基于随机森林模型的干旱绿洲区张掖盆地地下水水质评价

为合理准确评价地下水水质,建立了基于随机森林（RF）模型的地下水水质评价模型,并根据张掖盆地81个地下水采样点的pH值、Cl^-、SO₄^2-、NO₃^-、Na⁺、NH₄⁺含量及总硬度的监测数据,对研究区的地下水水质进行了综合评价。结果表明：盆地地下水水质主要为Ⅱ、Ⅲ、Ⅳ类水,其中甘州区地下水埋藏较深,水体不容易受到来自地面的污染,水质较好,大多数地方为Ⅱ类水;临泽县和高台县地下水埋藏较浅,水质较差,大多数地方为Ⅲ类水,尤其高台县的水位最浅,再加上地处河段下游,污染更为严重,部分地区达到Ⅳ类。根据指标的重要性度量发现影响研究区域地下水水质的主要因子是NO₃^-含量;其次是NH₄⁺、SO₄^2-、Na⁺、Cl^-含量及总硬度、pH值。为验证模型的有效性,将地下水水质评价结果与基于支持向量机（SVM）和人工神经网络（ANN）的地下水水质综合评价模型模拟结果进行对比,3个模型均能很好地评价研究区地下水水质,但RF模型的评价结果更为准确。     

基于遥感和机器学习的内陆水体水深反演技术

文章主要根据机器学习算法（随机森林算法和极端梯度提升算法）和遥感水深反演的原理,利用Sentinel_2多光谱卫星数据和无人船实测水深数据,对内陆水体——梅州水库建立了随机森林（RF）、极端梯度提升（XGBoost）和支持向量机（SVM）水深反演模型,并对反演结果进行对比分析。结果表明：1）RF的训练精度为97%,测试精度为0.80;XGBoost模型的训练精度为97%,测试精度为0.79;SVM的训练精度为90%,测试精度为0.78。说明了在水深预测方面RF模型和XGBoost模型比SVM模型表现更好,对各个区段的水深值较为敏感。2）根据运行时间考察各个模型的效率,其中RF模型从读取数据至输出结果耗时3.92 s;XGBoost模型4.26 s;SVM模型6.66 s。因此,在反演精度和效率上RF模型优于XGBoost模型优于SVM模型,且RF模型的预测结果图细节更加丰富,轮廓更加分明;XGBoost模型次之,但总体效果也较好;SVM模型表现最差。由此可知,机器学习水深反演模型获得的水深结果精度明显提高,解决了传统水深反演模型精度不高的问题。     

基于时变权重组合与贝叶斯修正的中长期径流预报

高精度的中长期径流预报信息是水资源规划管理与水利工程经济运行的重要基础支撑。论文在组合预报与误差修正2类径流预报后处理方法串联应用的技术框架下,考虑径流的高度非平稳与非线性等特征,提出了基于时变权重组合和贝叶斯修正的中长期径流预报方法。应用该方法开展了云南龙江水库年、月入库径流预报的实例研究,结果表明时变权重组合平衡了已建立的随机森林与支持向量机模型在建模期与检验期预报性能的差异,经贝叶斯修正后的预报精度接近或优于两阶段各自的最优单一模型。根据年径流预报结果判断水文年型的正确率达到77.2%,月预报径流的确定性系数超过0.90。因此,该方法在提升中长期径流预报精度方面具有积极效果。     

基于机器学习模型的区域土壤重金属空间预测精度比较研究

为识别区域土壤重金属的空间变异特征并厘清其影响因素,本研究构建了多元线性回归（MLR）、弹性网络回归（ENR）、随机森林（RF）、随机梯度提升（SGB）、堆叠（stacking）集成模型、反向传播神经网络（BP-ANN）、基于模型平均的神经网络集成（avNNet）、线性核支持向量机（SVM-L）和高斯核支持向量机（SVM-R）共九种机器学习模型,利用山东省中部土壤重金属（Cd、Cu、Hg、Pb和Zn）和环境辅助变量数据,开展区域土壤重金属空间预测精度比较研究。结果表明：RF对五种重金属空间预测的决定系数（R²）介于0.263~0.448之间,平均绝对误差（MAE）和均方根误差（RMSE）分别小于8.408和10.636,预测值/实际值（P/O）均接近于1,对五种重金属的预测效果均较为理想,是研究区土壤重金属空间预测的最优模型;SVM-R整体预测性能仅次于RF,各项精度评价指标均相对稳健,可作为备选模型;其余七种模型的预测性能均明显低于RF和SVM-R。RF的空间预测结果显示,研究区五种重金属呈现出相似的空间分布格局,含量均由研究区东北部向西南部递减,包括东北部、北部和南部3个高值区,且高值区与当地工业–交通密集区的分布格局一致,反映出人类活动是研究区土壤重金属空间分异的主要影响因素。本研究可为区域土壤污染调查、评价和管控提供科学参考。     

