Agris: Agricultural Remote Image Sensing

Abstract

Project AGRIS is an exploration in the field of Remote Image Sensing (RIS) used for crop yield prediction for better economic planning, basically it aims at analyzing different algorithms suitable and hence, using the most efficient one to provide an optimal prediction. For this purpose we researched the said domains with the decision of choosing Sentinel- 2 as the satellite for data due to it; being freely available with easy accessibility and its core competency of being especially designed to be used in agricultural applications. Hence, the following report provides a detailed and systematic explanation of how the project was carried out later followed by chapters focused on the understanding, learning and result discussion of the project models. The Literature Review for the first phase, has been constrained to papers regarding Crop prediction with Sentinel-2 which may involve LANDSAT and MODIS simulated datasets since Sentinel- 2 itself has been launched quite recently and so does not yet provide the temporal data for more than 3 years. Although, efforts have been made to create one dataset but that is only for Africa, which we do not want to use due to the landscape difference. Further discussion involves the basic problems like the inefficiency of current prediction methods used for prediction used by SUPARCO and NARC which use coarse resolution satellite images whereas our project using Sentinel – 2 images is low cost with higher and more accurate temporal, spectral and spatial resolution.