Enhancing USDA's Retrospective Analog Year Analyses Using NASA Satellite Precipitation and Soil Moisture Data

The USDA World Agricultural Outlook Board (WAOB) is responsible for monitoring weather and climate impacts on domestic and foreign crop development. One of WAOB's primary goals is to determine the net cumulative effect of weather and climate anomalies on final crop yields. To this end, a broad array of information is consulted, including maps, charts, and time series of recent weather, climate, and crop observations; numerical output from weather and crop models; and reports from the press, USDA attachés, and foreign governments. The resulting agricultural weather assessments are published in the Weekly Weather and Crop Bulletin, to keep farmers, policy makers, and commercial agricultural interests informed of weather and climate impacts on agriculture. Because both the amount and timing of precipitation significantly affect crop yields, WAOB has often, as part of its operational process, used historical time series of surface-based precipitation observations to visually identify growing seasons with similar (analog) weather patterns as, and help estimate crop yields for, the current growing season. As part of a larger effort to improve WAOB estimates by integrating NASA remote sensing observations and research results into WAOB's decision-making environment, a more rigorous, statistical method for identifying analog years was developed. This method, termed the analog index (AI), is based on the Nash-Sutcliffe model efficiency coefficient. The AI was computed for five study areas and six growing seasons of data analyzed (2003-2007 as potential analog years and 2008 as the target year). Previously reported results compared the performance of AI for time series derived from surface-based observations vs. satellite-retrieved precipitation data. Those results showed that, for all five areas, crop yield estimates derived from satellite-retrieved precipitation data are closer to measured yields than are estimates derived from surface-based precipitation observations. Subsequent work has compared the relative performance of AI for time series derived from satellite-retrieved surface soil moisture data and from root zone soil moisture derived from the assimilation of surface soil moisture data into a land surface model. These results, which also showed the potential benefits of satellite data for analog year analyses, will be presented.