Location: Hydrology and Remote Sensing LaboratoryTitle: Joint Leaf chlorophyll and leaf area index retrieval from Landsat data using a regularized model inversion system Author
Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/14/2014
Publication Date: 3/15/2015
Publication URL: http://handle.nal.usda.gov/10113/60936
Citation: Houborg, R., Mccabe, M., Cescatti, A., Gao, F.N., Schull, M.A., Gitelson, A. 2015. Joint Leaf chlorophyll and leaf area index retrieval from Landsat data using a regularized model inversion system. Remote Sensing of Environment. 159:203-221. Interpretive Summary: Leaf area index (LAI) and leaf chlorophyll (Chl) content provide critical information for modeling water use and carbon exchange. While algorithms for mapping LAI have become quite mature, the ability to estimate accurate leaf Chl content remains a difficult task. This paper presents a new technique to retrieve LAI and leaf Chl content simultaneously from Landsat data. The refined REGularized canopy reFLECtance (REGFLEC) retrieval system was applied to Landsat images over fields of maize and soybean in Nebraska during the 2001-2005 growing seasons. Results demonstrate that it is feasible to retrieve LAI and leaf Chl content simultaneously at the regional scale. The approach demonstrates a physically-based joint retrieval of LAI and leaf Chl content from Landsat for crop modeling at field scales which is important for yield monitoring by the National Agricultural Statistics Service and Foreign Agricultural Service.
Technical Abstract: Leaf area index (LAI) and leaf chlorophyll (Chl) content represent key biophysical and biochemical controls on water, energy and carbon exchange processes in the terrestrial biosphere. In combination, LAI and leaf Chl content provide critical information on vegetation density, vitality and photosynthetic potentials. Joint satellite-based retrievals of these variables could be used to inform land surface models and reduce uncertainties of predicted ecosystem fluxes in space and time. However, simultaneous retrieval of LAI and leaf Chl content from space observations is extremely challenging, as the combination of atmospheric effects, canopy characteristics and background reflectance may confound the detection of relatively subtle differences in canopy reflectances resulting from changes in leaf Chl content. Regularization strategies are therefore required to increase the robustness and accuracy of retrieved properties and enable more reliable separation of soil, leaf and canopy parameters. To address these issues, refinements to the REGularized canopy reFLECtance (REGFLEC) retrieval system were implemented by incorporating enhanced regularization techniques for exploiting ancillary LAI and temporal information derived from multiple satellite scenes over a given growing season. In this analysis, REGFLEC is applied to time-series of Landsat data, with retrieval results evaluated against in-situ LAI and leaf Chl observations collected over maize and soybean sites in central Nebraska over a 5-year period (2001-2005). Validation results demonstrate that it is feasible to retrieve leaf Chl content with a relative root-mean square-deviation (RMSD) on the order of 19% (RMSD=8.42 µg cm-2), although the level of accuracy depends on the configuration of REGFLEC. It was found that leaf Chl content retrievals were influenced by the version of the leaf optical properties model (PROSPECT) used, and the application of spatio-temporal regularization constraints is shown to be critical for estimating leaf Chl content with sufficient accuracy. REGFLEC reproduces the dynamics of in-situ measured LAI well (r=0.92), but the estimates are biased low, particularly over maize (LAI is underestimated by ~36 %), which may be largely attributed to differences between effective and true LAI caused by significant foliage clumping not properly accounted for by the canopy reflectance model (SAIL). Physically-based joint retrieval of LAI and leaf Chl content from Landsat observed reflectance data remains a challenging task due to the ill-posed nature of model inversion and the limited information carried by the radiometric signal (green, red and near-infrared in this study). A noteworthy and novel aspect of the REGFLEC approach is the fact that no site-specific data are required to calibrate the model, which may be run in a largely automated fashion using information extracted entirely from image-based and readily available datasets. Additional advances in the retrieval of canopy biophysical and leaf biochemical constituents will require innovative use of existing remote sensing data within physically realistic canopy reflectance models along with the ability to exploit the enhanced spectral and spatial capabilities of upcoming satellite systems.