings from the Colorado River Basin Pilot Project (CRBPP) increased the variance i.n daily precipitation explained by the model to 69% over 32% explained using only interpolated NWS R/S; (2) the use of the special Craig soundings (launched with, at best 6 hourly sequences) improved model performance from 10% to only 22% of th~,~~riance explained. Three factors responsible for the poor model performance in the Craig vicinity were isolated. (1) Only one precipi- tation measurement (Rabbit Ears Pass) was available for model verification in the north as opposed to 5 special CRBPP gauges in the south. (2) The number of special Craig soundings per precipitation event was much fewer than the CRBPP soundings. (3) A real "dry" airmass bias is introduced by using only the interpolated NWS R/S in this region, i.e. Lander and Denver are immediately downwind of major mountain. barriers, and Grand Junction is often too far south to detect true moisture conditions over the northern Colorado Rockies. Several case studies depict this real problem. These results point out the need for several special rawinsonde sites in any design for weather modification programs in the Upper Colorado River Basin. These sites must be upwind of major precipitation-receiving mountain barriers and able to detect significant moisture fluxes into the target areas. The special R/S must be launched with at least a 3 hourly frequency during precipitation events for forecasting and evaluation schemes ':..-'~~