Boosting for Mineral Prospectivity Modeling: A New GIS Toolbox

With an increasing demand for raw materials, predictive models that support successful mineral exploration targeting are of great importance. We evaluated different machine learning techniques with an emphasis on boosting algorithms and implemented them in an ArcGIS toolbox. Performance was tested on an exploration dataset from the Iberian Pyrite Belt (IPB) with respect to accuracy, performance, stability, and robustness. Boosting algorithms are ensemble methods used in supervised learning for regression and classification. They combine weak classifiers, i.e., classifiers that perform slightly better than random guessing to obtain robust classifiers. Each time a weak learner is added; the learning set is reweighted to give more importance to misclassified samples. Our test area, the IPB, is one of the oldest mining districts in the world and hosts giant volcanic-hosted massive sulfide (VMS) deposits. The spatial density of ore deposits, as well as the size and tonnage, makes the area unique, and due to the high data availability and number of known deposits, well-suited for testing machine learning algorithms. We combined several geophysical datasets, as well as layers derived from geological maps as predictors of the presence or absence of VMS deposits. Boosting algorithms such as BrownBoost and Adaboost were tested and compared to Logistic Regression (LR), Random Forests (RF) and Support Vector machines (SVM) in several experiments. We found performance results relatively similar, especially to BrownBoost, which slightly outperformed LR and SVM with respective accuracies of 0.96 compared to 0.89 and 0.93. Data augmentation by perturbing deposit location led to a 7% improvement in results. Variations in the split ratio of training and test data led to a reduction in the accuracy of the prediction result with relative stability occurring at a critical point at around 26 training samples out of 130 total samples. When lower numbers of training data were introduced accuracy dropped significantly. In comparison with other machine learning methods, Adaboost is user-friendly due to relatively short training and prediction times, the low likelihood of overfitting and the reduced number of hyperparameters for optimization. Boosting algorithms gave high predictive accuracies, making them a potential data-driven alternative for regional scale and/or brownfields mineral exploration.

     

运用小波变换与支持向量机耦合模型(WA-SVM)预测干旱区地下水埋深

准确预测干旱区地下水埋深,对区域地下水资源的合理开发利用与生态环境保护具有十分重要的意义。以额济纳盆地3个地下水埋深观测井为对象,运用小波变换与支持向量机耦合模型（WA-SVM）对观测井未来1个月的地下水埋深进行了短期预测。为检验WA-SVM的有效性,将模拟结果与未经小波变换的SVM模型进行了对比。结果表明：在对干旱区地下水埋深进行短期预测时,相较于SVM模型,WA-SVM模型的预测精度显著提高。WA-SVM模型在干旱区地下水埋深预测中有更好的适用性,可以为干旱地区地下水埋深动态预测提供新的方法和思路,是资料有限的条件下地下水埋深预测的有效方法。     

基于土壤水分和气象要素的林火预报研究——以广东省为例

基于国内现行的森林火险气象指数和单因子火险贡献度模型,以及逻辑回归模型和随机森林模型,在林火预报中引入微波遥感土壤水分信息,使用MCD14DL火点数据集和地面气象观测资料对广东省不同时间尺度的林火发生概率进行预测。结果表明：逻辑回归模型和随机森林模型构建的林火预测模型显著优于现行的森林火险气象指数和单因子火险贡献度模型,预测精度提升约20%。其中,随机森林模型对林火频数的解释程度最高（两者相关系数为0.476）。此外,加入微波土壤水分信息后,相较原有的基于气象要素的林火预测模型,2种机器学习模型的预测精度均略有提升,体现了表层土壤水分信息在林火预报中的重要性。研究可为高效提取对地观测信息,以改进华南地区不同时间尺度的林火预报工作提供参考。     

Machine learning predictions of passive microwave brightness temperature over snow-covered land using the special sensor microwave imager (SSM/I)

