Infilling of Missing Rainfall Data from a Long Term Monitoring Records

Abstract

Good and reliable information about the spatial and temporal distribution of rainfall is necessary for the construction of thematic maps of rainfall distribution which can be used for classifying wet and dry years and also for solving of problems related to predicting changes in the water regime. In Rwanda there are more than 51 meteorological stations with rainfall data records ranging from 15 to 66 years, with missing data from 4 months to about 15 years. Zaza meteostation have about 66 years record of data with only 4 months missing. This scenario justified the need for infilling of missing data to enable better decision-making in water resources management. The main aim of this work was to generate the missing rainfall data for different meteostations and construct reliable spatial and temporal thematic maps of rainfall distribution in Rwanda. Linear regression statistical methods were used in resolving this problem. Graphs of the relationship between average monthly rainfall and time were constructed for all the meteorological stations for the period of 1926 to 2006. The infilling of the missing data of the sum of the monthly rainfall for any given station was done using the monitoring data of a neighbouring meteostation with relatively the same elevation above sea level and with a monitoring record of not less than 40 years and have the best correlation. The missing rainfall data were infilled for all the stations with resulting regression coefficients ranging from 0.55 to 0.80. This indicates the acceptability of the performed regression.