Author(s): F. M. Woldemeskel; A. Sharma; B. Sivakumar
Linked Author(s):
Keywords: Spatial rainfall; Interpolation; Rain gauge; Satellite; Merging; Cross validation
Abstract: Accurate estimation of spatial rainfall is crucial for hydrological modelling and water resources assessment. Spatial rainfall can be estimated from rainfall observed using rain gauges or satellite-based techniques. However, both these rainfall estimates are uncertain due to either sampling or retrieval errors. The objective of this paper is to assess the benefit of merging rain gauge and satellite rainfall data so that the merged product possesses the strengths of both. To this end, spatial rainfall using monthly rain gauge data is first estimated at 0.05° x 0.05° longitude/latitude grid using Thin Plate Smoothing Splines (TPSS) and modified Inverse Distance Weighted (IDW) method for different rain gauge networks in Australia. The estimated rainfall is then merged with accumulated monthly Tropical Rainfall Measuring Mission (TRMM) 3B42 rainfall. Merging is carried out by linearly adding the two estimates after multiplying with weights. The weights are calculated based on error variances of each rainfall product. The modified Inverse Distance Weight (IDW) method helps to specify the weights more appropriately than the conventional method. Cross Validation (CV) errors reveal that merging helps to improve areal rainfall estimation, especially in areas where the rain gauge network is sparse.
Year: 2011