Incorporation of Stem Water Content into Vegetation Optical Depth for Crops and Woodlands

Estimation of vegetation water content (VWC) by optical remote sensing improves soil moisture retrievals from passive microwave radiometry. For a variety of vegetation types, the largest unknown for predicting VWC is stem water content, which is assumed to be allometrically related to the water content of the plant canopy. For maize and soybean, measured stem water contents were highly correlated to canopy water contents, so VWC was calculated directly from the normalized difference infrared index (NDII), which contrasts scattering at near-infrared wavelengths with absorption of shortwave infrared wavelengths by liquid water. Woodland tree height is linearly related to woody stem volume, and hence to stem water content. We hypothesized that tree height is positively correlated with canopy water content, and thus with NDII. Airborne color-infrared imagery was acquired at two study areas in a mixed agricultural and woodland landscape, and photogrammetric structure-from-motion point clouds were derived to estimate tree heights. However, estimated tree heights were only weakly correlated with measured data acquired for validation. NDII was calculated from Landsat 8 Operational Line Imager (30-m pixel) and WorldView-3 (7.5 m pixel); but contrary to the hypothesis, NDII was not correlated with woodland tree height. Lastly, the interaction of woodland and crops stem water contents on total VWC in a mixed landscape were simulated for 2 days, one in the early summer and one in the late summer. VWC for the region varied from 2.5 to 3.0 kg m−2, which was just under a threshold for accuracy for soil moisture retrievals using Coriolis WindSat. Woodland tree height should be included as an ancillary data set along with land cover classification for soil moisture retrieval algorithms.