基于经验模态分解与神经网络的信号预测

利用经验模态分解在处理非线性、非平稳信号以及人工神经网络可以较好地处理非线性问题的优点,通过经验模态分解把加入噪声的仿真信号分解成几个本征模态函数分量和一个趋势项,在分解过程中采用两种方法处理端点效应问题,结果表明两种方法都能很好的解决端点问题,然后对每个分量分别运用径向基函数神经网络进行预测,并重构出最后的预测结果。与不经EMD处理直接运用神经网络进行预测及真实数据进行对比,结果表明,相对于直接预测,该方法具有更好的预测效果。     

Hybrid denoising-jittering data processing approach to enhance sediment load prediction of muddy rivers

Successful modeling of hydro-environmental processes widely relies on quantity and quality of accessible data, and noisy data can affect the modeling performance. On the other hand in training phase of any Artificial Intelligence (AI) based model, each training data set is usually a limited sample of possible patterns of the process and hence, might not show the behavior of whole population. Accordingly, in the present paper, wavelet-based denoising method was used to smooth hydrological time series. Thereafter, small normally distributed noises with the mean of zero and various standard deviations were generated and added to the smooth time series to form different denoised-jittered data sets. Finally, the obtained pre-processed data were imposed into Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) models for daily runoff-sediment modeling of the Minnesota River. To evaluate the modeling performance, the outcomes were compared with results of multi linear regression (MLR) and Auto Regressive Integrated Moving Average (ARIMA) models. The comparison showed that the proposed data processing approach which serves both denoising and jittering techniques could enhance the performance of ANN and ANFIS based runoff-sediment modeling of the case study up to 34% and 25% in the verification phase, respectively.     

基于机器学习的稀疏样本下的土壤有机质估算方法

采用GRNN（Generalized Regression Neural Network）和RF（Random Forest）2种机器学习方法构建土壤有机质预测模型,以提高稀疏样本情况下的土壤有机质估算精度。依据北京市大兴区农用地2007年的土壤有机质采样数据,按MMSD准则（Minimization of the Mean of the Shortest Distances）抽稀为8种不同采样密度的样本（分别为2703、1352、676、339、169、85、43、22个样本）,分别采用GRNN、RF和Ordinary kriging对各采样密度下的未知采样点进行预测,采用交叉检验的方式验证各采样密度下未知样点的预测精度。随着采样点密度的下降,样点间的空间自相关性逐渐减弱,半变异函数的拟和精度变差,预测点结果误差增大,预测的置信度降低。当抽稀到43个和22个采样点时,样点间的空间自相关性接近歼灭,半变异函数的决定系数较低且残差较大。普通克里格受到采样点数量和采样密度、样点的空间结构的影响比较明显,其预测精度随采样点数量的下降而下降。在85个采样点及以下时,其预测值与观测值之间没有显著的相关性。GRNN和RF的预测精度受采样密度的影响不大,其预测精度在一个较小的范围内波动,其预测值围绕观测值在一定阈值空间内震荡波动,具有较好的相关性,在85个及以下的采样密度时,预测精度相对普通克里格有较大的提升。普通克里格法不适合在稀疏样本条件下空间插值计算,尤其是在空间自相关性比较弱的情况下。机器学习模型能充分学习土壤间环境信息、样点空间邻近效应信息,兼顾属性相似性和空间自相关,具有更好的稳定性和适应性,不容易受到采样点数量、构型和采样密度等因素的影响,即使在采样点空间自相关性很弱的情况下也能做出稳定预测精度。     

重庆雾的完全预报方法

本方法从动力、统计相结合的角度出发,利用多年历史资料,采用逐步回归方法并辅以技术处理,求得非线性回归方程为PP模型的预报方程,并且利用正压模式输出的两个月数值预报产品进行了试报,结果表明该模型对重庆雾的24小时预报具有一定的能力。     

