A MODIS-based scalable remote sensing method to estimate sowing and harvest dates of soybean crops in Mato Grosso, Brazil
Minghui Zhang, Gabriel Medeiros Abrahão, Avery Cohn, Jake Campolo, Sally Thompson
Abstract
) sowing and harvest date dataset for Mato Grosso from 2004 to 2014 using a novel time series analysis method for Moderate Resolution Imaging Spectroradiometer (MODIS) satellite imagery, adapted for implementation in Google Earth Engine (GEE). Our estimates reveal that soy sowing and harvest dates varied widely (about 2 months) from field to field, confirming the need for spatially resolved crop timing information. An interannual trend toward earlier sowing dates occurred independently of variations in wet season onset, and may be attributed to an improvement in logistic or economic constraints that previously hampered early sowing. As anticipated, double cropped fields in which two crops are grown in succession are planted earlier than single cropped fields. This difference shrank, however, as sowing of single cropped fields occurred closer to the wet season onset in more recent years. The analysis offers insights about sowing behavior in response to historical climate variations which could be extended to understand sowing response under climate change in Mato Grosso.