Abstract: Non point source pollution needs special attention to control the surface water quality. Knowledge of critical buffer zone is of great importance for watershed management. For this study annual average of phosphate concentration, landsat image and ASTER DEM was used. Image was classified in four land cover classes using ERDAS Imagine software. buffer zones at radial distances of 100m, 200m, 500m, 1000m and 2000m from the discharge point of stream into the lake, were generated by Arc GIS software. Area of different classes was measured using FRAGSTAT software. Six spatial interpolation techniques were compared and best fitted technique (SK) was used to predict the phosphate concentration at all discharge points. Statistical analysis software SPSS was used to correlate area of land use and phosphate concentration. Area of agriculture and barren land in 500m buffer zone was found most effectively correlated with phosphate concentration.