Skip to main content
ARS Home » Research » Publications at this Location » Publication #279294

Title: On the utility of land surface models for agricultural drought monitoring

item Crow, Wade
item Gupta, S
item Bolten, J

Submitted to: Hydrology and Earth System Sciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/1/2012
Publication Date: 9/24/2012
Publication URL:
Citation: Crow, W.T., Gupta, S.V., Bolten, J. 2012. On the utility of land surface models for agricultural drought monitoring. Hydrology and Earth System Sciences. 16(9):3451-2460.

Interpretive Summary: The ability to accurately monitor and predict the onset of agricultural drought at large scales it critical for drought mitigation and humanitarian applications. Recent research has focused on the use of complex land surface models for such monitoring. However, the added benefits associated with such models (above and beyond what is already obtainable using simpler approaches) has not yet been established. This paper applies a novel evaluation technique (based on remotely-sensed vegetation indices) to globally assess the performance of complex land surface models relative to much simpler approaches and concludes that the application of complex models is associated with only very minor benefits for agricultural drought monitoring. These types of assessment are critical for optimizing on-going USDA drought monitoring activities which support the department's goal of maintaining global food security.

Technical Abstract: The lagged rank cross-correlation between model-derived root-zone soil moisture estimates and remotely-sensed vegetation indices (VI) is examined from January 2000 until December 2010 to quantify the skill of various soil moisture models for agricultural drought monitoring. Examined modeling strategies range from a simple antecedent precipitation index to the application of modern land surface models (LSMs) based on complex water and energy balance formations. A quasi-global evaluation of lagged VI/soil moisture cross-correlation results suggests, when averaged in bulk across the annual cycle, little or no marginal added skill (<= 5% in relative terms) is associated with applying modern LSMs to off-line agricultural drought monitoring relative to simple accounting procedures based solely on antecedent precipitation. However, slightly larger amounts of added skill (5-15% in relative terms) is identified when focusing exclusively on the extra-tropical growing season and/or utilizing soil moisture values acquired by averaging across a multi-model ensemble.