Evaluation and Improvement of the AQUA/AMSR-E Soil Moisture Algorithm AMSR-E Science Team Meeting June, 2011, Asheville, NC I. E. Mladenova, T. J. Jackson, R. Bindlish, M. Cosh USDA-ARS, Hydrology and Remote Sensing Lab, Beltsville, MD E. Njoku, S. Chan NASA, Jet Propulsion Lab, Pasadena, CA
Introduction Overall Almost a decade of soil moisture data products Used for a wide range of applications Extensively validated Some validation issues* The ground area contributing (satellite footprint) is ambiguous. Day to day shifting of the satellite track results in different azimuth angles The elliptical shape of the footprint means that a somewhat different area contributes for each overpass. Nonlinearities in the radiative transfer processes as a result of land cover, terrain, and soil types variability within the satellite footprint. Issues associated with ground data include: different sampling depths, network density, accuracy of the sampling techniques, etc. Several well established retrieval algorithms Strengths and weaknesses in the currently available retrieval techniques Bias, narrow dynamic range, … Introduction Objectives Team Algorithms Evaluation Summary * Jackson et al Jackson et al. (2010) Validation of Advanced Microwave Scanning Radiometer Soil Moisture Products, IEEE TGRS, 48(12),
Objectives/Goals Evaluate the performance of the AMSR-E standard/baseline algorithm using ground based measurements, and assess its performance against alternative algorithms and soil moisture products. Research will provide continuity for the existing Aqua/AMSR-E product, basis for transition of the algorithm to near-future missions, and will contribute to establishing a community algorithm applicable to multiple instruments and platforms. Refine and test validation procedures & metrics. Develop a better understanding of the merits of the existing algorithms. Algorithm(s) improvement. Introduction Objectives Team Algorithms Evaluation Summary
Team & Collaborators Team: E. G. Njoku 1* T. J. Jackson 2* S. Chan 1 R. Bindlish 2 M. Cosh 2 I. E. Mladenova 2 1 NASA, Jet Propulsion Laboratory, Pasadena, CA 2 USDA-ARS, Hydrology and Remote Sensing Lab, Beltsville, MD * PI Collaborators D. Bosch, G. C. Heathman, M. S. Moran, J. H. Prueger, M. Seyfried, P. J. Starks USDA-ARS Introduction Objectives Team Algorithms Evaluation Summary
Available Algorithms Passive microwave algorithms suitable for soil moisture inversion from X-band brightness temperature observations NASA, National Aeronautics Space Administration ( Njoku & Chan ) USDA-SCA, U.S. Department of Agriculture - Single Channel Algorithm ( Jackson ) JAXA, Japan Aerospace Exploration Agency ( Koike ) VU-LPRM, Land Parameter Retrieval Model ( Owe & de Jeu ) UMo, University of Montana ( Jones & Kimball ) IFA, Istituto di Fisica Applicata ( Paloscia ) NRL WINDSAT, Naval Research Laboratory ( Li ) PrU, Princeton University ( Gao & Wood ) Introduction Objectives Team Algorithms overview previous work Evaluation Summary
Available Algorithms: Summary All algorithms are based on the τ-ω model. Each accounts for the effects of surface temperature and vegetation; however, the way how this is done varies between the different algorithms. Retrieved parameters: Soil moisture Additional (depending on algorithm): vegetation optical depth, surface temperature, water fraction… Major differences: Screening for RFI, frozen soils, dense vegetation, open water bodies. Assumptions and parameterization. Ancillary datasets, etc. Introduction Objectives Team Algorithms overview previous work Evaluation Summary
Available Algorithms: Overview and examples Image courtesy of the JAXA and IFAC maps: JAXA Aqua AMSR-E Descending 2007/06/28 JAXA IFA USDA-SCA VU-LPRM NASA UMo Introduction Objectives Team Algorithms overview previous work Evaluation Summary
Image courtesy: Jackson et al Jackson et al. (2010) Validation of Advanced Microwave Scanning Radiometer Soil Moisture Products, IEEE TGRS, 48(12), Available Algorithms: Overview and examples Introduction Objectives Team Algorithms overview previous work Evaluation Summary
Image courtesy: Jackson et al Jackson et al. (2010) Validation of Advanced Microwave Scanning Radiometer Soil Moisture Products, IEEE TGRS, 48(12), Available Algorithms: Overview and examples Introduction Objectives Team Algorithms overview previous work Evaluation Summary
VU-LPRM NASA USDA-SCA JAXA x: AMSR-E retrieval –: station data Image courtesy: Draper et al Draper et al. (2009) An evaluation of AMSR-E derived soil moisture over Australia, RSE 113(4), AMSR-E time series were re- scaled using in situ data. Available Algorithms: Overview and examples Introduction Objectives Team Algorithms overview previous work Evaluation Summary
Evaluation… Assessment includes two aspects: evaluate the performance of the individual retrievals, and asses the accuracy of the resulting soil moisture products. Previous AMSR-E evaluation studies Selecting proper data sets statistics Introduction Objectives Team Algorithms Evaluation data sets stats Summary
In situ Validation Data Sets International Soil Moisture Network Criteria to consider when selecting a soil moisture network Image courtesy: Dorigo et al Dorigo et al. (2011) The International Soil Moisture Network: a data hosting facility for global in situ soil moisture measurements, HESS, 15, Introduction Objectives Team Algorithms Evaluation data sets stats Summary
In situ Validation Data Sets International Soil Moisture Network Criteria to consider when selecting a soil moisture network Density Frequency Scale Point Local Regional Global Monthly Hourly Low High Optimum Image modified after Jackson 2005, IGWCO Soil Moisture Working Group (ISMWG) Most… USDA watersheds… Introduction Objectives Team Algorithms Evaluation data sets stats Summary
Additional Validation Data Sets Continental/Global scale evaluation Additional (independent data sets) Other passive-derived soil moisture products e.g. SMOS Radar-based soil moisture products e.g. ERS/ASCAT Modeled output e.g. Noah, ECMWF, … Antecedent Precipitation Index, API Introduction Objectives Team Algorithms Evaluation data sets stats Summary
Evaluation statistics Error, RMSE/ubRMSE Sample time series correlation, r Error analysis using tree-way collocation statistics, triple collocation estimates RMSE (e 2 ) while simultaneously solving for systematic differences in each colligated data set, is based on linear regression models, and requires independent data sets Entekhabi et al Scipal et al … … … Introduction Objectives Team Algorithms Evaluation data sets stats Summary
In depth evaluation of the NASA AMSR-E soil moisture product as well as available alternative retrieval methods that focuses on physical and algorithm sources of differences. Algorithm improvement Link between the current AMSR-E and upcoming missions (GCOM-W,…) Introduction Objectives Team Algorithms Evaluation Summary