Three-Dimensional Distance Based Geostatistical Models to Adjust Radar Rainfall Data
MetadataShow full item record
Accurate and reliable rainfall input is crucial for hydrological modeling studies. Rain gauge collection and weather radar rainfall estimate are two of the most common techniques used for receiving rainfall data at a watershed. This study focuses on the development of three-dimensional (3D) distances based geostatistical models, such as Regression Kriging (RK) and Merging methods, to perform the adjustments of radar rainfall data to the targeted gauge measurements. These models are tested at the Chenyulan River watershed using the rainfall events of five typhoons landed Taiwan in recent years. Two-dimensional (2D) distance based models are also simulated to compare the adjusted rainfall values with those from 3D distance approaches. Results from Ordinary Kriging (OK) and gauge data are also included for comparisons. It is found in general the radar rainfall data can be corrected more accurately using the developed RK or Merging models than OK. Additionally, the adjusted rainfall values from 3D distance based models are similar to those using 2D distance based calculations at most tested stations. Depending on the typhoon events, using 3D distances in the semivariogram and Kriging interpolations is shown to be able to produce improved estimations of radar rainfall rates than 2D distance based calculations.