Estimation of plant water content by spectral absorption features centered at 1,450 nm and 1,940 nm regions

Vegetation water content could possibly provide widespread utility in agriculture, forestry and hydrology. In this article, three species leaves were measured radiometrically in order to determine a relationship between leaf water status and the spectral feature centered at 1,450 and 1,940 nm where there are strong water absorptions. The first step of our research is to measure leaf spectra with a FieldSpec-FR. After the spectral analysis using the continuum removal technique, the spectral absorption feature parameters: absorption band depth (D (1450), D (1940)), the normalized band depth of absorption in 1,450 and 1,940 nm (BNA(1450), BNA(1940)), the ratio of the two reflectance of continuum line (R (1450i) /R (1940i) ), the ratio of the two band depth (D (1450)/D (1940)) and the ratio of the two absorption areas (A (1450)/A (1940)) in the two wavebands were extracted from each leaf spectrum. The fuel moisture content (FMC), specific leaf weight (SLW), equivalent water thickness (EWT) were measured for each leaf sample. A correlation analysis was conducted between the spectral absorption feature parameters and corresponding FMC, SLW and EWT. In addition, some existing indices for assessing water status such as WI (water index), WI/NDVI (water index/normalized difference vegetation index), MSI (moisture stress index), NDWI (normalized difference water index)were calculated and the correlation between them and water status were analyzed too. The results by comparing the correlations indicated that the spectral absorption feature indices we proposed were better. The indexes BNA(1940), D (1450)/D (1940), and A (1450)/A (1940) were well correlated with FMC, and the correlation between the indexes D (1450,) D (1940), R (1450i) /R (1940i) and EWT were strong. The index A (1450)/A (1940) was tested to be a good indictor for evaluating plant water content, because there was strongest positive correlation between it and FMC than other indices.