A Bayesian Model for Estimating the Effects of Human Disturbance on Wildlife Habitats Based on Nighttime Light Data and INLA-SPDE

Nighttime light (NTL) data records the nocturnal emission signals of human activities and provides an accurate and continuous basis for the study of social and economic development. A new geostatistical model using NTL data is proposed to estimate the effects of human disturbance on wildlife habitats. This paper presents an extended application of NTL data using integrated nested Laplace approximation with stochastic partial differential equation (INLA-SPDE) for modeling the spatial correlation. Several covariates, such as incision depth, road network, population, and land cover, were used to delineate the distribution of human disturbance. Our method includes an improved SPDE construction approach that allows the detection of non-stationary data in NTL datasets. Among four types of SPDE modes, the urban mode achieved the best fitting performance, revealing the aggregation effect of cities in NTL data. Comparison with previous research shows that, the estimation results of human disturbance coincided with Wilderness Quality Index, and demonstrated its potential capacity for application in ecological protection and biological conservation through an analysis of natural reserves in Yunnan Province.