Using temporal changes in drought indices to provide drought early warning over sub- seasonal time scales Jason Otkin University of Wisconsin-Madison, Cooperative Institute for Meteorological Satellite Studies Martha Anderson USDA-Agricultural Research Service, Hydrology and Remote Sensing Laboratory Chris Hain University of Maryland, Earth System Science Interdisciplinary Center Mark Svoboda University of Nebraska-Lincoln, Drought Mitigation Center

Rapid Drought Development Rapid drought development is most likely to occur when extreme weather anomalies remain entrenched over the same area Intense heat waves, below normal rainfall, strong winds, and sunny skies can quickly deplete root zone soil moisture The term “flash drought” has been used in recent years to better distinguish these events from those that develop more slowly Exceptional cases of flash drought development occurred across the central U.S. in 2011, 2012, and 2013 According to the U.S. Drought Monitor (USDM), some areas experienced up to a three category increase in drought severity over periods as short as one month World Weather Open Science Conference, Montreal, QB, 17 August 2014

ALEXI Model and Evaporative Stress Index (ESI) Remotely sensed land surface temperatures (LST) retrieved from GOES thermal infrared (TIR) observations can provide a reliable signal of incipient drought stress As root zone soil moisture decreases, canopy temperatures rise in comparison to unstressed vegetation because less energy is being used for evaporation The Atmospheric Land Exchange Inverse (ALEXI) model uses this relationship to estimate evapotranspiration (ET) at high resolution Drought severity is inferred from reductions in the ratio of real to potential ET, as expressed by the Evaporative Stress Index Because ET can only be computed for clear pixels, multi-week ET composites are used to compute standardized ESI anomalies Negative ESI anomalies indicate drier than normal conditions, whereas positive anomalies indicate wetter conditions World Weather Open Science Conference, Montreal, QB, 17 August 2014

Highly variable rainfall during April and May (wet in the east, dry in the west) Drought conditions rapidly expanded northward during June and July USDM indicates that parts of OK and AR experienced up to a 3 category increase in drought severity during July 2011 Flash Drought Example – OK & AR USDM ESI_04WKRainfall July June May April World Weather Open Science Conference, Montreal, QB, 17 August 2014

Time series of surface weather conditions across eastern OK and western AR Hot temperatures, strong winds and diminished cloud cover anomalies developed by the end of May and then persisted all summer Strongly negative ESI values by the middle of June indicate that the vegetation was unable to adequately respond to the extreme conditions 2011 Flash Drought Example – OK & AR DRIVERS RESPONSE World Weather Open Science Conference, Montreal, QB, 17 August 2014

Change indicators show that conditions were already starting to rapidly deteriorate at the end of May Rapid deterioration began one month before similar large changes occurred in the USDM First three columns show the USDM, weekly rainfall, and 3-month SPI Next three columns show 2, 4, and 8 week ESI composites Last twelve columns show weekly ESI change variables 2011 Flash Drought Example – OK & AR World Weather Open Science Conference, Montreal, QB, 17 August 2014

Rapid Change Index The Rapid Change Index (RCI) was developed to encapsulate the accumulated rate of moisture stress change for the full duration of a rapid change event It is designed to identify areas experiencing either rapid increases or rapid decreases in moisture stress RCI resets to zero if the sign of the  ESI change anomaly is opposite that of the prior week Negative RCI values correspond to periods of rapidly increasing moisture stress while positive values indicate improving conditions World Weather Open Science Conference, Montreal, QB, 17 August 2014

Large negative RCI values in the top row indicate that moisture stress was rapidly increasing at the beginning of summer Impressive scope of the unusually rapid decrease in the ESI anomalies is clearly depicted by the large area of negative RCI values Initial appearance of negative RCI values led the introduction of severe drought in the USDM by more than 4 weeks 2012 Central U.S. Flash Drought Example Rainfall ESIRCIUSDM World Weather Open Science Conference, Montreal, QB, 17 August 2014

Drought intensification probabilities computed as a function of the RCI Ratio of RCI-derived to climatological probabilities shown for a 4-week lead time Compared to climatology, drought much more likely to develop when the RCI is negative Ratios increase greatly for decreasing RCI values and for increasing USDM intensity changes 4-Week Drought Intensification Probabilities >1 USDM >2 USDM>3 USDM World Weather Open Science Conference, Montreal, QB, 17 August 2014

