WMOD00281

• Title: Application of a Simple Local-Scale Numerical Model in the Study of Altered Climate Impacts on Watershed Precipitation
• Prepared By: J. G. Medina
• Date: 1/1/1989
• State: AZ
• Weather Modification - Doc Type: Report

I 3 6 7 9 I I 13 16 17 19 21 23 25 Z1 29 31 33 :Ii
~ ~
<I

...1

'J9
37
36
33
31

21

19
17
16
13
II If
9
7
6
3
I I
I 3 6 7 9 f1 13 16 17 19 21 23 25 'Zl 29 31 33 :Ii

Fig. 3. Model topography for the Delaware River
watershed (contours in feet).

The Arizona study area is most distinguished
by the prominent near-horseshoe-shaped Mogollon
Rim that extends from the north-central part of the
state southward, then turning eastward to the New
Mexico border. The rim is marked by a generally flat
summit and relatively low elevations near 2150 m
( -7000 ft) with tbe exception of two peaks that extend
above 3050 m (-10 000 ft).

The Delaware River study area is marked by
the presence of the Atlantic Ocean on the southeast
and rising terrain to the north, with the highest area in
the northeast at the southern end of the Catskill
Mountains. Elevations at the headwaters of the
Delaware are generally well less than 1200 m
(- 3900 ft).

Regularly scheduled OOZ and 12Z rawinsonde
data were obtained and interpolated to 50-mb
intervals for use by the model. In the case of the
Delaware River watershed, soundings were obtained
from Pittsburgh (PIT), Albany (ALB), and Dulles
(lAD). For the Mogollon Rim, soundings from Desert
Rock (DRA), Winslow (INW), and Tucson (TUS)
were obtained. Additionally, soundings developed by
the MM4 regional model were used as input to the
model. The latter soundings are available at 6-hour
intervals as opposed to 12 hours for the regularly
scheduled observational data. The MM4 was
initialized at 12-hour intervals with European Center
for Medium Range Forecast's large-scale analysis of
January 1979 First Global Atmospheric Research
Project Global Experiment data. Matthews et al.
(these proceedings) discuss the MM4 analysis.

The two-dimensional feature of the model
demands for best results that, for each grid line, there
be representative sounding information at the upwind
border. Interpolation of the sounding data can
substantially affect the performance of the model.
Rhea (1977) interpolated from available sounding data
to 10 border positions and the center of the area of
interest. The border points were selected to well
represent important moisture sources and distribution.
Each grid line was then assigned to a particular
border-point sounding. In the current applications, a
similar procedure was generally applied but with
determination of the 11 soundings by use of a series of
weights based on inverse distance squared to real
soundings and general knowledge of weather
conditions accompanying winter precipitation events.
Assignment of grid lines to a border sounding was
based on distance to border points and general
weather knowledge.

The model requires for each run that one of
the 36 topographic grids be selected. Pittsburgh-
sounding 850-mb winds were found appropriate to
select the grid in the case of the Delaware River, and
the average of Tuscon and Winslow 700-mb winds
were best for the Mogollon Rim. With MM4-
produced soundings, 7oo-mb winds at a point near the
center of the model domain determined the
topographic grid to be used.

The vertical displacement of the air layers as
they cross the underlying terrain is estimated by the
model depending on airmass stability and terrain rise.
In the only attempt to simulate effects of convection,
an additional enhancement of air-layer rise over the
highest terrain is included in the model (thus leading
to additional condensate generation). Model runs
were performed to determine values of parameters
affecting air-layer displacement that led to improved
model estimations of gauge precipitation.

The model can estimate the large-scale
vertical motion provided that at least three real
soundings are available for each ob~ervational time.
For the Delaware River watershed, ::omputation and
inclusion of this factor into the model led to no
improvement, and it was thus eliminated. In fact, best
results were obtained using a single sounding for filling
all model border points as opposed to interpolated
estimates from three soundings. Similar unfruitful
results were obtained from attempts to incorporate the
large-scale vertical motion in the case of the Mogollon
Rim, with the exception of cases utilizing MM4 model-
produced soundings where little effect was observed.
More analysis is required on model runs using MM4
information.

Model efficiency (E) was determined
according to Rhea's procedure (1977). Attempts to
improve model capability by altering E's formulation
have generally produced little gain.