A COMPARISON OF SPATIAL RAINFALL ESTIMATION TECHNIQUES: A CASE STUDY OF NYANDO RIVER BASIN KENYA *Re‐published

Abstract

Many  hydrological  models  for  watershed  management  and  planning  require rainfall  as  an  input  in  a  continuous  format.  This  study  analyzed  four  different rainfall interpolation techniques in Nyando river basin, Kenya. Interpolation was done  for  a  period  of  30  days  using  19  rainfall  stations.  Two  geostatistical interpolation techniques (kriging and cokriging) were evaluated against inverse distance weighted (IDW) and global polynomial interpolation (GPI). Of the four spatial interpolators, kriging and cokriging produced results with the least root mean square error (RMSE). A digital elevation model (DEM) was introduced into the  cokriging  method  and  this  improved  the  results  considerably.  The  results demonstrate  that  for  low‐resolution  rain  gauge  networks,  geostatistical interpolation methods perform better than other techniques that ignore spatial dependence patterns. The use of secondary information improved the prediction results, as demonstrated by the inclusion of the DEM in this study.