WMOD00383

• Title: Simulation of Winter Precipitation in Western Mountain Watersheds with a Local-Scale Model
• Date: 11/1/1992
• Weather Modification - Doc Type: Report

."..

.
..

Comparison results for cases of the model initialized with MM4
soundings are given in table 2. The LSP~I runs were made at both 5- and
10-kIn resolutions. Correlation values fo:r: daily precipitation were less
than 0.4 at either resolution, indicating little skill in modeling daily
amounts. However, for monthly totals, t:he correlations were 0.71 and
0.72 for the 10-kIn and 5-kIn versions, respectively. These values
represent a bit over 50 percent of the precipitation variance. Though
not a large amount of the variance, they exhibit the useful application
of the models in simulating monthly precipitation.

Table 2. - Comparison of daily and monthly gauge measurements (Gunnison
watershed) with estimates from 5- and 10.-kIn resolution versions of the
model employing MM4 generated soundings for winters (Nov. -Apr.) of 1982-
83 and 1988-89 (period ends with March 1989).

winters
1982-83,
1988-89

Gauge
sum
(rom)

Model
pcp
(rom)

Cases

Correlation
with 0-0
pairs

Cases

Correlation
Without
0-0 pairs

10 km
Daily
Monthly

SkIn
Daily
Monthly

1717

1283

633
21

0.37
0.71

604
21

0.36
0.72

1717

1478

633
21

0.38
0.72

599
21

0.35
0.72

Study of daily results shows that the MM4 generated soundings
exhibit timing (e.g., moisture transport is not properly phased with
precipitation in the gauges)~and relative humidity errors. However, if
the nested models can simulate historical monthly precipitation with an
explained variance of at least 50 percent, and approximately reproduce
the daily precipitation distribution, it seems prudent to further pursue
the nested approach in modeling for a lon.g-term base run and companion
doubled CO2 run.

To study the daily precipitation dist:ribution, sample distribution
plots were developed for the gauge dail~' means and the corresponding
simulated precipitation for the winters of 1982-88, inclusive, for which
local soundings were available. These plots are given in figure 3.
Similarly, figure 4 contains the distribution plots for winter months
with soundings that were generated by MM4. Both figures express
essentially the same features, namely, that the models have simulation
problems in the first three precipitation categories that indicate zero
amounts and two small-amount intervals less than or equal to O.l. Study
of LSPM estimates and corresponding inputs suggests the problem is due
to difficulty in simulating precipitation in low shallow moist layers.
At moderate and higher precipitation amounts, the gauge and model
precipitation distributions are quite similar. These results suggest
that the nested modeling system may be capable of discerning changes in
precipitation distribution character arising from simulated impacts of
doubled CO2,