[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-2946":3,"related-2946":65},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":6,"published_at":11,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":24,"tags":26,"search_phrases":32,"slug":35,"view_count":36,"doi":37,"paper":38,"created_at":64},2946,"Forecasting cocoa export volumes in Indonesia using remote sensing, climate variability, and economic indicators: A comparative modeling approach","https:\u002F\u002Fdoi.org\u002F10.1177\u002F18747655261486415","Accurate forecasting of agricultural exports is essential for supporting production planning, trade management, and evidence-based policymaking. However, forecasting cocoa exports remains challenging because export volumes are influenced by interconnected environmental, climatic, and economic factors. This study proposes an integrated forecasting framework for Indonesian cocoa export volumes by incorporating remote sensing indicators, climatic indices, and economic variables into statistical, machine learning, deep learning, and hybrid forecasting approaches. The exogenous variables consist of Enhanced Vegetation Index (EVI), Land Surface Temperature (LST), rainfall, cocoa price, Oceanic Niño Index (ONI), Dipole Mode Index (DMI), and exchange rate. The study evaluates benchmark models (Naïve, Seasonal Naïve, and ETS), statistical time-series models (ARIMAX and SARIMAX), machine learning and deep learning models (XGBoost and LSTM), and a hybrid SARIMAX–XGBoost model. Model performance is evaluated using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) over an independent 17-month test horizon from November 2023 to March 2025. XGBoost achieved the overall best forecasting performance, obtaining the lowest test RMSE of 2,609,038 kg, MAE of 2,096,098 kg, and MAPE of 7.56%. Its advantage over ARIMAX, SARIMAX, and the SARIMAX–XGBoost hybrid was supported by pairwise comparisons using the modified Diebold–Mariano (DM) and Kolmogorov–Smirnov Predictive Accuracy (KSPA) tests. The comparison with LSTM showed no statistically significant difference according to the DM test, although the KSPA test indicated a significant difference. These findings demonstrate that XGBoost provided the strongest out-of-sample forecasting performance among the evaluated models.","农业出口的准确预测对于支持生产规划、贸易管理和循证政策制定至关重要。然而，可可出口预测仍然具有挑战性，因为出口量受到环境、气候和经济因素相互交织的影响。本研究提出了一种印度尼西亚可可出口量的集成预测框架，将遥感指标、气候指数和经济变量纳入统计、机器学习、深度学习和混合预测方法中。外生变量包括增强植被指数（EVI）、地表温度（LST）、降雨量、可可价格、海洋尼诺指数（ONI）、偶极子模态指数（DMI）和汇率。本研究评估了基准模型（Naïve、Seasonal Naïve和ETS）、统计时间序列模型（ARIMAX和SARIMAX）、机器学习和深度学习模型（XGBoost和LSTM）以及SARIMAX–XGBoost混合模型。模型性能通过均方根误差（RMSE）、平均绝对误差（MAE）和平均绝对百分比误差（MAPE）在2023年11月至2025年3月的独立17个月测试期内进行评估。XGBoost取得了总体最佳的预测性能，测试RMSE最低为2,609,038 kg，MAE为2,096,098 kg，MAPE为7.56%。其相对于ARIMAX、SARIMAX和SARIMAX–XGBoost混合模型的优势得到了使用修正Diebold–Mariano（DM）检验和Kolmogorov–Smirnov预测精度（KSPA）检验的成对比较的支持。与LSTM的比较显示，根据DM检验无统计学显著差异，尽管KSPA检验表明存在显著差异。这些发现表明，在所评估的模型中，XGBoost提供了最强的样本外预测性能。",null,"Statistical Journal of the IAOS","2026-09-18T00:00:00Z","论文",10,false,73,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},12,22,17,13,9,1,"将遥感、气候指数与经济变量整合进可可出口预测框架，XGBoost 表现最优，方法新颖且结论可靠，对农产品贸易信息化有参考价值。",[25],{"name":10,"url":6},[27,28,29,30,31],"机器学习","遥感监测","可可出口","农业预测","气候指数",[33,34],"印尼 可可 出口 预测","遥感 气候 可可 出口","印尼可可出口预测-2946",0,"10.1177\u002F18747655261486415",{"doi":37,"openalex_id":39,"authors":40,"venue":10,"cited_by_count":36,"oa_url":9,"card":56,"direction":62,"ingested_from":63},"W7213558756",[41,44,46,48,51,54],{"name":42,"orcid":43},"Erna Nurmawati","https:\u002F\u002Forcid.org\u002F0009-0002-3385-673X",{"name":45,"orcid":9},"Neli Agustina",{"name":47,"orcid":9},"Robert Kurniawan",{"name":49,"orcid":50},"Prana Ugiana Gio","https:\u002F\u002Forcid.org\u002F0000-0002-8155-005X",{"name":52,"orcid":53},"Rayhan Abyasa","https:\u002F\u002Forcid.org\u002F0009-0006-0476-7999",{"name":55,"orcid":9},"Aditya Hari Kurnia