| The object of this research is to show that long range forecasts of precipitation for California is possible using large-scale climatological indexes and that artificial neural networks (ANNs) are a viable tool for modeling and data extraction. For each of California's seven climate zones, ANNs were trained using a calendar year's input of parameters to predict the coming water year's total precipitation and to predict the following water year's. Activity by the El Niño-Southern Oscillation (ENSO) in the east Pacific and the 700 mb height anomaly over the northern hemisphere is known to be related to various phenomena in specific regions of California. These large-scale climatological parameters represent the global atmospheric circulation that, in a sense, bring the weather to a region. By determining how these parameters interact over time, we can determine the general weather conditions that will arrive in a region. Because of the large amount of data, the short time period the data covers, the unknown type of relationships involved, and the possibly extraneous data, common statistical methods are not easily applied. Artificial neural networks (ANNs) are powerful and useful tools, especially in cases where the complex relationship between the inputs and outputs cannot easily be determined by common modeling methods.; 0It was found that the pattern of rainfall predicted by the ANN model matched closely the observed rainfall with the nine month time lag for most California climate zones and for most years. This portion of the research shows the possibility of making long range forecasts using ANNs and large scale climatological parameters.; These artificial “brains” were then analyzed by two different methods to reveal their methods of forecasting. One method produced for each climate zone a reduced set of important global parameters that were used in a simple linear regression model with good results. The second method gave information about how the individual years were affected by the indices. The ANN methods proved their ability to “see” the important relations in the data and have provided a new tool for data extraction. |
