Water balance study for Nyumba ya Mungu reservoir and sub-catchments upstream in Pangani river basin in Tanzania.

The analysis carried out to check the variability of river flow upstream Nyumba ya Mungu reservoir in Pangani basin revealed no decreasing or increasing trends in the river flow. For major rivers into Nyumba ya Mungu reservoirs, Ruvu at 1 DC2A and Kikuletwa at 1 DD 1 it was observed that there was insignificant fluctuations in the river flows which subsequently showed no variability of the inflows to the Nyumba ya Mungu reservoir. The estimation of inflow to Nyumba ya Mungu reservoir was done by Multiple Input Linear Perturbation Model (MILPM) on daily basis, the inputs being the river flows at 1DD1 and 1DC2A as first and second input respectively. The model efficiency was 81.03% in the calibration period and 93.66% in the verification period. The monthly and annual fitting of the model for the period between 1960-1964 gave model efficiencies of 89.97% and 92.51 % respectively. The annual daily average inflow to the reservoir was found to be 38.58 cumecs. The water balance for the reservoir carried out on daily, weekly and annual basis in volume units could not yield good results. Some negative values for daily and weekly water balance were observed. The average daily water loss in daily and weekly basis was found to 12.48 mmlday and 12.03 mm/day respectively. The average annual evaporation was found to be 4513.09 mm or 467.21 MCM using outflow from the turbine and 262.99 MCM or 2640.61 min using discharge data of station 1 D8C. An attempt to do water balance in level units (without using the storage elevation curve) was tested. The daily water loss on daily and annual water balance was determined as 11.72 mm and 4280 mm respectively. The reasons for these unrealistic results were error in recording reservoir levels, the area-elevation and the storage elevation curves for the reservoir. The reservoir characteristic curves were used in the conversion of levels to area of water spread and volume of reservoir. The water balance for the sub-catchments upsteam Nyumba ya Mungu was carried on monthly and annual basis. It was observed that there is a carry over effect of the rainfall and runoff process, which gave rise to negative balance during the period of low rainfall. The water loss and rainfall correlations for each sub-catchment showed a linear relationship as evidenced by good model efficiencies (over 90%). The obtained catchment water loss-rainfall relationships can be used to predict water losses and ultimately runoff for the desired sub-catchments.