USING REMOTE SENSING & MODELING FOR EVALUATING HYDROLOGIC FLUXES, STATES, & CONSTITUENT TRANSPORT PROCESSES WITHIN AGRICULTURAL LANDSCAPES
Title: Remote Sensing of canopy water content during SMEX'04 and SMEX'05 using shortwave-infrared reflectances
Submitted to: IEEE Transactions on Geoscience and Remote Sensing
Publication Type: Abstract Only
Publication Acceptance Date: March 26, 2008
Publication Date: July 6, 2008
Citation: Hunt, E.R., Yilmaz, T.M., Jackson, T.J. 2008. Remote sensing of canopy water content during SMEX'04 and SMEX'05 using shortwave-infrared reflectances [abstract]. IEEE Geoscience and Remote Sensing Symposium. 2008 CDROM.
The Soil Moisture Experiments in 2004 (SMEX’04) and 2005 (SMEX’05) were conducted to validate algorithms on the retrieval of soil moisture content using microwave remote sensing. One of the biggest sources of error for estimation of soil moisture content is the amount of water in the vegetation. Typically, passive microwave radiometers can not detect changes in soil moisture when the vegetation water content is greater than 5 kg m-2. Canopy water content is the total foliage water mass per ground area and is only a fraction of the total vegetation water content. However, changes in canopy water content can be detected using reflected shortwave-infrared radiation from about 1200 nm to 2500 nm wavelength. The objectives of this study are to first determine if there is a constant relationship between canopy water content and reflectances at about 1650 nm wavelength and second to determine if total vegetation water content can be estimated from canopy water content.
SMEX’04 was conducted in Arizona, USA and Sonora, Mexico. Vegetation was sampled to obtain a large range of leaf area index (LAI) from different ecosystems from desert shrubland to irrigated agriculture. SMEX’05 was conducted in Iowa, USA. Soybean and corn were measured over time to obtain a large range of LAI; also a series of woodland sites were selected. Canopy water content was estimated from the product of LAI and leaf water content. Vegetation water content was only measured during SMEX’05. During both experiments, Landsat 5 Thematic Mapper (TM) data were acquired, geographically registered, and atmospherically corrected to land-surface reflectances. For SMEX’05, ResourceSat’s Advanced Wide Field Sensor (AWiFS) and Terra’s Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) were also acquired, registered, and atmospherically corrected. The Normalized Difference Infrared Index (NDII) was calculated:
NDII = (RNIR – RSW) / (RNIR + RSW)
where RNIR is the land-surface reflectance of the near-infrared channel (e.g. TM band 4) and RSW is the land-surface reflectance of the channel at 1600 nm (e.g. TM band 5).
During both SMEX’04 and SMEX’05, there were linear relationships between NDII and canopy water content. There was no significant difference between the two regression equations even though there was large changes in soil background reflectance and major differences in vegetation type. From SAIL canopy reflectance model simulations, the overall regression equation averages out the differences caused by soil background and leaf angle distribution. Therefore, the overall relationship may be valid for application to larger areas. From other data, NDII saturates at canopy water contents above 1.0 kg m-2. The standard error of the y-estimate is 0.05 kg m-2, which is about one-half the water content of a typical leaf. Thus, it is unlikely that the overall regression would be useful for detecting the onset of drought stress. The relationship between canopy water content and total vegetation water content varies with vegetation type. Corn and soybean both had linear relationships between these two variables; however, the regression slope was much higher for corn. Because these relationships were linear, NDII was linearly related to vegetation water content for corn and soybean. For woodland vegetation types, there was no relationship between canopy water content and total vegetation water content based on allometric equations with tree density and diameter. Therefore it is likely that canopy water content can be used to predict total vegetation water for only a few landcover types.