Characterizing the Uncertainty of Vegetation Moisture Content Retrieval through Radiative Transfer Model Inversion with Landsat 8 OLI Data

Abstract:

Systematic monitoring of vegetation moisture content (VMC) would improve early warning systems for conditions of water-stress (leading to decreased crop yield, increased fire risk, and amplified pest outbreaks and wildfires) as well as increase the accuracy of the retrieval of biomass consumed from satellite fire radiative power (FRP) (Smith et al. 2013). VMC is defined as the ratio between water content (weight of live biomass – weight of dry biomass) and either dry or living biomass weight. Despite the wide range of potential applications for systematic estimates of VMC, no VMC thematic product is currently generated (Yebra et al. 2013). The present paper presents the initial steps toward developing a modeling framework for the retrieval of VMC. A suite of field and laboratory measurements were collected weekly for various crop species throughout the 2014 and 2015 summer growing periods in the Southern Palouse region of Idaho. Measurements included: VMC, leaf area index (LAI), soil moisture content, leaf area, canopy height, and ground measured spectroradiometer data (350-2500 nm) of leaf and canopy reflectance. Field measurements were coincident with Landsat 8 OLI overpasses. PROSPECT- 5 and SAIL radiative transfer models were used to generate spectral signatures, simulating a variety of conditions and constrained by field and laboratory observations. Landsat 8 OLI data from coincident overpasses were used in an inversion approach to characterize VMC from field sites. The sources of uncertainty were characterized and examined for future model ensemble development.