Drought intensification probability ratios are lower than for 4-week lead times due to higher climatological probabilities Actual probabilities are in excess of 50% across large areas of the central and eastern U.S. Much higher ratios for larger RCI values shows that the accumulated time rate of moisture stress change can provide early warning of drought onset 8-Week Drought Intensification Probabilities World Weather Open Science Conference, Montreal, QB, 17 August 2014 >1 USDM >2 USDM>3 USDM

Problem: Because the maximum climatological value of the RCI varies across the U.S., it may be useful to convert the RCI values into drought intensification probabilistic forecasts RCI values tend to be larger across the south central U.S. due to the tendency for longer-lived and more extreme rapid change events that allow the RCI to become larger than other regions This means that the same RCI value may have different drought forecast implications for each region of the country Solution: A simple linear method was used to estimate the likelihood that the USDM will worsen by 1, 2, or 3 categories over 2, 4, and 8 week time periods based on the weekly RCI values Drought Intensification Forecasts World Weather Open Science Conference, Montreal, QB, 17 August 2014

2002 Central U.S. Flash Drought Example Large RCI and probabilities across South Dakota in June Drought spread northward into that region during June NW-SE oriented band of large RCI and probabilities developed across Nebraska by beginning of July Drought spread eastward into that region during July and August Large degradations also occurred in the NASS soil moisture and crop condition datasets Crop Condition Soil Moisture 12 August USDM > 1 CAT 4 weeks 29 July 15 July 01 July 17 June 03 June RCI > 2 CAT 8 weeks World Weather Open Science Conference, Montreal, QB, 17 August 2014

2002 Central U.S. Flash Drought Example Large RCI and probabilities across South Dakota in June Drought spread northward into that region during June NW-SE oriented band of large RCI and probabilities developed across Nebraska by beginning of July Drought spread eastward into that region during July and August Large degradations also occurred in the NASS soil moisture and crop condition datasets Crop Condition Soil Moisture 12 August USDM > 1 CAT 4 weeks 29 July 15 July 01 July 17 June 03 June RCI > 2 CAT 8 weeks World Weather Open Science Conference, Montreal, QB, 17 August 2014

Generate probabilistic forecasts using RCI variables derived from drought metrics sensitive to precipitation (e.g. the SPI) and to soil moisture anomalies (e.g. the NLDAS) Rapid changes in these variables also provide useful early warning signals of an elevated risk for drought development over sub-seasonal time scales Some combination of these RCI variables would likely provide the most accurate drought intensification forecasts Also actively working to improve our understanding of the causes and evolution of rapid onset, flash drought events Necessary to improve our ability to accurately forecast these damaging events Current and Future Work World Weather Open Science Conference, Montreal, QB, 17 August 2014

Flash Drought References Otkin, J. A., M. C. Anderson, C. Hain, and M. Svoboda, 2014: Using temporal changes in drought indices to generate probabilistic drought intensification forecasts. Conditionally accepted for publication in J. Hydrometeor. Otkin, J. A., M. C. Anderson, C. Hain, and M. Svoboda, 2014: Examining the relationship between drought development and rapid changes in the Evaporative Stress Index. J. Hydrometeor., 15, Otkin, J. A., M. C. Anderson, C. Hain, I. Mladenova, J. Basara, and M. Svoboda, 2013: Examining flash drought development using the thermal infrared based Evaporative Stress Index. J. Hydrometeor., 14, Anderson, M. C., C. Hain, J. A. Otkin, X. Zhan, K. Mo, M. Svoboda, W. Dulaney, and A. Pimstein, 2013: An intercomparison of drought indicators based on thermal remote sensing and NLDAS-2 simulations with U.S. Drought Monitor classifications. J. Hydrometeor., 14, Anderson, M.C., W.P. Kustas, C.R. Hain, C. Cammalleri, F. Gao, M.T. Yilmaz, I.E. Mladenova, J. A. Otkin, M. A. Schull, and R. Houborg, 2013: Mapping Surface Fluxes and Moisture Conditions from Field to Global Scales using ALEXI/DisALEXI. Remote Sensing of Land Surface Turbulent Fluxes and Soil Surface Moisture Content, Taylor and Francis, 562 pp.