WMOD00579

• Title: An Exploratory Forecast Experiment to Predict Supercooled Liquid Water in the Sierra Nevada
• Date: 11/1/1988
• Weather Modification - Doc Type: Report

and a value of zero when the rows all having the same
proportions when compared to their marginal totals (i.e., no
association). It should be noted that the TSS measure of
association in an ordered 2 x 2 table is identical to Somers'
statistic (Flueck, 1988).

Lastly, the experiment was conducted from 30/12/85 to
1~/03/86 (225 possible forecasts) in the first season and from
01/11/86 to 07/01/87 (211 possible forecasts) in the second
season. However, only about 67% of these forecast opportunities
were utilized typically due to one forecaster also having project
management duties and to the predesignated "down" days of the
experiment.

4.0 ANALYSIS Ai~D DISCUSSION

A. Liquid Water

Table I, Panel A presents the 3 x 3 contingency table
of counts and percentages (in parentheses) for the
predicted versus the observed LW concentration for the first
forecast period of 0-2 hours. We note that 76% of the
forecasts exactly matched the observed results (highlighted
by the -dotted line) suggesting some forecasting ability.
However, this percentage is dominated by the LW = 0 joint
cell (i.e., 54.5%). Figure 3 presents this data in a 3-
dimensional frequency diagram.

Looking at the marginal percentages of the table
indicates there was a tendency to under-forecast the 0
(i.e., 57.9 versus 61.0%) and the 0-.10 rom (i.e., 17.1
versus 25%) categories. Correspondingly, there was a
tendency to over-forecast the LW > .10 mm 25 versus 14%).

8