Towards Improvement of Seasonal Rainfall Forecasting Through Model Output Statistics (Mos) Downscaling of Echam Forecasts Over Tanzania

Abstract

Many economic activities such as Agriculture and hydroelectric power generation are dependent on the availability of water. Most of the water comes from rainfall The seasonal rainfall over Tanzania is highly variable both in time and space. Hence there is a strong need forecasting model. The main objective of this study was towards improvement of seasonal rainfall forecasting through the use global model forecast from the ECl-IAM model and application of Model Output Statistics (MOS) downscaling on the model forecasts to improve the model scale of Tanzania. The data used in this study consists of monthly rainfall data from 15 stations and ECHAM outputs of circulation's and rainfall for the period of 1971 to 2004. Among this station only four had missing and they were estimated using arithmetic method. Single mass curve analysis was used for data quality control and it has been shown that the data from most stations are homogenous and consistent. The result from annual cycles of atmospheric circulation and rainfall occur Tanzania have shown that the model is capable of simulating the observed climatology over Tanzania. Experiments were carried out to determine whether the model cam stimulate the extreme dry and wet years over Tanzania. The skill of simulation was high during the October to December (OND) season with- correlation between the simulated rainfall and observed of 0.86. During the March to May (MAM) season the skill was generally low. This is could be due to the fact that the influence of sea surface temperature (SST) boundary forcing on then season rainfall is dominant. during the on season. The ECHAM model driven with observed SSTs. Past studies have indicated strong correlatio~~ betw'een global SSTs and seasonal rainfall over East Africa during October to December months compared to march to may period, in a addition to the influence of ENSO events which tend to be mature during October to December (Mutemi 2003) .. The MOS equations (model) were developed using stations data as predicted and ECHAM output as predictors various sub-sectors of the country. The rainfall simulation by the MOS was closer to the observed as compared to ECHAM model"in most regions