WMOD00268 - Laserfiche WebLink

Home Browse Search

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . the mountains. These problems became noticeable about mid-January and persisted into February 2004. The problems were basically solved through making the following changes: allowing the initialization of frozen soil moisture at sub-freezing temperatures; fixing the coding error in the thermal energy content formulation for soil; and switching back to the standard mass- conserving and horizontal-diffusion scheme. These changes were combined with a doubling of the low-level vertical grid spacing (delta-z in the model) that generally prevents the runaway cooling that occurs frequently using that scheme with the original 150 m vertical grid spacing. For the February 14, 2004 real-time model forecast run (still a pretty cold regime), delta-z was increased to 300 m. A test showed that this change lessened, but did not eliminate, the excessive cooling problem; there was still unreasonable surface cooling in the February 14 real-time run. After February 14, there were no additional changes made to the real-time forecast model; all three fixes were operative, with a delta-z of 300 m. Even after the model fixes were implemented on February 14, there still remained a low-level warm temperature bias and a simulated precipitation over prediction bias. 2.8 post-operational RAMS Control and Seeding Runs After the model fixes were implemented in mid-February 2004, the real- time model forecasts improved significantly. However, as CSU began experimenting with the model seeding runs. it became evident that even the improved real-time forecasts were unusable as control no-seed runs. This was because the model code that was developed to simulate seeding effects through a second IFN category was substantially different from the model code used for the real-time forecasts, with inconsistencies in microphysical options that made evaluation of subtle seed/no-seed effects difficult. Due to these inconsistencies and the earlier problems described previously, it was determined that after-the-fact control no-seed as well as seed model runs would have to be performed for the entire DW 2003-2004 Program's operational period (November 2003 through March 2004). This was necessary in order to get completely consistent pairs of control/seed runs, which differed only due to the Introduction of Agl and its activation in seeding runs. The initial sets of control no-seed and seed runs indicated unexpected seeding effects. There were very small differences in simulated precipitation fields. More unexpectedly, the patterns of the difference fields on some of the days were generally organized into positive and negative bands aligned more or less with the mean wind and extending across much of the 3-km fine grid, far upwind and laterally from the target area. Figure 2.15 shows an example of such a seed - control simulated precipitation difference analysis. The targeting wind for this seeding event was 275 to 325 degrees. In this figure, the + signs followed by numbers are locations of SNOTEL sites. The difference scale along the right side of the figure provides for a maximum value of 2.00 mm (0.08 in), but the maximum 24-hr difference indicated in the figure is only about 33 -