Abstract

Two different forms of machine learning – an artificial neural network (ANN) and a support vector machine (SVM) – are used to estimate passive microwave (PMW) brightness temperatures (Tb) as observed by the special sensor microwave imager (SSM/I) satellite sensor over snow- covered land in North America. Both techniques reasonably reproduce unbiased estimates of SSM/I observations at 19.35 and 37.0 GHz for both vertically- and horizontally-polarized channels. When compared against SSM/I observations not used during training, domain-averaged statistics from 1 September 1987 to 1 September 2002 yielded a root mean squared error (RMSE) of less than 9 K for all frequency and polarization combinations examined in this study. Even though both ML techniques reasonably reproduced SSM/I Tb observations, the SVM outperformed the ANN because the SVM: (1) better captured the high-frequency (i.e. day-to-day) temporal characteristics in the Tb observations across the majority of the study domain, (2) better reproduced the spatial variability as a function of snow classification, and (3) yielded greater sensitivity to snow-related input variables during the estimation of PMW Tb. These findings reinforce previous research of SVM-based estimation of PMW Tb employing observations from the advanced microwave scanning radiometer.     

ORELM: A Novel Machine Learning Approach for Prediction of Flyrock in Mine Blasting

Blast-induced flyrock is a hazardous and undesirable phenomenon that may occur in surface mines, especially when blasting takes place near residential areas. Therefore, accurate prediction of flyrock distance is of high significance in the determination of the statutory danger area. To this end, there is a practical need to propose an accurate model to predict flyrock. Aiming at this topic, this study presents two machine learning models, including extreme learning machine (ELM) and outlier robust ELM (ORELM), for predicting flyrock. To the best of our knowledge, this is the first work that investigates the use of ORELM model in the field of flyrock prediction. To construct and verify the proposed ELM and ORELM models, a database including 82 datasets has been collected from the three granite quarry sites in Malaysia. Additionally, artificial neural network (ANN) and multiple regression models were used for comparison. According to the results, both ELM and ORELM models performed satisfactorily, and their performances were far better compared to the performances of ANN and multiple regression models.

     

基于机器学习模型的海河北系干旱预测研究

提高干旱预测精度能为流域干旱应对及风险防范提供可靠数据支撑,构建比选合适的干旱模型是当前研究的热点。研究以4个时间尺度（3、6、9、12月）标准化降水指数（SPI）为表征指标,利用小波神经网络（WNN）、支持向量回归（SVR）、随机森林（RF）三种机器学习算法分别构建了海河北系干旱预测模型,利用Kendall、K-S、MAE三种检验方法判定模型表现及其稳定性。研究表明：（1） WNN、SVR模型呈现结果在不同时间尺度SPI存在差异,WNN最适合12个月尺度SPI干旱预测;SVR最适合6个月尺度SPI干旱预测。（2） 对3、12个月尺度SPI,RF预测性能最优（Kendall>0.898,MAE<0.05）;对6、9个月尺度SPI,SVR预测性能最优（Kendall>0.95,MAE<0.04）。（3） 模型预测性能稳定性存在区别,RF预测稳定性最高,其次为SVR。（4） 构建的三种模型表现异同主要是因为SVR转为凸优化问题解决了WNN易陷入局部最优解的不足,从而提高了模型预测性能,RF集成多样化回归树,降低了弱学习器的负面影响,提高了模型预测准确率及稳定性,同时,RF处理包含噪声的降水数据的能力更强。     

Strength of Stacking Technique of Ensemble Learning in Rockburst Prediction with Imbalanced Data: Comparison of Eight Single and Ensemble Models

Rockburst is a common dynamic geological hazard, severely restricting the development and utilization of underground space and resources. As the depth of excavation and mining increases, rockburst tends to occur frequently. Hence, it is necessary to carry out a study on rockburst prediction. Due to the nonlinear relationship between rockburst and its influencing factors, artificial intelligence was introduced. However, the collected data were typically imbalanced. Single algorithms trained by such data have low recognition for minority classes. In order to handle the problem, this paper employed stacking technique of ensemble learning to establish rockburst prediction models. In total, 246 sets of data were collected. In the preprocessing stage, three data mining techniques including principal component analysis, local outlier factor and expectation maximization algorithm were used for dimension reduction, outlier detection and outlier substitution, respectively. Then, the pre-processed data were split into a training set (75%) and a test set (25%) with stratified sampling. Based on the four classical single intelligent algorithms, namely k-nearest neighbors (KNN), support vector machine (SVM), deep neural network (DNN) and recurrent neural network (RNN), four ensemble models (KNN–RNN, SVM–RNN, DNN–RNN and KNN–SVM–DNN–RNN) were built by stacking technique of ensemble learning. The prediction performance of eight models was evaluated, and the differences between single models and ensemble models were analyzed. Additionally, a sensitivity analysis was conducted, revealing the importance of input variables on the models. Finally, the impact of class imbalance on the prediction accuracy and fitting effect of models was quantitatively discussed. The results showed that stacking technique of ensemble learning provides a new and promising way for rockburst prediction, which exhibits unique advantages especially when using imbalanced data.

