Long-term effects of water stress on hyperspectral remote sensing indicators in young radiata pine

Abstract

The development of methods to spatially describe long-term water stress in plantations is going to become increasingly important as drought events increase in frequency under climate change. Despite this we are unaware of any research that has used hyperspectral imagery to describe long-term water stress for the most widely established exotic pine species, Pinus radiata D. Don (radiata pine). Hyperspectral imagery and foliage water content were repeatedly sampled over a five month period in one-year old radiata pine that were allocated to well-watered and drought treatments. These data were used to (i) determine how rapidly equivalent water thickness (EWT) and leaf fresh weight (LWCF) change in response to drought and (ii) identify the key hyperspectral indicators associated with changes in EWT and LWCF.Both EWT and LWCF exhibited little treatment variation until 91 days after treatment (DAT) but diverged strongly after this time. By 108 DAT, mean values in the control exceeded the droughted treatment by 41.3% for EWT (0.0175 vs. 0.0124 g cm−2, P < 0.001) and 17.2% for LWCF (67.7 vs. 57.8%, P < 0.001). Using hyperspectral data captured 108 DAT, highly significant treatment differences (P < 0.001) in reflectance were found in the SWIR region with the most significant differences occurring between 1139 and 2497 nm.There were strong correlations between a wide range of water indices and both measures of water stress using data captured 108 DAT. The indices that were most strongly correlated with EWT were found to be GVMI, NDWI1640 and SIWSI with R2 values ranging from 0.86 to 0.87. Indices that had the strongest correlations with LWCF were SRWI1, GVMI and MSI with R2 values ranging from 0.89 to 0.90. Water indices were more strongly related to both measures of water stress than the physiological index PRI which was moderately correlated to both EWT (R2 = 0.47) and LWCF (R2 = 0.62). A wide range of structural and chlorophyll indices were less well correlated with both measures of water content and long-term water stress than either PRI or the water indices.