Special session of California 2000 field study - Air Pollution Modeling and Its Application - page 523

RAWS sites, such that simulations characterizing one station type well necessarily
handled the other poorly . After little difference overnight, the average RAWS
observation was 4°C warmer than ASOS by local noon ( Fig. 2c) . Even after
adjustment for elevation ( Table 1) and timing issues, the former addressed via a
32-station, lower altitude RAWS subset ( Fig. 1) , a substantial discrepancy bet-
ween RAWS and ASOS averages remained (Fig. 2c) .

41N

40.5N

2400
2100
1800
1500
1200
900
600
300
0

40N

39.5N

39N

38.5N

38N

37.5N

37N

36.5N

36N

35.5N

ASOS

RAWS

35N

RAWS (lower elevation)

34.5N

124W 123W 122W 121W 120W 119W 118W 117W

Fig. 1. Terrain of 4 km domain with locations of ASOS and RAWS stations shown

This analysis raises questions about these remote, unattended RAWS sites,
which appear to permit greater variation in temperature sensor placement height.
The most skillful simulation with respect to ASOS stations alone (WRF-55,
Table 2) evinced a warm bias overnight for RAWS stations and a pronounced cold
bias for the following afternoon (Fig. 2d) . This suggests RAWS sensors may be
systematically placed too close to the ground, invalidating model physics similarity
assumptions. Subsequent analyses involve ASOS stations only.
The WRF-55 simulation, which used WRF v.3.1, was an unnudged 48 h run
from 00Z July 31 using the QNSE PBL scheme and CAM radiation. This was one
of the few WRF runs that bested MM5 without resorting to nudging. We hoped
that skillful WRF simulations could be obtained without that commonly employed
technique for model limitation compensation. Further work is needed to determine
if this physics combination has wider applicability.