Field-scale mapping of evaporative stress indicators of crop yield: an application over Mead, Nebraska

Authors: YANG, YANG - US Department Of Agriculture (USDA); Anderson, Martha; Gao, Feng; WARDLOW, B. - University Of Nebraska; HAIN, C. - Goddard Space Flight Center; OTKIN, J. - University Of Wisconsin; Alfieri, Joseph; Yang, Yun; SUN, L. - US Department Of Agriculture (USDA); Dulaney, Wayne

Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/4/2018
Publication Date: 6/1/2018
Citation: Yang, Y., Anderson, M.C., Gao, F.N., Wardlow, B., Hain, C., Otkin, J., Alfieri, J.G., Yang, Y., Sun, L., Dulaney, W.P. 2018. Field-scale mapping of evaporative stress indicators of crop yield: an application over Mead, Nebraska. Remote Sensing of Environment. 210:387-402.

Interpretive Summary: Soil moisture deficiency is a major factor in determining crop yields in water-limited agricultural production regions. Satellite remote sensing has the potential to provide spatially distributed and timely information about crop moisture stress, benefiting yield monitoring programs in the United States and abroad. In particular, the Evaporative Stress Index (ESI), developed by USDA-ARS scientists, provides spatial maps of stress-induced reductions in evapotranspiration (ET) - a measure of crop water use and availability. The ESI signals areas where crop transpiration is anomalously low, indicating severe depletion of the soil moisture reserves. ESI products are currently routinely produced over the United States at relatively coarse spatial scales (4 to 8-km pixels) using geostationary satellite data, and have demonstrated good capability in capturing flash drought impacts on crop condition. However, over intensively managed agricultural landscapes where the growing seasons can be quite compact, ESI performance can be degraded. Spurious anomalies in ET can arise due to shifts in the growing season from year-to-year. Furthermore, the signal at 4-8 km can be corrupted by mixed landcover at the sub-pixel level (including crops of different types, as well as lakes, wetlands, forest patches, etc.). Here, we test an ESI product generated at 30-m spatial resolution - a scale where pure crop pixels can be extracted, and simple corrections for variations in emergence date can be applied. The ET data used in this high-resolution ESI are developed by fusing information from multiple satellite platforms providing a complement of spatial and temporal detail. The results indicate that the high-resolution 30-m ESI provides excellent correlation with field- and county-scale yield anomalies - a significant improvement over the coarse, 4-km ESI. The study prototypes a methodology that could benefit operational yield monitoring within USDA.

Technical Abstract: The Evaporative Stress Index (ESI) quantifies temporal anomalies in a normalized evapotranspiration (ET) metric describing the ratio of actual-to-reference ET (fRET) as derived from satellite remote sensing. At coarse, regional scales (5-10 km resolution), the ESI has demonstrated capacity to capture developing crop stress and impacts on regional yield variability in water-limited crop growing regions. However, its performance in some regions where the vegetation cycle is intensively managed appears to be degraded. In this study, we generated maps of ET, fRET, and ESI at high spatiotemporal resolution (30-m pixels, daily timesteps) using a multi-sensor data fusion method, integrating information from satellite platforms with good temporal coverage and other platforms that provide field-scale spatial detail. The study was conducted over the period 2010-2014, covering a region around Mead, Nebraska that includes both rainfed and irrigated crops. Correlations between ESI and measurements of corn yield were investigated at both the field and county level to assess the value of ESI as a yield forecasting tool. To examine the role of crop phenology in ESI-yield correlations, annual input fRET time series were aligned by both calendar day and by biophysically relevant dates (e.g. days since planting or emergence). Results demonstrate that mapping of fRET and ESI at 30-m has the advantage of being able to resolve different crop types with varying phenology. The study also suggests that incorporating phenological information significantly improves yield-correlations by accounting for effects of phenology such as variable planting date and emergence date. The yield-ESI relationship in this study well captures the inter-annual variability of yields and thus has potential to be used for yield prediction, or for ingestion into a crop simulation model as a crop-specific moisture stress function.