The influence of landscape's dynamics on the Oriental Migratory Locust habitat change based on the time-series satellite data. (15th July 2018)

• Record Type: Journal Article
• Authors: Shi, Yue; Huang, Wenjiang; Dong, Yingying; Peng, Dailiang; Zheng, Qiong; Yang, Puyun
• Is Part Of: Journal of environmental management. Volume 218(2018)
• Page Start: 280
• Page End: 290
• Publication Date: 2018-07-15
• Subjects: Data fusion -- Random forest -- Time-series analysis -- Locust habitat mapping -- Change detection; Environmental policy -- Periodicals; Environmental management -- Periodicals; Environment -- Periodicals; Ecology -- Periodicals; 363.705
• DOI: 10.1016/j.jenvman.2018.04.028
• Languages: English
• ISSNs: 0301-4797

Abstract

Abstract: Landscape structure and vegetation coverage are important habitat conditions for Oriental Migratory Locust infestation in East Asia. Characterizing the landscape's dynamics of locust habitat is meaningful for reducing the occupation of locusts and limiting potential risks. To better understand causes and consequences of landscape pattern and locust habitat, it is not enough to simply detect locust habitat of each year. Rather, landcover transitions causing the change of locust habitat area must also be explored. This paper proposes an integrated implement to quantify the influence of landscape's dynamics on locust habitat changes based on three tenets: 1) temporal context can provide insight into the land cover transitions, 2) the detection of locust habitat area is operated on patches rather than pixels with full consideration of landscape's ecology, 3) the modeling must be flexible and unsupervised. These ideas have not been previously explored in demonstrating the possible role of changes in landscape characteristics to drive locust habitat transitions. The case study focuses on the Dagang district, a hot spot of locust infestation of China, from 2000 to 2015. Firstly, the seasonal characteristics of typical landcovers in NDVI, TVI, and LST were extracted from fused Landsat-MODIS surface reflectance imagery. Subsequently, a landscape membership-based random forest (LMRF) algorithm was proposed to quantify the landscape structure and hydrological regimen of locust habitat at the patch level. Finally, we investigated the correlations between the specific landcover transitions and habitat changes. Within the 16 years observations, our findings suggest that the sparse reeds and weeds in the vicinity of beach land, riverbanks, and wetlands are the dominant landscape structure associated with locust habitat change (R 2 > 0.68), and the fluctuation in the water level is a key ecological factor to facilitate the locust habitat change (R 2 > 0.61). These results are instrumental for developing precision pesticide use to reduce environmental degradation, and providing positive perspectives for ecological management and transformation of locust habitats. Highlights: Details of locust habitats are captured from fused Landsat-MODIS dataset. A patch-focused LMRF approach is proposed for habitat extraction. The landscape and hydrological structure of locust habitats are analyzed. The influence of landscape's dynamics on locust habitat are quantified.