Putra",{"tldr":57,"method":58,"finding":59,"direction":60,"opportunity":61},"融合遥感、气候与经济指标，比较多种模型预测印尼可可出口量，XGBoost表现最佳。","用EVI、LST、降水、ONI、DMI、价格、汇率等变量，比较ARIMAX、SA","XGBoost预测精度最高，MAPE为7.56%，显著优于统计模型和混合模型。","农业人工智能与决策模型","可探索多源遥感与气候指数融合的混合模型，提升农产品出口预测精度与可解释性。","农业遥感与作物表型","openalex","2026-09-19T23:30:33.176995Z",{"total":66,"page":22,"page_size":66,"items":67},6,[68,116,163,211,267,318],{"id":69,"title":70,"url":71,"summary":72,"summary_zh":73,"content":9,"source_name":74,"source_url":71,"published_at":75,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":76,"score_detail":77,"sources":82,"tags":84,"search_phrases":88,"slug":91,"view_count":36,"doi":92,"paper":93,"created_at":115},2871,"Spatial and Temporal characteristics and driving force analysis of vegetation cover change in Shanxi Province, China based on kNDVI and XGBoost-SHAP model","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffenvs.2026.1948848","Introduction Systematically clarifying the spatiotemporal evolution of vegetation coverage and elucidating its response mechanisms to climatic fluctuations and human activities carries substantial practical implications for advancing the Dual Carbon Strategy and optimizing the pattern of territorial spatial development and conservation across Shanxi Province, China. Methods Based on kernel Normalized Difference Vegetation Index (kNDVI) datasets, multi-source meteorological, topographic and socioeconomic datasets, this study integrated Theil-Sen slope estimation, Mann-Kendall significance test, Hurst exponent, standard deviational ellipse and gravity center migration model to systematically characterize the spatiotemporal patterns of vegetation coverage for the period 2000 to 2024 and predict the potential persistence of its future evolution. Meanwhile, an interpretable XGBoost-SHAP machine learning framework was constructed to quantify the independent contributions, nonlinear marginal responses, and temporal evolutionary characteristics of seven driving factors: elevation, slope, aspect, annual mean temperature, annual mean precipitation, population density and nighttime light intensity. Results (1) Temporally, the multi-year average kNDVI of the whole province reached approximately 0.1699, presenting an extremely significant fluctuating upward trend with an annual growth rate of 0.0041. Annual kNDVI ranges from 0.1132 recorded in 2001 to a peak value of 0.2227 in 2024. (2) Spatially, vegetation coverage exhibited a remarkable differentiation pattern of “low values in the north and high values in the south”. Higher kNDVI values were predominantly distributed within the forest-covered mountainous regions of the Taihang, Lüliang, Zhongtiao, and Taiyue Mountains, whereas relatively low values were distributed across the sandy-hilly regions of northern Shanxi and urban agglomerations along the Fen River Valley. Vegetation restoration was observed across 98.02% of the study area, among which roughly 92.18% exhibited an extremely significant improvement. (3) Over 2000–2024, the major axis of vegetation spatial distribution maintained a stable northeast-southwest orientation, accompanied by moderate outward expansion of the standard deviational ellipse, and a general northwestward shift of the vegetation gravity centre. (4) The Hurst exponent indicated that the vegetation improvement trend in 98.03% of the region would persist, while scattered patches of industrial, mining and urban land (accounting for 1.97%) faced the potential vegetation degradation risk. (5) The regional vegetation driving system experienced three evolutionary stages: a single precipitation-dominated natural driving stage (2000–2005), a climate-human coupled transitional stage (2005–2010), and a multi-factor synergistic balanced stage (2010–2020). Precipitation and temperature acted as core climatic drivers, and elevation largely determined the vertical differentiation baseline of vegetation distribution. Population density and nighttime light intensity exerted persistent suppressive effects on vegetation