WMOD00472

• Title: Estimation of Cloud Liquid Water in Winter Storms on the Mogollon Rim
• Date: 5/1/1993
• Weather Modification - Doc Type: Report

'The results in table l(c) suggest that local-sounding lower-level relative humidity, such as from
CVR, when combined with INW lower-level wind speed, can yield useful estimates of the HJ
CLW as measured by the vertically-pointing radiometer. The variance explained was 61
percent, a noticeable amount particularly for CLW, a complex and highly variable cloud
feature. The application of this analysis process on local and INW sounding data, to estimate
CLW, should be useful to cloud seeding studies particularly when used in combination with
other sensor cloud measurements.

Because the CVR data set is limited to the winter of 1987, MLR application to sounding data
Bolely from INW and TUS was pursued. Given in table 1 are the results obtained with a
number of independent variable candidates where, as previously, HJ CLW is the dependent
variable. A best multiple cOITelation coefficient of 0.67 was obtained with the mean ofTUS 700-
Imd 750-mb relative humidity and INW 750-mb wind speed. The variance explained by the
more remote sounding data is lower at 45 versus 61 percent when CVR data was included. The
MLR establishes the coefficients by which estimates of HJ CLW can be developed with
soundings of the 10-winter data set.

'rhough the INW-TUS MLR equation represented effects of relative humidity and wind speed,
some temperature profile effects on HJ CLW may have been absent and might have improved
results if included. Although some incorporating expression could be developed through a
selection and testing process, an alternative and more sophisticated approach was the
a.daptation and application of an existing operational model that estimates CLW. This avenue
was pursued with expectations that such a model could lead to improved estimates.

Table 1. - Multiple regression results with sounding derived relative humidities (RH) and
wind speeds (WS) as independent variables and HJ CLW as the dependent variable!.

Independent
variables

Multiple
correlation

Adjusted R2

Root mean
squared error

(a)

Xl: Mean ofCVR 700- and 750-mb RH
Xz: Mean ofCVR 700- and 750-mb WS

0.67 0.44 0.043
Regression equation: Y' = -0.0654+0.1306XI+0.0033Xz

(b)

XI: Mean of CVR 700- and 750-mb RH
Xz: INW 750-mb WS

0.79 0.61 0.037
Regression equation: Y' = -0.0753+0.1375XI+0.0043Xz

(c)

Xl: Mean ofTUS 700- and 750-mb RH
Xz: INW 750-mb WS

0.67 0.45 0.043
Regression equation: Y' = 0.0512+0.1262XI+0.0039Xz

1 Dependent variable - a HJ CLW estimate cOnsists of the mean for 7 hours that are centered on the
sounding time.

12