Rainfall analysis using conventional and non-conventional rainfall information on monthly scale

摘要

The aim of this paper is to describe the technique used to create the merged analysis of rainfall over the Indian and adjoining region (1.5° to 35.5° N and 63.5° to 97.5° E). The technique is tested for monthly gridded fields of rainfall for a 2-year monsoon period (2001 and 2003) on a 1° × 1° latitude-longitude grid by merging rainfall estimates from different sources, viz satellite based estimates, rain gauge analysis and numerical weather prediction model rainfall. First, in order to reduce the random error involved in the satellite rainfall estimates and the model predictions, satellite and model estimates are combined linearly based on a maximum likelihood estimate method. In this case the weight for each component is inversely proportional to the squares of the individual random errors. The weight is determined by comparing the components with the concurrent gauge analysis. As the combined analysis contains bias from the individual input data sources, the combined analysis is then blended with the analysis based on gauge observations. It is seen that the merged analysis produced here is closer to the observations than the individual sources. It is observed that the magnitude and distribution of the orographic heavy rainfall along the Western Ghats of India is very different and more realistic compared to the Global Precipitation Climatology Project (GPCP) and the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP). When compared with the India Meteorological Department analysis, it is found that the merged analysis shows higher correlation than the satellite and model predicted rainfall. From the results it can be concluded that this study has shown promising results and the analyses can be used as a bench mark for evaluating model simulations which serves as a basis for real-time monitoring. Based on these promising results, long term datasets on high resolution grid for daily and monthly scale over Indian and adjoining region will be generated, which in turn can be used to study spatial and temporal variability of rainfall over Indian and adjoining region.