growth. Empirically derived tentative thresholds are identified for major predictors: elevation ∼1,200 m, slope ∼8°, annual mean temperature ∼7.5 °C, annual mean precipitation ∼500 mm, population density 500 persons\u002Fkm 2 , and nighttime light intensity ∼5. These values can serve as reference boundaries for differentiating ecological conservation zones from human-disturbed zones. Conclusion This research identified the intertemporal differentiation and nonlinear coupling laws governing vegetation dynamics in Shanxi Province, a temperate transition zone on the Loess Plateau. Differentiated ecological governance strategies targeting climate adaptation, topographic zoning, and anthropogenic-pressure regulation are proposed, which provide observational scientific support for the construction of ecological security barriers on China’s Loess Plateau.","引言 系统厘清植被覆盖的时空演变规律并阐明其对气候波动与人类活动的响应机制，对推进双碳战略、优化山西省国土空间开发保护格局具有重要现实意义。方法 基于核归一化植被指数（kNDVI）数据集及多源气象、地形和社会经济数据，本研究综合运用Theil-Sen斜率估计、Mann-Kendall显著性检验、Hurst指数、标准差椭圆和重心迁移模型，系统刻画了2000—2024年植被覆盖的时空格局，并预测其未来演变的潜在持续性。同时，构建了可解释的XGBoost-SHAP机器学习框架，量化了高程、坡度、坡向、年平均气温、年平均降水量、人口密度和夜间灯光强度7个驱动因子的独立贡献、非线性边际响应及时间演变特征。结果 （1）时间上，全省多年平均kNDVI约为0.1699，呈极显著波动上升趋势，年增长率为0.0041。年kNDVI从2001年的0.1132变化至2024年的峰值0.2227。（2）空间上，植被覆盖呈现“北低南高”的显著分异格局。较高kNDVI值主要分布于太行山、吕梁山、中条山和太岳山等森林覆盖山区，而较低值分布于晋北沙丘丘陵区和汾河谷地城市群。研究区98.02%的区域植被呈恢复态势，其中约92.18%表现为极显著改善。（3）2000—2024年间，植被空间分布的主轴保持稳定的东北—西南走向，标准差椭圆呈中度向外扩张，植被重心总体向西北方向迁移。（4）Hurst指数表明，98.03%区域的植被改善趋势将持续，而零星斑块区域的工业","Frontiers in Environmental Science","2026-09-17T00:00:00Z",82,{"impact":78,"substance":79,"depth":80,"authority":20,"freshness":21,"relevant":22,"comment":81},18,23,19,"基于kNDVI与XGBoost-SHAP的山西植被时空演变与驱动力研究，方法新颖、数据扎实，对黄土高原生态治理有参考价值。",[83],{"name":74,"url":71},[27,85,28,86,87],"生态保护","植被覆盖","黄土高原",[89,90],"山西 kNDVI 植被覆盖","XGBoost-SHAP 植被驱动","山西kNDVI植被覆盖-2871","10.3389\u002Ffenvs.2026.1948848",{"doi":92,"openalex_id":94,"authors":95,"venue":74,"cited_by_count":36,"oa_url":109,"card":110,"direction":62,"ingested_from":63},"W7213469060",[96,98,100,102,104,107],{"name":97,"orcid":9},"Jie Chen",{"name":99,"orcid":9},"Yi Hou",{"name":101,"orcid":9},"Jianhua Xue",{"name":103,"orcid":9},"Jianhua Ni",{"name":105,"orcid":106},"Hao Liu","https:\u002F\u002Forcid.org\u002F0000-0001-8903-0983",{"name":108,"orcid":9},"Pengxiang Gao","https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fenvironmental-science\u002Farticles\u002F10.3389\u002Ffenvs.2026.1948848\u002Fpdf",{"tldr":111,"method":112,"finding":113,"direction":62,"opportunity":114},"基于kNDVI与XGBoost-SHAP分析山西2000-2024年植被覆盖时空变化及驱动机制。","kNDVI数据结合Theil-Sen、Mann-Kendall、Hurst指数与","山西植被呈显著上升趋势，98.02%区域改善，驱动因子具非线性与时空差异。","可引入多源遥感与作物物候数据，将kNDVI驱动分析拓展至农田尺度精准管理与碳汇评估。","2026-09-18T23:30:29.212735Z",{"id":117,"title":118,"url":119,"summary":120,"summary_zh":121,"content":9,"source_name":122,"source_url":119,"published_at":75,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":123,"score_detail":124,"sources":128,"tags":130,"search_phrases":134,"slug":137,"view_count":36,"doi":138,"paper":139,"created_at":162},2804,"A Novel Data-Driven Framework for Stubble Burning Detection Using NASA FIRMS and Machine Learning","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs12524-026-02584-x","A Novel Data-Driven Framework for Stubble Burning Detection Using NASA FIRMS and Machine Learning。Journal of the Indian Society of Remote Sensing","一种基于NASA FIRMS和机器学习的新型数据驱动秸秆焚烧检测框架。