Print Email Facebook Twitter Error Analysis of TRMM, WFD and APHRODITE datasets using Triple Collocation Title Error Analysis of TRMM, WFD and APHRODITE datasets using Triple Collocation Author Rathore, P. Contributor Van de Giesen, N.C. (mentor) Faculty Civil Engineering and Geosciences Department Water Management Programme Water Resources Date 2014-10-29 Abstract The use of global precipitation datasets such as TRMM, WFD etc. for data scarce regions is gaining popularity since they provide forcing input for hydrological models. They make up for the lack of ground based data or the poor quality of whatever is available in many parts of the world. Using these datasets would be perfect if they were free of errors. Unfortunately, this is not the case. The geo-spatial data obtained from satellites or reanalysis products are not direct measures of precipitation. They are derived from atmospheric parameters such as cloud depth, brightness temperature etc. (Huffman 2007). The conversion of these to precipitation is done using complex algorithms. Efforts are made to calibrate this data but still errors sneak in. Similarly the interpolated gauge data like APHRODITE also has errors because of the inability of interpolation techniques to capture the high spatio-temporal variability in Precipitation. Hence the error estimation of these datasets remains a big problem. Lack of ground based data ensures there is no reference to check these global datasets against. In this research, Triple collocation technique is applied to 3 datasets namely APHRODITE, TRMM and WFD for the river basins in Myanmar. The technique gives an estimate of the residual errors in the 3 datasets (with uncorrelated errors) without using any ground measurements or true values (R. A. Roebeling 2012). This is the first time tit has been used to estimate errors in Precipitation datasets on a daily scale. Though the errors are not absolute, the results give an insight into the relative quality of these datasets. The errors have been calculated in space and time. Hence both temporal and spatial error characteristics are analysed. The study period is from 1998-2001. The results obtained show that for TRMM and WFD, the errors are concentrated and of higher magnitude. For APHRODITE, the errors are more evenly distributed in space. All three datasets showed high errors in the central dry parts and the delta region. Overall, APHRODITE seems to show lowest error values in space. The temporal error characteristics were also different for the 3 datasets. WFD showed highest average and maximum errors. TRMM had some very high error peaks but was in general better than WFD. Looking at the maximum and Average errors, APHRODITE seems to be the best of the three. WFD also shows some error peaks at the onset and end of Monsoon season. This shows its inability to estimate the localized pre and post monsoon storms. The assumption of uncorrelated errors was also verified post analysis. Errors for 2 locations, Bago and Yangon were used to make scatter plots. No strong correlation is visible in the scatter plots reinforcing the assumption that the errors are uncorrelated. The research shows that it is possible to make qualitative and quantitative inferences about the errors in the global precipitation datasets in the absence of in-situ measurements. Based on this research, it is concluded that overall, APHRODITE is the best of the 3 datasets. The possibility of a merged dataset formed by combining these 3 based on the error patterns observed in this study should be explored further. Subject TRMMWFDAPHRODITETriple CollocationError AnalysisPrecipitation DatasetsHydrological Modeling To reference this document use: http://resolver.tudelft.nl/uuid:b6c87b6e-f736-4130-a813-b4f9e170a6fc Part of collection Student theses Document type master thesis Rights (c) 2014 Rathore, P. Files PDF Thesis_Report_Final_P_Rat ... 240987.pdf 2.38 MB Close viewer /islandora/object/uuid:b6c87b6e-f736-4130-a813-b4f9e170a6fc/datastream/OBJ/view