Estimating Precipitation Data Using a Fuzzybased Technique

As a flexible tool that may be adapted to many systems, fuzzy logic was employed to develop a new technique for estimation of precipitation data. This fuzzy-based technique estimates precipitation for any point upon available data from the neighboring meteorological stations. The contributions of each station is weighed due to its differential longitudes, latitudes, and altitudes with respect to the point of interest. This was accomplished through two membership functions for distance and elevation. Each of these functions are composed of four fuzzy sets (triangular and trapezoidal shapes with partial overlaps), which in turn led to sixteen fuzzy rules. Computing weights for each station activates a minimum of two and a maximum of four rules. The technique was tested for Khorasan province in eastern Iran using the data from 48 meteorological stations. Finally, the results from the fuzzy-based technique are compared with the results of two other commonly used methods, namely simple average and inversed-distance. The results generally showed the minimum error for fuzzy technique. The minimum meanabsolute- error and the model deviation of the estimated values were found for the fuzzy-based technique. Inverseddistance and simple average methods showed higher values for these parameters. The role of the number of stations involved in the estimation process was also discussed. The optimal number of stations are found to be four.