《印度遥感学会杂志》","Journal of the Indian Society of Remote Sensing",76,{"impact":125,"substance":126,"depth":78,"authority":20,"freshness":13,"relevant":22,"comment":127},15,20,"该论文提出基于NASA FIRMS卫星数据与机器学习的秸秆焚烧检测新框架，方法新颖、数据规模明确，对农业遥感监测有实质参考价值，但属细分领域学术进展，影响力限于专业圈层。",[129],{"name":122,"url":119},[131,27,28,132,133],"农业人工智能","秸秆焚烧","卫星数据",[135,136],"农业人工智能 卫星数据 机器学习 秸秆焚烧","农业人工智能 卫星数据","农业人工智能卫星数据机器学习秸秆焚烧-2804","10.1007\u002Fs12524-026-02584-x",{"doi":138,"openalex_id":140,"authors":141,"venue":122,"cited_by_count":36,"oa_url":9,"card":157,"direction":60,"ingested_from":63},"W7213455429",[142,145,147,150,152,155],{"name":143,"orcid":144},"Mohit Dua","https:\u002F\u002Forcid.org\u002F0000-0001-7071-8323",{"name":146,"orcid":9},"Oshin Rastogi",{"name":148,"orcid":149},"Ashish Saini","https:\u002F\u002Forcid.org\u002F0000-0003-3061-2342",{"name":151,"orcid":9},"Raviya",{"name":153,"orcid":154},"Nidhi Chakravarty","https:\u002F\u002Forcid.org\u002F0000-0002-5454-1561",{"name":156,"orcid":9},"Shelza Dua",{"tldr":158,"method":159,"finding":160,"direction":62,"opportunity":161},"提出基于NASA FIRMS与机器学习的数据驱动框架，用于检测秸秆焚烧。","使用NASA FIRMS火点数据结合机器学习分类算法。","该框架能有效识别秸秆焚烧事件，提升检测精度。","可结合多源遥感与深度学习，提升小尺度焚烧检测与实时预警能力。","2026-09-17T23:30:59.313408Z",{"id":164,"title":165,"url":166,"summary":167,"summary_zh":168,"content":9,"source_name":169,"source_url":166,"published_at":170,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":171,"score_detail":172,"sources":174,"tags":176,"search_phrases":180,"slug":183,"view_count":36,"doi":184,"paper":185,"created_at":210},2791,"Integrating multi-source data and support vector machine to assess the spatio-temporal pattern of land degradation in the Eastern Cape of South Africa","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.indic.2026.101524","Land degradation remains a major environmental challenge, particularly in semi-arid and heterogeneous landscapes, where interactions between vegetation loss and soil exposure are complex and spatially dynamic. This study, therefore, seeks to evaluate the spatial extent of land degradation and drivers over time (2005 - 2025) using Landsat data series and support vector machine (SVM) in the Keiskamma Catchment of South Africa. Degraded land followed a non-monotonic trajectory: it declined from ∼197 km 2 in 2005 to ∼157 km 2 in 2015 (a temporary contraction of 20.3%, consistent with short-term restoration and land-use shifts), before rising sharply and unsustainably to ∼328 km 2 by 2025 (a 108.9% increase relative to 2015, and a net increase of 66.5% over the full two-decade period), largely at the expense of grassland and agricultural land. Furthermore, the findings show that soil-sensitive indicators, particularly BSI and SWIR spectral bands, play a crucial role in determining degraded land. In contrast, vegetation indices such as NDVI contribute less under degraded conditions because degraded areas were severely dominated by exposed soil rather than vegetation. Correlation matrix analysis further reveals a temporal shift from mixed soil–vegetation spectral relationships toward strong soil-dominated reflectance patterns by 2025, indicating advanced degradation stages. Overall, the integration of SVM classification with VIF and SHAP provides a transparent, reliable, and spatially explicit framework for monitoring land degradation. The findings support land degradation neutrality monitoring and provide critical insights for sustainable land-management planning in support of Sustainable Development Goal (SDG) 15.3.","土地退化仍然是一项重大环境挑战，尤其是在半干旱和异质性景观中，植被丧失与土壤裸露之间的相互作用复杂且具有空间动态性。因此，本研究旨在利用Landsat数据序列和支持向量机（SVM），评估南非Keiskamma集水区2005—2025年间土地退化的空间范围及其驱动因素。退化土地呈非单调变化轨迹：从2005年的约197 km²下降至2015年的约157 km²（暂时收缩20.3%，与短期恢复和土地利用变化相一致），随后急剧且不可持续地上升至2025年的约328 km²（较2015年增加108.9%，在整个二十年期间净增加66.5%），且主要以草地和农用地为代价。此外，研究结果表明，土壤敏感指标，尤其是BSI和SWIR光谱波段，在判定退化土地方面发挥着关键作用。相比之下，NDVI等植被指数在退化条件下贡献较小，因为退化区域严重以裸露土壤为主，而非植被。相关矩阵分析进一步揭示，到2025年，光谱关系由土壤—植被混合关系向强烈的土壤主导反射模式发生时间转变，表明退化已进入后期阶段。总体而言，将SVM分类与VIF和SHAP相结合，为监测土地退化提供了一个透明、可靠且具有空间显式性的框架。