Landslide Susceptibility Assessment in the Brazilian Atlantic Forest 2020

The research employs various datasets to develop a susceptibility map for shallow landslides in Guarujá, a significant city within the Brazilian Atlantic Forest biome. Morphometric features extracted through ArcGIS software, such as slope, hillside curvature, topographic moisture index, and aspect, were derived from a digital terrain model (DTM). Geological data sourced from the Geological Map of São Paulo and land use data from the Environmental Planning Office of São Paulo complement these morphometric features. Additionally, the scar inventory of landslides documented by the Municipal Coordinator of Protection and Civil Defense of Guarujá for 2020 was utilized. To ensure effective model training, both occurrence (landslide) and nonoccurrence (not prone to landslide) samples were acquired. Nonoccurrence data were generated employing three methodologies: random distribution, point generation within specified radii around occurrence points, and buffer creation around landslide scars. This comprehensive approach to dataset selection and generation aims to enhance the accuracy and generalization ability of the landslide susceptibility model.

本研究依托多源数据集，为巴西大西洋森林生物群落内的核心城市瓜鲁雅（Guarujá）构建浅层滑坡敏感性制图模型。研究基于数字地形模型（Digital Terrain Model, DTM），通过ArcGIS软件提取坡度、山坡曲率、地形湿度指数、坡向等地形形态特征参数。配套数据集还包括源自《圣保罗地质图》的地质数据，以及圣保罗环境规划局提供的土地利用数据。此外，本研究还采用了瓜鲁雅市保护与民防协调办公室2020年记录的滑坡疤痕调查清单。为保障模型训练的有效性，研究同时采集了滑坡发生样本（正样本）与非发生样本（负样本，即不易发生滑坡的区域样本）两类样本。非发生样本通过三种方法生成：随机分布法、在发生点周边指定半径范围内生成点集法、以及围绕滑坡疤痕创建缓冲区法。本数据集的遴选与生成采用了系统性方案，旨在提升滑坡敏感性模型的预测精度与泛化能力。