Fusion of radar and rain gage measurements for an accurate estimation of rainfall

With the increased availability of rainfall measurements from multiple sensors having different spatiotemporal characteristics, issues of sensor fusion and intercomparison of different estimation methods are emerging as critical research questions. Cokriging is perhaps the most widely used method to fuse measurements from two sensors, for example, radar and rain gages. Cokriging offers a minimum variance estimate and can be shown to be the best linear estimator. It, however, requires the estimation of semivariograms which are usually not well behaved for rain gages. In addition, semivariograms and cross variograms estimated for radar and rain gages are subjected to constraints which are not easily met for most of the cases we examined. Here an alternative fusion methodology, based on recent developments in artificial neural networks (ANNs) is presented. ANNs are nonlinear estimators and thus have a distinct advantage over traditional statistical methods. Intercomparison of rainfall estimation, using cokriging and ANN methods, suggests that ANNs provide a more attractive and robust fusion of rainfall measurements from radar and rain gages for several storms from Oklahoma.