研究结果支持土地退化零增长监测，并为支持可持续发展目标（SDG）15.3的可持续土地管理规划提供了关键见解。","Environmental and Sustainability Indicators","2026-09-16T00:00:00Z",77,{"impact":125,"substance":18,"depth":78,"authority":20,"freshness":21,"relevant":22,"comment":173},"基于Landsat时序与SVM\u002FSHAP的南非土地退化监测研究，方法透明可复现，对农业遥感与土地退化中性监测有参考价值，但属区域案例、非国内三农直接政策信息。",[175],{"name":169,"url":166},[177,27,178,179,28],"农业遥感","可持续发展","土地退化",[181,182],"可持续发展 农业遥感 土地退化 机器学习","可持续发展 农业遥感","可持续发展农业遥感土地退化机器学习-2791","10.1016\u002Fj.indic.2026.101524",{"doi":184,"openalex_id":186,"authors":187,"venue":169,"cited_by_count":36,"oa_url":204,"card":205,"direction":62,"ingested_from":63},"W7213298151",[188,190,193,196,198,201],{"name":189,"orcid":9},"Mandisa Zameko",{"name":191,"orcid":192},"Kgabo Humphrey Thamaga","https:\u002F\u002Forcid.org\u002F0000-0002-2305-9975",{"name":194,"orcid":195},"Mthunzi Mndela","https:\u002F\u002Forcid.org\u002F0000-0002-2384-6856",{"name":197,"orcid":9},"Matthieu Tshanga",{"name":199,"orcid":200},"Nobert Tafadzwa Mukomberanwa","https:\u002F\u002Forcid.org\u002F0009-0003-1896-9813",{"name":202,"orcid":203},"Mohamed Zhran","https:\u002F\u002Forcid.org\u002F0000-0002-1112-387X","https:\u002F\u002Fwww.sciencedirect.com\u002Fscience\u002Farticle\u002Fpii\u002FS2665972726004137\u002Fpdf",{"tldr":206,"method":207,"finding":208,"direction":62,"opportunity":209},"用Landsat与SVM评估南非Keiskamma流域2005-2025年土地退化时空格局。","Landsat时序数据、SVM分类，结合VIF与SHAP做特征解释。","退化面积先降后升，2025年达328km²，土壤光谱指标比NDVI更关键。","可将该SVM-SHAP框架迁移到其他半干旱区，并耦合气候与土地利用驱动做退化预警。","2026-09-17T23:30:36.054539Z",{"id":212,"title":213,"url":214,"summary":215,"summary_zh":216,"content":9,"source_name":217,"source_url":214,"published_at":170,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":76,"score_detail":218,"sources":221,"tags":223,"search_phrases":227,"slug":230,"view_count":36,"doi":231,"paper":232,"created_at":266},2657,"Retrieval of Optically Active and Inactive Water Quality Parameters Using Remote Sensing and Machine Learning: Evidence from Water Hyacinth-Infested Lake Tana, Ethiopia","https:\u002F\u002Fdoi.org\u002F10.3390\u002Frs18183185","Monitoring water quality is critical for protecting freshwater ecosystems and supporting sustainable water resource management. Lake Tana, Ethiopia’s largest freshwater lake, faces increasing agricultural and urban pressures, while conventional monitoring remains costly and spatially constrained. This study developed an integrated Sentinel-2 remote sensing and machine learning framework to estimate chlorophyll-a (Chl-a), turbidity (TU), total nitrogen (TN), and total phosphorus (TP) using 858 in situ observations and Google Earth Engine. Random Forest (RF), Extreme Gradient Boosting (XGB), Artificial Neural Networks (ANN), and Support Vector Regression (SVR) were evaluated using spectral bands, band combinations, and indices. RF provided the best predictions for Chl-a (R2 = 0.94 ± 0.01; RMSE = 2.11 ± 0.18 µg L−1; MARE = 5%) and TP (R2 = 0.91 ± 0.01; RMSE = 0.26 ± 0.01 mg L−1; MARE = 8.7%), whereas XGB performed best for TU (R2 = 0.93 ± 0.01; RMSE = 5.17 ± 0.43 NTU; MARE = 7%) and TN (R2 = 0.94 ± 0.02; RMSE = 0.18 ± 0.02 mg L−1; MARE = 9.9%). The strong predictive performance of RF and XGB across both optically active and inactive parameters demonstrates the capability of the framework to capture complex spectral water quality relationships and support spatially continuous assessment. Significant seasonal differences (p \u003C 0.001) showed higher dry season Chl-a (137.1%) and higher wet season TP (21.7%), TU (7.5%), and TN (3.9%). Long-term paired observations further indicated increases in Chl-a (73.7%), TN (30%), and TP (14.3%) from December 2016 to December 2025 (p \u003C 0.001). Spatial hotspot analysis revealed strong clustering of TU, TN, and TP, particularly around tributary mouths and nearshore areas, highlighting priority zones for monitoring and intervention. Overall, integrating field observations, Sentinel-2 imagery, and machine learning provides an accurate, scalable, and cost-effective approach for monitoring diverse water quality parameters. The framework offers a transferable solution for strengthening freshwater monitoring in data-scarce regions and supporting sustainable management of lakes under increasing water quality pressures.","监测水质对于保护淡水生态系统和支撑可持续水资源管理至关重要。埃塞俄比亚最大的淡水湖——塔纳湖（Lake Tana）面临着日益加剧的农业和城市压力，而传统监测手段仍然成本高昂且受空间限制。本研究开发了一套集成Sentinel-2遥感与机器学习的框架，利用858个原位观测数据和Google Earth Engine估算叶绿素a（Chl-a）、浊度（TU）、总氮（TN）和总磷（TP）。采用光谱波段、波段组合和指数，评估了随机森林（RF）、极端梯度提升（XGB）、人工神经网络（ANN）和支持向量回归（SVR）的性能。