Murugesan Bagyaraj, Balasumbramaniyan Gurugnanam, Swaminathan Bairavi, Karuppannan Shankar
Centre for Applied Geology, The Gandhigram Rural Institute (Deemed to Be University), Gandhigram, Dindigul, 624302, Tamil Nadu, India.
Department of Applied Geology, College of Applied Natural Science, Adama Science and Technology University, P.O. Box: 1888, Adama, Ethiopia.
Sci Rep. 2025 Jul 21;15(1):26463. doi: 10.1038/s41598-025-10948-5.
This research aims to define the potential of using natural networks for groundwater mapping. While neural networks have proven effective for various perceptual tasks, the difficulty in identifying data points below the surface remains a key challenge. The area under study encompasses a mountainous region in the Western Ghats. The most efficient, practical, along sensible methods for defining the GWPZ (Groundwater Potential Zones) in the Nilgiri's hard rock terrain are Geographic Information Systems (GIS) as well as analytic hierarchy process (AHP) of multicriteria decision making. To create various thematic layers, we utilized Indian topographical maps, satellite imagery, and field observations. We collected data on ten factors influencing groundwater(GW), including LULC(Land Use Land Cover), elevation, slope, soil type, geomorphology(GM), rainfall(RF), geology(GL), LD(lineament density), as well as DD(drainage density). Based on the weight assignment, all the thematic maps influencing GW events were assessed and compiled using GIS analysis. The weighted index overlay (WIO) approach and PCM (pairwise comparison matrix) within the AHP were used for a hierarchical ranking to identify the possible GW zones. The outcome revealed that the sample region could be divided into 5 separate groundwater potential (GWP) areas, i.e., very good (10%), good (32%), moderate (21%), low (26%), as well as very low (11%) potentials. Well and spring data were used to validate the model, and the ROC (Receiver Operating Characteristic) curve method was applied. The results showed a good accuracy of 70.03%. chance of correctly distinguishing a randomly chosen true positive from a false positive. This research is useful for improved preparation and control of GW supplies and offers swift guidelines for the discovery of GW in the hard rock aquifer region.
本研究旨在确定利用自然网络进行地下水测绘的潜力。虽然神经网络已被证明在各种感知任务中有效，但识别地表以下数据点的困难仍然是一个关键挑战。研究区域包括西高止山脉的一个山区。在尼尔吉里硬岩地形中定义地下水潜力区（GWPZ）最有效、实用且合理的方法是地理信息系统（GIS）以及多准则决策的层次分析法（AHP）。为了创建各种专题图层，我们利用了印度地形图、卫星图像和实地观测。我们收集了影响地下水（GW）的十个因素的数据，包括土地利用土地覆盖（LULC）、海拔、坡度、土壤类型、地貌（GM）、降雨量（RF）、地质（GL）、线性构造密度（LD）以及排水密度（DD）。基于权重分配，使用GIS分析对所有影响GW事件的专题地图进行评估和编制。层次分析法中的加权指数叠加（WIO）方法和成对比较矩阵（PCM）用于分层排序，以确定可能的GW区。结果表明，样本区域可分为5个独立的地下水潜力（GWP）区域，即潜力非常好（10%）、好（32%）、中等（21%）、低（26%）以及非常低（11%）。利用水井和泉水数据对模型进行验证，并应用ROC（接收者操作特征）曲线法。结果显示准确率良好为70.03%（即正确区分随机选择的真阳性与假阳性的概率）。本研究有助于改进地下水供应的准备和管理，并为在硬岩含水层区域发现地下水提供快速指导。
