پهنه بندي اطلاعات بارندگي با استفاده از روش هاي آمار كلاسيك و زمين آمار و مقايسه با شبكه عصبي مصنوعي

Estimating Spatial Distribution of Rainfall Using Statistical and Geostatistical Methods and Comparison with Artificial Networks

In most hydrological and water resources studies, rainfall estimation is an important parameter. Different interpolation methods such as nearest neighbors, inverse distance, minimum curvature etc. can be used to show the distribution of rainfall over the watershed. Since the number of rain gauges are usually insufficient and there are high uncertainties in measurements, the estimated rainfall is not accurate. In the last few decades, geostatistics have been developed and applied in many engineering fields successfully. In this research, spatial distribution of rainfall over the Maroon Watershed has been estimated using statistical and geostatistical methods in GIS environment. On the other hand, artificial neural networks (ANN) have been applied as an interpolation tool and in combination with geostatistical methods. Results proved that geostatistical methods especially Kriging and Co-Kriging have better ability for estimation.