RF对Chl-a（R2 = 0.94 ± 0.01；RMSE = 2.11 ± 0.18 µg L−1；MARE = 5%）和TP（R2 = 0.91 ± 0.01；RMSE = 0.26 ± 0.01 mg L−1；MARE = 8.7%）的预测效果最佳，而XGB对TU（R2 = 0.93 ± 0.01；RMSE = 5.17 ± 0.43 NTU；MARE = 7%）和TN（R2 = 0.94 ± 0.02；RMSE = 0.18 ± 0.02 mg L−1；MARE = 9.9%）的预测效果最佳。RF和XGB在光学活性和非光学活性参数上均表现出强大的预测性能，表明该框架能够捕捉复杂的光谱水质关系并支持空间连续评估。显著的季节性差异（p \u003C 0.001）显示旱季Chl-a较高（137.1%），雨季TP（21.7%）、TU（7.5%）和TN（3.9%）较高。长期配对观测进一步表明，从2016年12月至2025年12月，Chl-a（73.7%）、TN（30%）和TP（14.3%）均有所增加（p \u003C 0.001）。空间热点分析揭示了TU、TN和TP的强烈聚集性，尤其是在支流河口和近岸区域，凸显了优先监测和干预区域。总体而言，整合实地观测、Sentinel-2影像和机器学习为监测多种水质参数提供了一种准确、可扩展且具有成本效益的方法。该框架为加强数据稀缺地区的淡水监测以及支持在水质压力日益增大背景下湖泊的可持续管理提供了一种可迁移的解决方案。","Remote Sensing",{"impact":78,"substance":18,"depth":78,"authority":219,"freshness":13,"relevant":22,"comment":220},14,"基于Sentinel-2与机器学习实现湖泊多参数水质反演，方法可迁移至国内农业面源污染与渔业水域监测，数据规模与精度均具参考价值。",[222],{"name":217,"url":214},[27,224,225,28,226],"农业面源污染","智慧渔业","水质监测",[228,229],"农业面源污染 智慧渔业 机器学习 水质监测","农业面源污染 智慧渔业","农业面源污染智慧渔业机器学习水质监测-2657","10.3390\u002Frs18183185",{"doi":231,"openalex_id":233,"authors":234,"venue":217,"cited_by_count":36,"oa_url":214,"card":261,"direction":62,"ingested_from":63},"W7213230686",[235,238,240,243,245,248,250,252,254,256,258],{"name":236,"orcid":237},"Lakachew Y. Alemneh","https:\u002F\u002Forcid.org\u002F0009-0004-3471-3778",{"name":239,"orcid":9},"Daganchew Aklog",{"name":241,"orcid":242},"Ann van Griensven","https:\u002F\u002Forcid.org\u002F0000-0002-2105-6287",{"name":244,"orcid":9},"Minychl G. Dersseh",{"name":246,"orcid":247},"Goraw Goshu","https:\u002F\u002Forcid.org\u002F0000-0001-9629-0126",{"name":249,"orcid":9},"Seleshi Yalew",{"name":251,"orcid":9},"Demesew A. Mhiret",{"name":253,"orcid":9},"Sisay B. Asress",{"name":255,"orcid":9},"Tigistu Wassie Agegnehu",{"name":257,"orcid":9},"Shawl Abebe Desta",{"name":259,"orcid":260},"Samuel Berihun Kassa","https:\u002F\u002Forcid.org\u002F0009-0004-5618-9743",{"tldr":262,"method":263,"finding":264,"direction":62,"opportunity":265},"用Sentinel-2与机器学习反演埃塞俄比亚塔纳湖四类水质参数并分析时空变化。","Sentinel-2影像、Google Earth Engine、858个实测点","RF与XGB精度最高（R²达0.91-0.94），水质参数呈显著季节与年际上升趋势。","可迁移至其他数据稀缺湖泊，探索水葫芦覆盖下水体光谱混合与多源遥感协同反演。","2026-09-16T23:30:28.415573Z",{"id":268,"title":269,"url":270,"summary":271,"summary_zh":272,"content":9,"source_name":217,"source_url":270,"published_at":273,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":274,"score_detail":275,"sources":278,"tags":280,"search_phrases":283,"slug":286,"view_count":36,"doi":287,"paper":288,"created_at":317},2536,"Mapping Native Grass Cover with Random Forest Models: Sentinel-2 Versus Sentinel-2 Combined with Sentinel-1 SAR-Derived GLCM Texture Metrics","https:\u002F\u002Fdoi.org\u002F10.3390\u002Frs18183150","Temperate native grasslands in southeastern Australia have been extensively cleared for agriculture, and the remaining patches are under growing pressure from further land use change, climate variability, and invasive species. Mapping and monitoring their distribution and the cover of native and exotic grasses are critical for their conservation and management. Field-based methods are not always scalable or time-effective, and this study aimed to develop a scalable method to map and monitor the fractional cover-class maps of native C3 and native C4 grass cover as a component of remnant native grasslands on the western outskirts of Melbourne, Victoria, Australia. Field-based reference data for training and validation of random forest machine learning models were collected across multiple sites in 2021. Sentinel-2 optical spectral bands and vegetation indices were used as the primary input data, and Sentinel-1 Synthetic Aperture Radar (SAR)-derived Grey Level Co-occurrence Matrix (GLCM) texture metrics were assessed for their capacity to improve the model. Results show that random forest models trained on Sentinel-2 data without GLCM texture information derived from Sentinel-1 SAR data provided a moderate overall accuracy (C3: 59.1%, C4: 78.1%). Class-specific