WMOD00309 ‎(2)‎

• Title: Weather Modification Programme - Catalogue of Data Sets for the International Cloud Modelling Workshop/Conference
• Date: 6/1/1984
• Weather Modification - Doc Type: Report

2. MODEL EXPERIMENTS AND OBJECTIVES

The planning session recommended that three classes of experiments be perfor-
med: (a) Model verification against observations, (b) sensitivity tests, and
(c) application of models to increase physical understanding.

Table 1 shows a summary of the proposed experiments within each category.

2.1 Model Verification against Observations

Models are commonly constructed with the purpose of simulating certain charac-
teristics of the atmosphere. The confidence one places on the predictions (the
"output") of these models commonly is proportional to the ability of the model
to match nature. Generally, the greater the fidelity of the model with regard
to the greater number of cloud and environmental parameters, the greater is the
confidence in model outputs. Therefore, examination will be made of how well
the vari ous model s simul ate nature. The primary test quantity wi 11 be the pre-
diction of the amount and rate of precipitation. This, plus other cloud parame-
ters will be compared against observations of nature and against other model
outputs with the view to determining reasons causing variations in models. This
exercise will provide an opportunity to test the model's response to both
natural and seeded conditions and the breadth of the differences between natural
and seeded conditions should provide some indication of the consistency of the
models with regard to the prediction of changes in precipitation brought about
by seeding. The experiments should also contribute to development of the tools
for quantitative precipitation forecasts. Experiments dealing with convective
and orographic clouds were recommended.

2.1.1 Experimental approach

Both open (confirmation information supplied) and closed (confirmation
information withheld) data sets were recommended. Open sets would permit
examination of how well a model could be forced to simulate nature. Closed
data sets would permit examination of the ability of the model to predict
nature.

Open data sets, i.e, those which contain the observations to verify the fidelity
of the model, will be used to change adjustable parameters within the models to
optimize them so that the model replicates the observed conditions as closely as
possible. This exercise will permit comparisons of the model outputs under
ideal circumstances. Through these comparisons and knowledge of the nature of
the optimization needed, information on the critical and sensitive aspects of
the models should be obtained.

Closed data sets, ; .e., those which contain the initial conditions necessary to
run the model but not the observations to velri fy thei r output, shoul d provi de
information on how well the models predict c"loud characteristics. It was recom-
mended that each closed data set be from the same area and season as its com-
panion open data set. This would justify thl~ use of the optimized adjustments
from the open set with the closed set.

2.