In this paper a complex model building procedure is developed. The idea is to split complex physical processes into simpler, individual physical processes. The basic procedure is to develop rules for each individual physical process in a numerically stable and efficient manner, and then assemble these processes in a cascading sequence to form a complete numerical solution of a complex model. The major advantage is that complicated mathematical models can be solved as a sequence of much simpler and less computationally intensive processes. The numerical accuracy of the solution method is excellent as long as the time step is relatively small.
The method is applied to the filters in the Earth radiation budget instrument (CERES), a device designed to study global climate change. The instrument measures earth temperatures through temperature changes on a detector caused by absorbed radiation from the earth scans. Extremely accurate temperature predictions are required due to the low signal to noise ratio. Parameter studies are conducted to examine the relative advantages of a two-filter system compared to the single filter arrangement. It is shown that a split filter is necessary to meet the stringent design requirements usually associated with Earth radiation budget applications.