Multi-sensor model-data fusion for estimation of hydrologic and energy flux parameters

Luigi J. Renzullo, Damian J. Barrett, Alan S. Marks, Michael J. Hill, Juan P. Guerschman, Qiaozhen Mu, Steve W. Running

Research output: Contribution to journalArticlepeer-review

50 Scopus citations

Abstract

Model-data fusion offers considerable promise in remote sensing for improved state and parameter estimation particularly when applied to multi-sensor image products. This paper demonstrates the application of a 'multiple constraints' model-data fusion (MCMDF) scheme to integrating AMSR-E soil moisture content (SMC) and MODIS land surface temperature (LST) data products with a coupled biophysical model of surface moisture and energy budgets for savannas of northern Australia. The focus in this paper is on the methods, difficulties and error sources encountered in developing an MCMDF scheme and enhancements for future schemes. An important aspect of the MCMDF approach emphasized here is the identification of inconsistencies between model and data, and among data sets. The MCMDF scheme was able to identify that an inconsistency existed between AMSR-E SMC and LST data when combined with the coupled SEB-MRT model. For the example presented, an optimal fit to both remote sensing data sets together resulted in an 84% increase in predicted SMC and 0.06% increase for LST relative to the fit to each data set separately. That is the model predicted on average cooler LST's (∼ 1.7 K) and wetter SMC values (∼ 0.04 g cm- 3) than the satellite image products. In this instance we found that the AMSR-E SMC data on their own were poor constraints on the model. Incorporating LST data via the MCMDF scheme ameliorated deficiencies in the SMC data and resulted in enhanced characterization of the land surface soil moisture and energy balance based on comparison with the MODIS evapotranspiration (ET) product of Mu et al. [Mu, Q., Heinsch, F.A, Zhao, M. and Running, S.W. (in press), Development of a global evapotranspiration algorithm based on MODIS and global meteorology data, Remote Sensing of Environment.]. Canopy conductance, gC, and latent heat flux, λE, from the MODIS ET product were in good agreement with RMSEs for gC = 0.5 mm s- 1 and for λE = 18 W m- 2, respectively. Differences were attributable to a greater canopy-to-air vapor pressure gradient in the MCMDF approach obtained from a more realistic partitioning of soil surface and canopy temperatures.

Original languageEnglish
Pages (from-to)1306-1319
Number of pages14
JournalRemote Sensing of Environment
Volume112
Issue number4
DOIs
StatePublished - Apr 15 2008

Keywords

  • Biophysical modeling
  • Canopy conductance
  • Energy flux
  • Land surface temperature
  • Multiple constraints model-data fusion
  • Soil moisture content

Fingerprint

Dive into the research topics of 'Multi-sensor model-data fusion for estimation of hydrologic and energy flux parameters'. Together they form a unique fingerprint.

Cite this