Feasibility of estimating leaf water content using spectral indices from WorldView-3’s near-infrared and shortwave infrared bands

E. Raymond Hunt,Craig S. T. Daughtry,Li Li

DOI: https://doi.org/10.1080/01431161.2015.1128575

IF: 3.531

2015-12-23

International Journal of Remote Sensing

Abstract:A recently-launched high-resolution commercial satellite, DigitalGlobe’s WorldView-3, has 8 bands in the shortwave infrared (SWIR) wavelength region, which may be capable of estimating canopy water content at 3.7-m spatial resolution. WorldView-3 also has 8 multispectral bands at 1.24-m resolution with two bands in the near-infrared (NIR). The relative spectral response functions for WorldView-3 were provided by DigitalGlobe, Inc., and band reflectances were determined for reflectance spectra of PROSPECT model simulations and leaf data from maize, trees, grasses, and broadleaf herbaceous eudicots. For laboratory measurements, the range of leaf water contents was extended by including drying leaves and leaf stacks of corn, soybean, oaks, and maples. Correlations between leaf water content and spectral indices from model simulations suggested that indices using SWIR band 1 (center wavelength 1210 nm) had low variability with respect to leaf water content, but also low sensitivity. Other indices using SWIR band 5 (2165 nm) had the highest sensitivity, but also had high variability caused by different values of the leaf structure parameter in PROSPECT. Indices using SWIR bands 2, 3 and 4 (1570, 1660, and 1730 nm, respectively) had high correlations and intermediate variability from the leaf structure parameter. Spectral indices calculated from the leaf data had the same overall patterns as the simulations for variation and sensitivity; however, indices using SWIR band 1 had low correlations, and the best correlations were from indices that used SWIR bands 2, 3 and 4. Spectral indices for maize, grasses, and herbaceous crops and weeds had similar responses to leaf water content; tree leaves had higher index values and saturated at lower leaf water contents. The specified width of NIR band 2 (860–1040 nm) overlaps the water absorption feature at 970 nm wavelength; however, the normalized difference of NIR band 1 and 2 was insensitive to water content because NIR band 2’s spectral response was most heavily weighted to wavelengths less than 930 nm. The high spatial resolution of the WorldView-3 SWIR data will help analyze how variation among plant species and functional groups affects spectral responses to differences in canopy water content.

imaging science & photographic technology,remote sensing