Mapping Water Use and Drought with Satellite Remote Sensing

Transcription

1 Xiwu Zhan NOAA-NESDIS-STAR Jason Otkin University of Wisconsin Mapping Water Use and Drought with Satellite Remote Sensing Christopher Hain NASA-MSFC Li Fang, Mitch Schull Earth System Science Interdisciplinary Center, University of Maryland, NOAA-NESDIS-STAR Martha C. Anderson, Bill Kustas, Feng Gao, Yun Yang, Yang Yang, Liang Sun USDA-Agricultural Research Service Hydrology and Remote Sensing Laboratory Beltsville, MD Thomas Holmes NASA-GSFC

2 Approaches to mapping ET PRECIPITATION SURFACE TEMPERATURE transpiration & evaporation Veg stress parms T soil & T veg transpiration & evaporation T veg soil evaporation Bare soil evap parms Sfc moisture infiltration Soil hydraulic parms runoff T Soil soil evaporation Rootzone moisture drainage Soil moisture holding capacity Root uptake Root distribution parms Given known radiative energy inputs, how much water loss is required to keep the soil and vegetation at the observed temperatures? WATER BALANCE APPROACH (prognostic modeling) ENERGY BALANCE APPROACH (diagnostic modeling)

3 ALEXI: regional ET modeling system ABL T a Blending height Blending height z ABL Closure T a Energy balance: ET = (R NET - G) - H Two-Source Model R a T ac R soil T s H = H c + H s H s H c R x E T T c T RAD (φ), f c 5 km ALEXI (Atmosphere-Land Exchange Inverse model) z blend 30 m a t 2 H(t)dt t 1 R a,i H 1 H 2 T A1 T RAD1 T A2 T RAD2 T DisALEXI RAD,i (φ i ), f c,i i q Surface temp: Air temp: Regional scale DT RAD T a - Geostationary - ABL model

4 Comparison of ET approaches COMPARISON of ET from energy and water balance models (ALEXI vs. Noah) (Green indicates energy balance ET is persistently wetter than expected based on local water balance) Differences are primarily related to: % Irrigation Hain, et al. (2014) Depth to water table (m) (as well as density of subpixel water bodies)

5 GEO (ISCCP) GEO (GOES Sounder) GEO (GOES Imager) Polar (MODIS) Polar (Landsat) Airborne (USU aircraft) SURFACE TEMPERATURE Temperature (C) EVAPOTRANSPIRATION Global (25km) Continental (10km) Hourly DATA FUSION: daily ET at field scale Regional (5km) Basin (1km) Daily Watershed (60m) 1 LS 16 day 2 LS 8 day corn soy 1 July :30AM LST Latent Heat (Wm -2 ) Field scale (30m)

6 LANDSAT (DisALEXI) MODIS (DisALEXI) GOES (ALEXI) GOES/MODIS/Landsat FUSION Daily Evapotranspiration Orlando, FL, 2002 DOY Landsat 5 Landsat 7 Spatial Temporal Adaptive Reflectance Fusion Model (STARFM) (Gao et al, 2006)

7 Evaluation of fused ET fluxes SMEX02 SMEX02 Soil Moisture Experiment 2002 Ames, Iowa Rainfed corn and soybean 4 km BEAREX08 Bushland ET and Remote sensing Experiment 2008 Bushland, Texas Rainfed and irrigated cotton MEAD Ameriflux site (S. Verma) Mead, NE Rainfed and irrigated corn and soybean

8 Model performance on Landsat dates SMEX02 BEAREX08 MEAD MAE: RE: 1.08 MJ m -2 d -1 8% 1.3 MJ m -2 d -1 10% 1.3 MJ m -2 d -1 11%

9 Validation using flux tower data Cumulative ET (mm) Daily ET (mm per day) Daily ET (mm per day) Rainfed soybean SMEX02 (Iowa) Reference ET Observed ET Landsat retrievals Landsat-only Landsat-MODIS fusion RAIN Day Day of Year of Year Day of Year

10 Irrigated vs. rainfed crop water use Bushland, TX Unirrigated cotton Mead, NE Unirrigated corn Irrigated cotton Irrigated corn

11 MONITORING DROUGHT Crop stress and yield impacts

12 Satellite ET Drought Indicator

13 ESI Methodology ALEXI ESI represents temporal anomalies in the ratio of actual ET to potential ET. ESI does not require precipitation data, the current surface moisture state is deduced directly from the remotely sensed LST, therefore it may be more robust in regions with minimal in-situ precipitation monitoring. Signatures of vegetation stress are manifested in the LST signal before any deterioration of vegetation cover occurs, for as example as indicated in NDVI, so TIR-based indices such as ESI can provide an effective early warning signal of impending agricultural drought. ALEXI ESI inherently includes non-precipitation related moisture signals (such as irrigation; vegetation rooted to groundwater; lateral flows) that need to be modeled a priori in prognostic LSM schemes.

14 Central Development US Flash of a Drought Multi-Scale of 2012 Remote-Sensing Based Framework for Mapping Drought over North America Christopher Hain (U. of Maryland) Flash drought are rapid onset events typically driven by precipitation deficits, high temperature anomalies and often strong winds. ESI has the potential to provide an early warning component during such events as water stress is able to be detected in the LST signal before degradation in the vegetation health occurs. 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

15 Yield anomaly ESI (JFM) Yield anomaly ESI (JFM) ANNUAL MUNICIPAL LEVEL SOYBEAN YIELD ANOMALIES ESI (s) Yield anomaly (kg/ha)

16 Conclusions LST-Based Evapotranspiration Diagnostically captures non-precipitation related moisture sources/sinks (irrigation, shallow groundwater, drainage) Capacity to map from global to sub-field scales using TIR-based data fusion Can be combined with remotely sensed soil moisture and precipitation data to interpret changes in other hydrologic variables ALEXI: Website: Christopher Hain Martha Anderson