Efficient water management is a major concern in rice cropping. Controlling the use of excessive water in irrigation field is essential for the protection of underground water that will also be the part of climate change adaptation. The sustainable use of water resources is the prior task in Bangladesh. Imbalances between demand and supply are the main region for degradation of surface and groundwater. The human readability of checking the water level on irrigation field is considerable for these circumstances. In this paper I discussed the procedure for monitoring of surface water level in irrigation field, continuous monitoring of weather condition like temperature, air pressure, sunlight, rainfall etc. by using sensor network. The aim is to create a machine learning mechanism for farmers that can be given a forecast of water demand of irrigation field by the collection of IoT based data. In turn, this will help the farmer to prepare them to give water and on the other hand it will be helpful to use appropriate ground water and also it can be used for predict energy utilization. In this research Multiple linear regression algorithm is used for this prediction. Data from the irrigation field of North-West part in Bangladesh is used here to find the result of prediction.