metrics showed that reliability was highest for better represented lower-cover classes, particularly the 6–25% native C3 class and the 0–5% native C4 class, while higher-cover classes were less reliable because of the limited number of training and validation samples. Grass cover fractions were modelled well for sparse to moderate grass cover, but dense grass cover was not modelled accurately, probably due to limited high-cover samples in the training dataset. Model performance was not improved by the inclusion of Sentinel-1 SAR-derived GLCM texture metrics, indicating that C-band VH-polarised SAR is not sensitive to the fine-scale structural heterogeneity that characterises native grassland ecosystems. Sparse native C3 and C4 grasses could be mapped most reliably in the lower-cover classes as a component of grasslands with optical remote sensing, and the method developed here can now be applied to enable evidence-based management of grasslands, biodiversity conservation and the monitoring of grassland composition in the WGR and elsewhere. Higher-resolution structural datasets and more sophisticated machine learning approaches may be required to accurately predict native C3 and C4 grass cover fractions in denser grasslands.","澳大利亚东南部的温带原生草原已被大面积开垦用于农业，残余斑块正面临土地利用进一步变化、气候变率和入侵物种日益增大的压力。对其分布以及原生和外来草类覆盖度进行制图和监测，对于草原的保护和管理至关重要。基于实地调查的方法并非总是可扩展或省时的，本研究旨在开发一种可扩展的方法，以制图和监测作为残余原生草原组成部分的原生C3和原生C4草类覆盖度的分数覆盖等级图，研究区位于澳大利亚维多利亚州墨尔本西郊。用于训练和验证随机森林机器学习模型的实地参考数据于2021年在多个样点采集。研究以Sentinel-2光学光谱波段和植被指数作为主要输入数据，并评估了Sentinel-1合成孔径雷达（SAR）衍生的灰度共生矩阵（GLCM）纹理指标对提升模型性能的能力。结果表明，仅使用Sentinel-2数据（不含Sentinel-1 SAR衍生的GLCM纹理信息）训练的随机森林模型提供了中等的总体精度（C3：59.1%，C4：78.1%）。分类别指标显示，对于代表性较好的低覆盖度类别，可靠性最高，尤其是6–25%原生C3类别和0–5%原生C4类别，而较高覆盖度类别的可靠性较低，原因是训练和验证样本数量有限。草类覆盖度分数在稀疏至中等草类覆盖条件下建模效果良好，但茂密草类覆盖未能准确建模，可能是由于训练数据集中高覆盖度样本有限。纳入Sentinel-1 SAR衍生的GLCM纹理指标并未改善模型性能，表明C波段VH极化SAR对原生草原生态系统所特有的精细尺度结构异质性不敏感。作为草原的组成部分，稀疏的原生C3和C4草类在低覆盖度类别中利用光学遥感可最可靠地制图，本研究开发的方法现可应用于西维多利亚草原（WGR）及其他地区，以实现基于证据的草原管理、生物多样性保护和草原组成监测。准确预测原生C3和C4草类覆盖度分数可能需要更高分辨率的结构数据集和更先进的机器学习方法。","2026-09-13T00:00:00Z",71,{"impact":17,"substance":126,"depth":19,"authority":219,"freshness":276,"relevant":22,"comment":277},8,"方法新颖、结论明确（SAR纹理未提升精度），对草地遥感监测有参考价值，但属细分领域研究，影响范围有限。",[279],{"name":217,"url":270},[281,27,28,282,86],"智慧农业","草原生态",[284,285],"智慧农业 机器学习 植被覆盖 草原生态","智慧农业 机器学习","智慧农业机器学习植被覆盖草原生态-2536","10.3390\u002Frs18183150",{"doi":287,"openalex_id":289,"authors":290,"venue":217,"cited_by_count":36,"oa_url":270,"card":312,"direction":62,"ingested_from":63},"W7212561645",[291,294,297,300,302,304,307,309],{"name":292,"orcid":293},"Sabah Sabaghy","https:\u002F\u002Forcid.org\u002F0000-0002-9453-8922",{"name":295,"orcid":296},"M. Abuzar","https:\u002F\u002Forcid.org\u002F0000-0002-6101-1307",{"name":298,"orcid":299},"Steve J. Sinclair","https:\u002F\u002Forcid.org\u002F0000-0002-4282-1021",{"name":301,"orcid":9},"Tony Dugdale",{"name":303,"orcid":9},"Vanessa Hutchins",{"name":305,"orcid":306},"Yogendra K. Karna","https:\u002F\u002Forcid.org\u002F0000-0002-2120-4710",{"name":308,"orcid":9},"Jonathan Wilson",{"name":310,"orcid":311},"Kathryn Sheffield","https:\u002F\u002Forcid.org\u002F0000-0003-2624-9739",{"tldr":313,"method":314,"finding":315,"direction":62,"opportunity":316},"用随机森林结合Sentinel-2与Sentinel-1纹理特征，绘制澳洲原生草地C3\u002FC4草覆盖度","2021年野外样点训练随机森林，Sentinel-2光谱与植被指数为主，Sent","仅用Sentinel-2精度中等（C3 59.1%、C4 78.1%），加入SAR纹理未提升，高覆盖","高覆盖度草地样本不足且C波段SAR不敏感，可探索高分辨率结构数据与深度模型提升密草覆盖反演。","2026-09-15T23:30:21.287053Z",{"id":319,"title":320,"url":321,"summary":322,"summary_zh":323,"content":9,"source_name":324,"source_url":321,"published_at":325,"category":12,"cover_url":9,"hotness":13,"is_selected":14,"score":36,"score_detail":326,"sources":328,"tags":330,"search_phrases":332,"slug":335,"view_count":36,"doi":336,"paper":337,"created_at":356},2440,"Mapping urban expansion and land transformation in Dhaka, Bangladesh by fusing night-time lights, thermal, and spectral data via machine learning