Team Location: International Research Institute for Climate and Society, Palisades, New York
Sunny Ng (Columbia University)
Yifang Yang (Columbia University)
Pietro Ceccato, Ph.D. (International Research Institute for Climate and Society)
Walter Baethgen, Ph.D. åÊ(International Research Institute for Climate and Society)
Andrea Baraldi, Ph.D. (University of Maryland)
Luigi Boschetti, Ph.D. åÊ(University of Idaho)
The Uruguay Ministry of Agriculture has expressed concern about the need to detect and predict crop yield at early stages of the crop planting season in Uruguay for economic and food security reasons.åÊ During the Spring 2013 DEVELOP session, we determined that Moderate Resolution Imaging Spectroradiometer (MODIS) imagery at 250 meter resolution is too coarse to identify crop fields and Landsat imagery at 30 meter resolution would be more appropriate to identifying crop fields. In this study, we further investigated the possibility of using Landsat imagery to detect crops in Uruguay using prior knowledge of the evolution of spectral properties of crops over time and also using an object-oriented approach to identify crop fields.åÊ Using the satellite image automatic mapper (SIAM) to process Landsat 7 images, we identified crops through the evolution of SIAM spectral categories from barren land to vegetation.åÊ We also explored ways to identify crop fields based on the size, shape, and angle properties of crop fields.åÊ This study used Landsat 7 imagery from November to March for two planting seasons: 2010-2011 and 2011-2012. The accuracy of this methodology was tested using crop field measurements from the Instituto Nacional de InvestigaciÌ_n Agropecuaria (INIA) in Uruguay.åÊ This work can help INIA improve the detection of crops at an early stage in the growing season.