Evaluation of different interpolation methods on dailyrainfall estimation, Case study: Fars Province

Rainfall is one of the most important natural sources of water used by mankind foragricultural and industrial applications. Because of the difficulty of rainfall monitoring,which has high spatial and temporal variability, our knowledge of both the its synopticand climatological distribution over most of the country is not well known.Climatological investigations in Iran lack adequate meteorological observations tounderstand and interpret diverse climatic features. In the country as a whole, theground observation network is not dense enough to provide the detailed informationrequired, especially in rugged regions. Combined with this is the problem of pooraccessibility to the mountainous regions, which has resulted in the areal distribution ofrainfall being poorly known. In addition, the data collected by the existingmeteorological stations, with a few exceptions, are mostly discontinuous, nonhomogenousand short period observations. Such limited data sets, make it difficult toboth accurately delimit different climatic regimes across the country and identifysignificant departures from normal conditions, whereas many climatologicalapplications and investigations should ideally be based on the data collected at givenpoints over long periods and should have a good spatial and temporal coverage.Perhaps the most urgent problem facing rainfall measurement at the present time isdata collection - one of the most costly areas in meteorology and hydrology.In this paper, the central aim is to investigate the potential of GeographicalInformation System for monitoring Fars province’s rainfall with special reference toKriging and IDW (Inverse Distance Weighting), to describe pre-processingapproaches, including relative calibration, and to examine various techniques for bothrainfall monitoring.Analyses of the GIS techniques provided very useful information on both thespatial and temporal distributions of rainfall over regional space, although datacollected by Kriging technique showed more accurate than IDW technique. In order toexplore the appropriate method, the accuracy of the obtained results from differentmethods using RMS was examined. The results showed that among different usedkriging methods, the circular and exponential ordinary had lowest error.