approaches","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.cacint.2026.100472","This study evaluates urban expansion and land transformation in Dhaka District, Bangladesh, between 2016 and 2024 by integrating multi-source geospatial data with machine learning. The aim was to map urban growth more precisely than conventional spectral indices allow and to characterize its spatial patterns through landscape metrics and change detection. Landsat 8 thermal and spectral data were fused with Visible Infrared Imaging Radiometer Suite (VIIRS) night-time lights, a suite of spectral indices (NDVI, NDBI, SAVI, MNDWI), Gray-Level Co-occurrence Matrix (GLCM) texture features, and road-network data. Three supervised classifiers, Random Forest (RF), Support Vector Machine (SVM), and Gradient Tree Boosting (GTB), together with a majority-voting ensemble, were compared. Random Forest performed best in every study year (overall accuracy 99.57–99.62%; Kappa 0.991–0.992) and was adopted as the urban layer for all subsequent analyses. The urban area expanded by 39.53 km 2 over the period. Vegetation and agricultural land were the predominant source of new urban land (75.6%), whereas direct water-to-urban conversion was minor (below 2%); total water-body area nonetheless declined by roughly 25% (from 79.0 to 58.9 km 2 ), largely through indirect conversion to vegetation and seasonal land rather than through direct urbanization. Landscape metrics indicate increasingly fragmented growth, with patch density rising from 2.46 to 7.82 per 100 ha and Shannon's entropy from 0.45 to 0.49. Zonal analysis identifies the northeastern and southeastern peri -urban fringes as the fastest-growing, dominated by leapfrog and edge development. The findings depict an unsustainable expansion trajectory whose environmental costs stem from the direct loss of vegetation and agricultural land alongside indirect pressure on floodplain water bodies, and they offer a transferable multi-modal framework for monitoring sprawl in Global South megacities. The application of the machine learning and geospatial approach in this study will help urban planners, policymakers, and stakeholders manage Dhaka's rapid and fragmented urban growth.","本研究通过整合多源地理空间数据与机器学习方法，评估了2016年至2024年间孟加拉国达卡地区的城市扩张与土地转型。研究旨在比传统光谱指数更精确地绘制城市增长图景，并通过景观格局指标与变化检测刻画其空间模式。研究将Landsat 8热红外与光谱数据与可见光红外成像辐射仪套件（VIIRS）夜间灯光数据、一组光谱指数（NDVI、NDBI、SAVI、MNDWI）、灰度共生矩阵（GLCM）纹理特征以及道路网络数据进行融合。比较了三种监督分类器——随机森林（RF）、支持向量机（SVM）和梯度树提升（GTB），以及多数投票集成方法。随机森林在各研究年份均表现最佳（总体精度99.57%–99.62%；Kappa系数0.991–0.992），并被采纳为后续所有分析的城市图层。研究期内城市面积扩张了39.53 km²。植被和农业用地是新增城市用地的主要来源（75.6%），而直接的水体向城市用地转化较少（低于2%）；然而，水体总面积仍下降了约25%（从79.0 km²降至58.9 km²），这主要是通过间接转化为植被和季节性土地而非直接城市化实现的。景观格局指标表明增长日益破碎化，斑块密度从每100 ha 2.46上升至7.82，香农熵从0.45上升至0.49。分区分析识别出东北部和东南部城市边缘区为增长最快的区域，以跳跃式和边缘式开发为主。研究结果描绘了一条不可持续的扩张轨迹，其环境代价既来自植被和农业用地的直接丧失，也来自对洪泛平原水体的间接压力，并为监测全球南方特大城市蔓延提供了一个可迁移的多模态框架。本研究中的机器学习与地理空间方法的应用将有助于城市规划者、政策制定者和利益相关者管理达卡快速且破碎化的城市增长。","City and Environment Interactions","2026-09-11T00:00:00Z",{"impact":36,"substance":36,"depth":36,"authority":36,"freshness":36,"relevant":36,"comment":327},"研究孟加拉国达卡城市扩张与土地转化，属城市遥感与城市规划领域，与三农、农业信息化、智慧农业无直接关联，不建议进入每日精选。",[329],{"name":324,"url":321},[27,28,331],"城市扩张",[333,334],"城市扩张 机器学习 遥感监测","城市扩张 机器学习","城市扩张机器学习遥感监测-2440","10.1016\u002Fj.cacint.2026.100472",{"doi":336,"openalex_id":338,"authors":339,"venue":324,"cited_by_count":36,"oa_url":321,"card":351,"direction":60,"ingested_from":63},"W7212257463",[340,342,345,347,349],{"name":341,"orcid":9},"Arpon Sarkar",{"name":343,"orcid":344},"Mafrid Haydar","https:\u002F\u002Forcid.org\u002F0009-0003-9229-0171",{"name":346,"orcid":9},"Farzana Islam Mitu",{"name":348,"orcid":9},"Al Hossain Rafi",{"name":350,"orcid":9},"Sakib Hosan",{"tldr":352,"method":353,"finding":354,"direction":62,"opportunity":355},"融合夜间灯光、热红外与光谱数据，用机器学习精准监测达卡2016-2024年城市扩张与土地转化。","Landsat 8热红外与光谱、VIIRS夜间灯光、NDVI\u002FNDBI等指数、G","城市扩张39.53 km²，75.6%来自植被和农地，水体间接减少约25%，增长日益破碎化。","可将该多模态融合框架迁移至耕地流失预警与城郊农地保护，结合时序模型预测扩张对农业的长期影响。","2026-09-14T23:30:47.470495Z"]