Most parts of world are currently facing an acute water shortage that is likely to become worse in the coming years. Climate change and global warming has a significant impact on the hydrological cycle. Both these factors climate change and global warming effect on the rainfall patterns and temperature. As the temperature increase from 2 Celsius to 4 Celsius it will rise the evaporation from the land and sea. The rainfall will be in higher intensity in higher latitudes and decrease in mid latitudes. The areas of the world which has scarce water will become drier and hotter. Global water withdrawal for agricultural sector is approximately 70%. However, most of this fresh water approximately 50 % is wasted due to inefficient and poorly managed irrigation system. The farming community in under develop countries of the world is wasting a huge amount of fresh water by using outdated and poorly managed flood irrigation (surface irrigation). Runoff estimation/prediction can be very valuable in water management and irrigation scheduling management. In this research an optimal reservoir precision irrigation system based on runoff estimation between two farms (farm1 and farm2) has been proposed to reduce water waste and to utilize the runoff water in nearby farm i.e farm2 or divert it back to reservoir through back runoff channels from both the farms in case of surplus amount of water left from either irrigation or there is an excessive rainfall. NRCS (Natural Resources Conservation Service), ANN (Artificial Neural Network), DT (Decision Tree),SVR (Support Vector Regression) and MLR (Multiple Liner Regression) are used to predict discharge, peak discharge and time to peak at farm1 and farm2 outlets. The performance of these algorithms is evaluated using different performance metrics. Overall, ANN show good performance for different datasets and scenarios while MLR show worse performance. Beside this an IOT (Internet of Things) based model is developed which remotely retrieved data from different environmental and agricultural based sensors such as temperature sensor, soil moisture sensor and crop stage sensor. The current conditions of farms is retrieved from sensors on mobile application, the end user has to only enter the precipitation depth/irrigation depth and the predication results are displayed in form of table showing NRCS predication, and other machine learning algorithms predication for total discharge, peak discharge and time peak, their comparison and also their respective hydrographs are displayed for different farm conditions.