Optimizing spectral index to estimate the relative chlorophyll content of the forest under the damage of Erannis jacobsoni Djak in Mongolia

Abstract

Jas's Larch Inchworm (Erannis jacobsoni Djak) is a Lepidopteran insect pest that seriously threatens larch forest ecosystems in Mongolia. Damage caused by E. jacobsoni changes the chlorophyll content of forest trees, leading to significant changes in the color of the larch canopy. Chlorophyll content, an important parameter that reflects the physiological state of plants, is expressed as the relative chlorophyll content (RCC) evaluated using a chlorophyll meter. In this study, we estimate the relative chlorophyll content of forest trees damaged by E. jacobsoni by optimizing the spectral index. Four larch forest areas affected by outbreaks of this pest in Ikhtamir, Battsengel, and Tsenkher in the Mongolia Houhangai Province and Binder in the Khentii Province were selected as study areas. Based on the RCC and measured hyperspectral data of forest trees, hyperspectral features such as spectral index (SI) and continuous wavelet coefficients were analyzed. Partial least squares regression (PLSR), support vector machine regression (SVMR), and stepwise multiple linear regression (SMLR) were used to estimate the RCC of the total damage process and different degrees of damage. The optimized spectral index (OSI) showed the highest potential for estimating the total damage process, exhibiting good estimation accuracy and model stability. For example, in the SVMR model, the R2A of OSI-39 SVMR was 0.077, 0.074, 0.014, and 0.115 higher than those of the traditional spectral index (TSI), bior1.5, coif1, and sym3, respectively, while the RMSE was 0.017, 0.021, 0.014, and 0.048 lower than those of TSI, bior1.5, coif1, and sym3, respectively. In the estimation of different degrees of damage, the estimation performance of OSI was significantly improved compared with that of TSI and had the same potential as that of coif1. TSI, OSI, and coif1 showed the best estimation potential in a moderate degree of damage, of which OSI-SVMR had the best effect (R2A = 0.710 44 and RMSE = 0. 137).The optimization and combination method of SI used in this study will help to facilitate future research. Our findings provide insights into the estimation of RCC at the regional scale and for the effective monitoring of forest pest severity.