Wade CrowUSDA ARS Hydrology and Remote Sensing Laboratory

John BoltenNASA Goddard Space Flight Center

Thanks: Xiwu Zhan (NOAA NESDIS), Curt Reynolds (USDA FAS), Christa Peters-Lidard (NASA GSFC), John Eylander (AFWA) and Sujay Kumar (NASA

GSFC/SAIC)

Funded by the NASA Applied Sciences Water Resources Application Area

Application of Terrestrial Microwave Remote Sensing to Agricultural Drought Monitoring

• Potential agricultural users are diverse (irrigation, crop insurance, yield forecasting, management practice assessment, etc).

• (Arguably) the most well-devopled agricultural applications is regional-scale crop condition and production estimates [appropriate spatial scales].

• Within the United States, this activity is carried out by the USDA Foreign Agricultural Service (FAS) International Production Assessment Division (IPAD).

Potential Agricultural Users of Satellite Soil Moisture

The USDA Foreign Agricultural Service (FAS) International Production Assessment Division (IPAD):

•Monthly global production estimates for commodity crops.

•Regional/national scales.

•Vital for economic competitiveness, national security and food security applications.

•Utilizes a wide-range of satellite data sources, input databases, climate data, crop models, and data extraction routines to arrive at yield and area estimates.

•Analyst-based decision support system.

•Characterizing the extent and impact of agricultural drought (i.e. root-zone soil moisture limitations) is critical for monitoring variations in agricultural productivity.

Crop Stress (Alarm) Models

Crop Models

2-Layer Soil Moisture Model Analysts

Global Rain and Met Forcing Data

Baseline USDA FAS Treatment of Soil Moisture

Goal: Use global soil moisture products (among many other things) to forecast variations in international agricultural productivity and yield.

Baseline Approach: Global application of a (simple) soil water balance model.

Crop Stress (Alarm) Models

Crop Models

“Modern” Land Surface Model Analysts

Global Rain and Met Forcing Data

Remotely-Sensed Soil Moisture

Modifications Examined by Project

Data Assimilation

What is the added value of assimilating remotely-sensed soil moisture information?

What is the added value of applying a more complex land surface model?

How do We Evaluate These Modifications?

1) Obtain a multi-year, monthly, 0.25° root-zone soil moisture (SM) product.2) Obtain a multi-year, monthly, 0.25° vegetation indices (NDVI) product.3) Sort both by month-of-year and rank across all years of the multi-year data set.(e.g., count all June’s in 2000-2010 that are drier than June 2005).4) Calculate the cross-correlation of SM/VI ranks.

For a 0.25° OK box:Soil Moisture is blackNDVI is red.

Degree of cross-correlation depends on:1) Climate (water versus energy limited growth conditions).2) Accuracy of the NDVI product.3) Accuracy of the SM product [Peled et al., 2010].

Global Rank Correlations for Model and Data Assimilation

Rank correlation between soil moisture for month i versus NDVI for month i+1

2002-2010 Performance in Data-Poor Regions

6 of the 10 most “food insecure” countries in

the world.

Seasonality Impacts

Rank correlation between soil moisture for month i versus NDVI for month i+1

Remotely-Sensed Soil Moisture Resources

Sensor: Band: Resolution: Dates:AMSR-E C/X (P) 30 km 2002-2011SMOS L (P) 45 km 2010 -ASCAT C (A) 50 km 2008 - SMAP L (A/P) 10 km 2015 -

Past/Current/Future

Repeat time (for all sensors) on the order of 2-3 days (at mid-latitudes)…latency is on the order of 12 -24 hours.

Remotely-Sensed Soil Moisture Resources

Sensor: Band: Resolution: Dates:AMSR-E C/X (P) 30 km 2002-2011SMOS L (P) 45 km 2010 - ASCAT C (A) 50 km 2008 - SMAP L (A/P) 10 km 2015 -

Past/Current/Future

Repeat time (for all sensors) on the order of 2-3 days (at mid-latitudes)…latency is on the order of 12 -24 hours.

• L-band passive

• Non-scanning - aperture synthesis using a 2-D radiometer array.

• Launched in October 2009.

• Planned operations until (at least) mid-2017.

ESA Soil Moisture Ocean Salinity (SMOS) Mission

Current SMOS Data Assimilation Product

Model-only SMOS+ Model (EnKF)

Current SMOS Data Assimilation Product

Operationally Implemented in Crop Explorer as of April 2014.www.pecad.fas.usda.gov/cropexplorer/

Current SMOS Data Assimilation Product

Remotely-Sensed Soil Moisture Resources

Sensor: Band: Resolution: Dates:AMSR-E C/X (P) 30 km 2002-2011SMOS L (P) 45 km 2010 - ASCAT C (A) 50 km 2008 - SMAP L (A/P) 10 km 2015 -

Past/Current/Future

Repeat time (for all sensors) on the order of 2-3 days (at mid-latitudes)…latency is on the order of 12 -24 hours.

• L-band unfocused SAR and radiometer system, offset-fed 6-m, light-weight deployable mesh reflector. Shared feed for:

1.26 GHz HH, VV, HV Radar at 1-3 km (30% nadir gap)

1.4 GHz H, V, 3rd and 4th Stokes Radiometer at 40 km

• Conical scan, fixed incidence angle across swath

• Contiguous 1000 km swath with 2-3 days revisit (less for combined ascending/descending)

• Sun-synchronous 6am/6pm orbit (680 km)

• Launch: January 29, 2015

• Mission duration 3 years

• Improved resolution relative to SMOS (10-km versus 45-km).

NASA SMAP Mission Concept

Microwave SM

Drought

Thermal SM

VIS/NIR SM

Time/Space Scale Considerations

Data Fusion Techniques

SAR SM

Thank you…

Bolten, J.D. and W.T. Crow, "Improved prediction of quasi-global vegetation conditions using remotely-sensed surface soil moisture," Geophysical Research Letters, 39, L19406, 10.1029/2012GL053470, 2012.

Crow, W.T., S.V. Kumar and J.D. Bolten, "On the utility of land surface models for agricultural drought monitoring," Hydrologic and Earth System Sciences, 16, 3451-3460, 10.5194/hess-16-3451-2012, 2012.

“SMAP Passive” = SMOS

“SMAP Active” = ASCAT

Clear benefits to integrating active and passive soil moisture

products!ACTIVE + PASSIVEPASSIVE-ONLY

Prototype SMAP-Based Data Assimilation System

• SMOS L2 (0-5 cm) surface soil moisture. • ~45-km spatial resolution.• L2 to L3 conversion (0.25°) by NESDIS SMOPS.• ~2 retrievals every 3-days.• ~12-hour latency.

• Modified 2-Layer Palmer Model at 0.25°.• AFWA LIS forcing/daily time step.• Assimilate SMOS L3 into model.• 30-member Ensemble Kalman filter (EnKF).• Use EnKF to update surface and root-zone.• 3-day composite delivered at ~4-day latency (from start of composite period).

Soil Moisture Remote Sensing:

Soil Moisture Data Assimilation:

Model

Observation

Observation

ModelModel

Current SMOS Data Assimilation Product

Global Rank Correlations for Various Models

Rank correlation between soil moisture for month i versus NDVI for month i+1

COMPLEX

COMPLEX

COMPLEX

SIMPLE