While the vast computational requirements of numerical forecast models dictates the use of relatively coarse resolution, there is still a need to provide a forecast tailored to local areas. Currently, the forecasting of local weather elements such as precipitation, surface temperature, winds, and so forth are done by a statistical adaptation of a large-scale forecast model. The principal investigator and his colleagues will develop a model that computes the physical processes that locally modulate the large-scale meteorological fields. The parameters of such a model need to be adjusted to the local conditions. In doing this work, these scientists will employ the adjoint model technique, a tool of optimal control theory, to do this adjustment, that is, to find the set of parameters that minimize the local forecast errors. This research will allow the development of models that can be tailored to the specific needs of users of local forecasts such as utility industries, airports, agricultural entities, etc. There are many other potential applications of the technique such as the calibration of satellite observations of ground characteristics, and the indirect measurement of quantities that are difficult to measure routinely in situ such as soil moisture.