     

Predictive modelling of gold potential with the integration of multisource information based on random forest: a case study on the Rodalquilar area,Southern Spain

Mineral exploration activities require robust predictive models that result in accurate mapping of the probability that mineral deposits can be found at a certain location. Random forest (RF) is a powerful machine data-driven predictive method that is unknown in mineral potential mapping. In this paper, performance of RF regression for the likelihood of gold deposits in the Rodalquilar mining district is explored. The RF model was developed using a comprehensive exploration GIS database composed of: gravimetric and magnetic survey, a lithogeochemical survey of 59 elements, lithology and fracture maps, a Landsat 5 Thematic Mapper image and gold occurrence locations. The results of this study indicate that the use of RF for the integration of large multisource data sets used in mineral exploration and for prediction of mineral deposit occurrences offers several advantages over existing methods. Key advantages of RF include: (1) the simplicity of parameter setting; (2) an internal unbiased estimate of the prediction error; (3) the ability to handle complex data of different statistical distributions, responding to nonlinear relationships between variables; (4) the capability to use categorical predictors; and (5) the capability to determine variable importance. Additionally, variables that RF identified as most important coincide with well-known geologic expectations. To validate and assess the effectiveness of the RF method, gold prospectivity maps are also prepared using the logistic regression (LR) method. Statistical measures of map quality indicate that the RF method performs better than LR, with mean square errors equal to 0.12 and 0.19, respectively. The efficiency of RF is also better, achieving an optimum success rate when half of the area predicted by LR is considered.     

Application of Artificial Neural Networks to Complex Groundwater Management Problems

As water quantity and quality problems become increasingly severe, accurate prediction and effective management of scarcer water resources will become critical. In this paper, the successful application of artificial neural network (ANN) technology is described for three types of groundwater prediction and management problems. In the first example, an ANN was trained with simulation data from a physically based numerical model to predict head (groundwater elevation) at locations of interest under variable pumping and climate conditions. The ANN achieved a high degree of predictive accuracy, and its derived state-transition equations were embedded into a multiobjective optimization formulation and solved to generate a trade-off curve depicting water supply in relation to contamination risk. In the second and third examples, ANNs were developed with real-world hydrologic and climate data for different hydrogeologic environments. For the second problem, an ANN was developed using data collected for a 5-year, 8-month period to predict heads in a multilayered surficial and limestone aquifer system under variable pumping, state, and climate conditions. Using weekly stress periods, the ANN substantially outperformed a well-calibrated numerical flow model for the 71-day validation period, and provided insights into the effects of climate and pumping on water levels. For the third problem, an ANN was developed with data collected automatically over a 6-week period to predict hourly heads in 11 high-capacity public supply wells tapping a semiconfined bedrock aquifer and subject to large well-interference effects. Using hourly stress periods, the ANN accurately predicted heads for 24-hour periods in all public supply wells. These test cases demonstrate that the ANN technology can solve a variety of complex groundwater management problems and overcome many of the problems and limitations associated with traditional physically based flow models.     

基于反射率和亮度温度的锡林郭勒积雪深度估算

积雪是我国西北干旱半干旱区重要的水资源,也是影响全球气候变化的重要因子之一。 目前光学影像反射率和雷达亮温数据是积雪遥感领域的主要数据,本文首次结合两类遥感数据估 算积雪深度,并比较偏最小二乘法和机器学习算法（人工神经网络、支持向量机和随机森林算法） 在积雪深度估算方面的表现。以锡林郭勒盟 2012—2015 年积雪深度数据为例,基于反射率和亮度 温度相结合的积雪深度估算精度优于单个数据源,且随机森林算法表现最好,均方根误差为 2.93 cm,满足实际应用的需求。研究结果对我国西北地区水资源分布、生态环境评估等研究具有重要 意义。     

Application and comparison of two prediction models for groundwater levels: A case study in Western Jilin Province,China

Evaluation and prediction of groundwater levels through specific model(s) helps in forecasting of groundwater resources. Among the different robust tools available, the Integrated Time Series (ITS) and Back-Propagation Artificial Neural Network (BPANN) models are commonly used to empirically forecast hydrological variables. Here, we discuss the modeling process and accuracy of these two methods in assessing their relative advantages and disadvantages based on Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and coefficient of efficiency (CE). The arid and semi-arid areas of western Jilin province of China were chosen as study area owing to the decline of groundwater levels during the past decade mainly due to overexploitation. The simulation results indicated that both ITS and BPANN are accurate in reproducing (fitting) the groundwater levels and the CE are 0.98 and 0.97, respectively. In the validation phase, the comparison of the prediction accuracy of the BPANN and ITS models indicated that the BPANN models is superior to the ITS in forecasting the groundwater levels time series in term of the RMSE, MAE and CE.     

