Journal of Earth System Science ( IF 1.9 ) Pub Date : 2021-04-09 , DOI: 10.1007/s12040-021-01566-6 Sheelu Jones , A K Kasthurba , Anjana Bhagyanathan , B V Binoy
Abstract
Landslides are global natural hazards with significant social and economic impact. The study is conducted in the wake of landslide occurrences in Kerala state, India, during 2018–2019. This study aims to demarcate landslide hotspots in the study area, identify anthropogenic activities, and critically examine their role in landslide occurrences. The identified landslide hotspots are utilized to comprehend spatial patterns of landslide distribution throughout the state. The landslide hotspots are concentrated in Idukki, Ernakulam, Kottayam, Wayanad, Kozhikode and Malappuram districts. Anthropogenic conditioning factors stimulating landslide occurrences are identified from land-use activities and the results obtained manifest that about 59.38% of total landslides have occurred in the plantation area. Human-induced parameters accelerating landslides are examined along with other natural parameters using various regression methods in the latter part of the study. The modelling techniques used in the study are the ordinary least square (OLS), spatial autoregressive model (SAR), and geographically weighted regression (GWR). All models point to the fact that anthropogenic activities such as plantation, quarry, road density, and cropland influence landslide occurrences. Results of both spatial models give excellent predictive capability compared to the OLS model. SAR analyzes the spatial interaction of parameters globally, whereas GWR considers the local aspect. The study encapsulates the importance of global and local spatial models in landslide studies based on their applicability.
Research Highlights
-
1.
The landslide hotspots of the state can be classified into two clusters, and they are concentrated in Idukki, Ernakulum, Kottayam, Wayanad, Kozhikode and Malappuram districts of the state.
-
2.
All hotspots recline in the Western Ghats region.
-
3.
About 59.38% of total landslides in Kerala have occurred in the plantation area.
-
4.
64.76% of the state's total landslides have happened in the human-modified land-uses.
-
5.
According to the spatial models, presence of human-modified landscapes has influence in landslide occurrences.
-
6.
However, it was found that the model performance of the global regression models (SAR) is higher than local regression model (GWR); predicted value of landslides are more accurate in GWR model.
中文翻译：
人为活动对印度西南部滑坡发生的影响：使用空间模型的调查
摘要
滑坡是具有重大社会和经济影响的全球自然灾害。这项研究是在2018-2019年印度喀拉拉邦发生滑坡之后进行的。这项研究旨在划定研究区域内的滑坡热点，确定人为活动，并严格审查其在滑坡发生中的作用。识别出的滑坡热点用于了解整个州的滑坡分布的空间格局。滑坡热点地区集中在Idukki，Ernakulam，Kottayam，Wayanad，Kozhikode和Malappuram地区。从土地利用活动中确定了促进滑坡发生的人为条件因素，获得的结果表明，人工林中约有59.38％发生了滑坡。在研究的后半部分，使用各种回归方法对人为加速滑坡的参数以及其他自然参数进行了检验。研究中使用的建模技术是普通最小二乘（OLS），空间自回归模型（SAR）和地理加权回归（GWR）。所有模型都指出这样一个事实，即人工活动（例如种植园，采石场，道路密度和农田）会影响滑坡的发生。与OLS模型相比，两个空间模型的结果均具有出色的预测能力。SAR会全局分析参数的空间交互作用，而GWR会考虑局部方面。该研究基于其适用性，概括了全球和局部空间模型在滑坡研究中的重要性。
研究重点
-
1。
该州的滑坡热点可分为两个类，它们集中在该州的Idukki，Ernakulum，Kottayam，Wayanad，Kozhikode和Malappuram地区。
-
2。
所有热点都在西高止山脉地区倾斜。
-
3。
喀拉拉邦大约59.38％的滑坡发生在人工林地区。
-
4，
全州发生的山体滑坡发生率达64.76％，发生在人为改变的土地利用方式中。
-
5，
根据空间模型，人为改变的景观的存在会影响滑坡的发生。
-
6，
但是，发现全局回归模型（SAR）的模型性能要高于局部回归模型（GWR）。在GWR模型中，滑坡的预测值更准确。
京公网安备 11010802027423号