Applications of Satellite Data to Assess Agricultural Production and Quantify Fallowed Agricultural Acreage during Drought Events in Washington and Nevada

Recent drought events in the western United States have highlighted the agricultural sector's vulnerability to changes in the frequency or duration of extreme events, and underscored the need to improve monitoring and mitigation of drought impacts to agriculture. Previous studies in California have shown that remote sensing can provide consistent and accurate estimates of changes in fallowing of agricultural lands during drought events. In response to a request from the California Department of Water Resources, we previously developed an approach for monitoring land fallowing in California that applies decision tree and neural network algorithms to timeseries of satellite data from the Landsat TM, ETM+ and OLI instruments, and the MODIS instrument onboard the Terra and Aqua satellites. Comparisons with data collected during field surveys of 680 fields in California demonstrated uncertainties of +/-10% or less in the monthly and seasonal satellite-derived estimates in both the winter/spring and summer growing seasons.

We present an extension of this approach to the states of Washington and Nevada, and describe modifications to the algorithms to account for different crop types and growing seasons in these states. Final accuracies for the modified approaches will be calculated at the end of the 2018 growing season using data from field surveys of 435 fields in Washington conducted in April and July, 2018, and surveys of 280 fields in the Nevada in July 2018.

The satellite data processing workflow was originally developed on the NASA Earth Exchange (NEX) and has been transferred to Google Earth Engine to facilitate sustained operational use by state agencies. We present an automated workflow that has been developed for state agency partners that uses the Google Earth Engine python-based application programming interface to perform the satellite data extraction and a python-based workflow for data processing and production of summary tables and maps. Co-development of tools for fallowed area mapping with state agency partners has been an important and valuable part of this project, and we describe lessons learned from the development of the algorithms and automated workflow