ARIMA和ANN模型的干旱预测适用性研究

开展干旱预测是有效应对干旱风险的前提基础,根据1960-2016年三江平原7个站点逐日降水和气温数据,利用ARIMA和ANN模型对不同时间尺度标准化降水蒸散指数（SPEI）序列进行分析建模预测。借助相关系数R、纳什效率系数NSE、Kendall秩相关系数τ、均方误差MSE和Kolmogorov-Smirnov （K-S）检验对模型的有效性进行了判定,然后分别用ARIMA和ANN模型进行12步预测,并将预测值与实际值进行比较。结果表明：（1） ARIMA模型和ANN模型对SPEI的预测能力都随时间尺度的增加而逐渐提高。（2）两种模型对3、6个月尺度SPEI的预测精度偏低,9、12、24个月的SPEI的预测精度在70%以上;（3）SPEI-9、SPEI-12、SPEI-24三个时间尺度ANN模型的预测精度优于ARIMA模型。     

基于SVM的泥石流危险度评价研究

选取泥石流一次(可能)最大冲出量(L₁)、泥石流发生频率(L₂)、流域面积(S₁)、主沟长度(S₂)、流域最大相对高差(S₃)、流域切割密度(S₆)和泥沙补给段长度比(S₉)7个因子作为泥石流沟谷危险度评价因子,运用支持向量机理论,以云南省37条泥石流沟的259个基础数据为样本进行学习训练和测试,建立泥石流危险度评价的支持向量机模型,通过实例验证,取得良好效果。     

多源遥感影像土地覆盖分类结果一致性评价与集成应用

针对多源遥感影像土地覆盖分类结果一致性与分类精度改进的要求,对两组中等空间分辨率的光学影像进行土地覆盖分类,以支持向量机分类结果为基础,采用Kappa统计量、双错误测量、Q统计量、相同错误率从不同角度评价了不同分类结果的一致性。实验表明,多源遥感数据分类结果总体上常规一致性程度较好,二值先验一致性程度尚可,错误一致性程度较小;不同土地覆盖类别的一致性程度并不相同,有的类别甚至出现不一致现象。提出组合法和替换法两种策略以综合数据优点、实现多传感器数据集成应用,能够有效提高分类精度。     

Extracting urban landmarks from geographical datasets using a random forests classifier

ABSTRACT

Urban landmarks are of significant importance to spatial cognition and route navigation. However, the current landmark extraction methods mainly focus on the visual salience of landmarks and are insufficient for obtaining high extraction accuracy when the size of the geographical dataset varies. This study introduces a random forests (RF) classifier combining with the synthetic minority oversampling technique (SMOTE) in urban landmark extraction. Both GIS and social sensing data are employed to quantify the structural and cognitive salience of the examined urban features, which are available from basic spatial databases or mainstream web service application programming interfaces (APIs). The results show that the SMOTE-RF model performs well in urban landmark extraction, with the values of recall, precision, F-measure and AUC reaching 0.851, 0.831, 0.841 and 0.841, respectively. Additionally, this method is suitable for both large and small geographical datasets. The ranking of variable importance given by this model further indicates that certain cognitive measures – such as feature class, Weibo popularity and Bing popularity – can serve as crucial factors for determining a landmark. The optimal variable combination for landmark extraction is also acquired, which might provide support for eliminating the variable selection requirement in other landmark extraction methods.     

基于随机森林的山西省柳林县黄土滑坡空间敏感性评价

基于随机森林模型,以GF-6影像和ALOS DEM数据为基本信息源,结合高程,地形起伏度及地形湿度等11项因子,对山西省柳林县进行滑坡敏感性空间区划。模型精度评价表明：随机森林模型精度为0.75,支持向量机模型精度为0.7,表明随机森林更适合柳林县的滑坡敏感性评价。指标重要性分析结果表明：高程、坡度、距道路距离以及距河流距离,是影响柳林县滑坡发育的主要因素。敏感性空间区划结果表明：高度敏感区约占柳林县总面积的28%,主要分布在三川河流域的南北边界及邻近区域内,其中贾家垣乡分布面积最广。从时间成本、训练难度、稳定度以及精确度考虑,随机森林模型更适合滑坡敏感性评价这类非线性计算问题。