Vegetation Water Content Estimation Using Landsat Observations
T.J. Jackson, J. Tao, R. Bindlish, L. McKee and P. O’Neill
Vegetation water content (VWC) is a valuable input to most microwave based soil moisture algorithms. Previous research, both theoretical and experimental, has established that VWC can be estimated using multispectral remote sensing. There are limits on the reliability of these methods that are related to canopy characteristics. For some crops and other canopies, index based techniques typically saturate before peak VWC occurs. Therefore, a careful evaluation using validation data is useful, especially during an intensive field campaign. As part of National Airborme Field Experinment in 2006 (NAFE06), ground based VWC sampling was conducted in conjunction with ground based reflectance observations and Leaf Area Index (LAI) measurements. These data were used to establish relationships between the Normalized Difference Water Index (NDWI) and VWC for different land cover and crop types. In order to apply these relationships over the larger NAFE06 region it was also necessary to develop a land cover classifier using satellite data that incorporates the temporal changes that may have occurred during the study (i.e. flooding of rice fields). This was accomplished using multiple Landsat Thematic Mapper images and a decision tree classifier. The relationships between NDWI and VWC were applied to the available Landsat TM observations to estimate spatially distributed VWC datasets for October 6, November 7 and November 23, 2006. The values on these dates were used to interpolate a VWC for each pixel on each date of NAFE06. Our efforts to date have only involved the Yanco area because of the availability of ground data.