Neuroid BP-type Model Applied to the Study of Monthly Rainfall Forecasting

A neuroid BP-type three-layer mapping model is used for monthly rainfall forecasting in terms of 1946-1985 Nanjing monthly precipitation records as basic sequences and the model has the form i × j ＝ 8 × 3, K = 1; by steadily modifying the weighing coefficient, long-range monthly forecasts for January to December, 1986 are constructed and 1986 month-to-month predictions are made based on, say, the January measurement for February rainfall and so on, with mean absolute error reaching 6,07 and 5,73 mm, respectively. Also, with a different monthly initial value for June through September, 1994, neuroid forecasting is done, indicating the same result of the drought in Nanjing dur-ing the summer, an outcome that is in sharp agreement with the observation.