Flood damage assessment for cultivated lands: BVDA based adaptive thresholding and local statistics

Abstract

Determining the land lost due to excess moisture for a harvest season is very important for the farmers. Quick and easy way of doing that is to utilize satellite imagery which provides accurate moisture damage calculations. In this study, an automated analysis technique is developed to process SPOT 5 Panchromatic data for delineating excess moisture on cultivated lands. The proposed method was validated with the maps that were manually digitized by imagery technicians. A total of six different SPOT 5 Panchromatic scenes acquired from May to July in 2011 over six different farming regions in Alberta and Saskatchewan, Canada were used in excess moisture damage assessment experiments. The preliminary results show that using BVDA as a dynamic threshold method resulted in more consistent and accurate results than CFAR. Incorporating local statistics with the global ones decreases the false alarm rate and increases the robustness of the results. Excess moisture layer over cultivated lands that is produced with the proposed methodology can be useful to farmers applying for crop insurance programs which might result in potential payments for unseeded acres.