Effect of Differently Interpolated Geographical Data on WRF-ARW Forecast

Abstract

With the advancement of computing power now a days Numerical Weather Forecasting (NWP) has emerged as an important tool and is playing its role for short to medium even for long range weather forecasting. Like other models theWeather Research and Forecasting (WRF) model has three components, i) Pre-processing, ii) Model running and iii) Post-processing. At the preprocessing stage one has to decide/declare the date and time, the region of interest, grid spacing (in case of nested run parent-child ratio), map projection etc. In this stage the geographical data is placed to defined grids and blended with the meteorological data. In numerical calculation/prediction the final output depends on the input. Thus it is assumed that the output has dependencies on the input values assigned at the grids. The value at the grids are depending on the interpolation technique. In this thesis we have compared the results of different predicted parameters (temperature, sea level pressure (SLP), relative humidity, cumulus accumulated precipitation and wind speed) of different events (rainfall, temperature and cyclone) calculated on the basis of different interpolation techniques. For the purpose default results are compared with the results obtained by using 4-point, 16-point and nearest neighborhood interpolation techniques. For the comparison purpose graphical/visual difference is considered. After completion of the model running GrADS(Grid Analysis and Display System) has been used for the graphing purpose. It has been found that except SLP the results of different interpolation schemes have similar pattern though the shape of the patterns are different in size. But in the case of SLP there are tremendous difference in result with respect to change of interpolation scheme.