西藏高原积雪覆盖空间分布及地形影响

利用2000-2014年MOD10A2积雪产品和数字高程模型DEM数据,以积雪覆盖率为指标,在分析西藏高原积雪空间分布特点的基础上,定量研究了高程、坡度和坡向等地形要素对高原积雪时空分布的影响。主要结论有：① 西藏高原积雪的空间分布差异显著,具有中东部念青唐古拉山和周边高山积雪丰富,覆盖率高,而南部河谷和羌塘高原中西部积雪少,覆盖率低的特点。② 海拔越高积雪覆盖率越高,积雪持续时间越长,年内变化越稳定。海拔2 km以下积雪覆盖率不足4%,海拔6 km以上覆盖率达75%。海拔4 km以下年内积雪覆盖呈单峰型分布特点,海拔越高,单峰型越明显;而海拔4 km以上则为双峰型,海拔越高,双峰型越明显。海拔6 km以下积雪覆盖率最低值出现在夏季,而6 km以上则出现在冬季。③ 总体上,高原地形坡度越高积雪覆盖率越高。不同坡向中,北坡积雪覆盖率最高,南坡最低,年内分布呈双峰型,而无坡向的平地积雪覆盖率要小于有坡向的山地,其年内变化呈单峰型分布特点。     

A multiobjective evolutionary optimization method based critical rainfall thresholds for debris flows initiation

At present,most researches on the critical rainfall threshold of debris flow initiation use a linear model obtained through regression.With relatively weak fault tolerance,this method not only ignores nonlinear effects but also is susceptible to singular noise samples,which makes it difficult to characterize the true quantization relationship of the rainfall threshold.Besides,the early warning threshold determined by statistical parameters is susceptible to negative samples(samples where no debris flow has occurred),which leads to uncertainty in the reliability of the early warning results by the regression curve.To overcome the above limitations,this study develops a data-driven multiobjective evolutionary optimization method that combines an artificial neural network(ANN) and a multiobjective evolutionary optimization implemented by particle swarm optimization(PSO).Firstly,the Pareto optimality method is used to represent the nonlinear and conflicting critical thresholds for the rainfall intensity I and the rainfall duration D.An ANN is used to construct a dual-target(dual-task) predictive surrogate model,and then a PSO-based multiobjective evolutionary optimization algorithm is applied to train the ANN and stochastically search the trained ANN for obtaining the Pareto front of the I-D surrogate prediction model,which is intended to overcome the limitations of the existing linear regression-based threshold methods.Finally,a double early warning curve model that can effectively control the false alarm rate and negative alarm rate of hazard warnings are proposed based on the decision space and target space maps.This study provides theoretical guidance for the early warning and forecasting of debris flows and has strong applicability.     

Using very long-range terrestrial laser scanner to analyze the temporal consistency of the snowpack distribution in a high mountain environment

This study demonstrated the usefulness of very long-range terrestrial laser scanning (TLS) for analysis of the spatial distribution of a snowpack, to distances up to 3000 m, one of the longest measurement range reported to date. Snow depth data were collected using a terrestrial laser scanner during 11 periods of snow accumulation and melting, over three snow seasons on a Pyrenean hillslope characterized by a large elevational gradient, steep slopes, and avalanche occurrence. The maximum and mean absolute snow depth error found was 0.5-0.6 and 0.2-0.3 m respectively, which may result problematic for areas with a shallow snowpack, but it is sufficiently accurate to determine snow distribution patterns in areas characterized by a thick snowpack. The results indicated that in most cases there was temporal consistency in the spatial distribution of the snowpack, even in different years. The spatial patterns were particularly similar amongst the surveys conducted during the period dominated by snow accumulation (generally until end of April), or amongst those conducted during the period dominated by melting processes (generally after mid of April or early May). Simple linear correlation analyses for the 11 survey dates, and the application of Random Forests analysis to two days representative of snow accumulation and melting periods indicated the importance of topography to the snow distribution. The results also highlight that elevation and the Topographic Position index (TPI) were the main variables explaining the snow distribution, especially during periods dominated by melting. The intra- and inter-annual spatial consistency of the snowpack distribution suggests that the geomorphological processes linked to presence/absence of snow cover act in a similar way in the long term, and that these spatial patterns can be easily identified through several years of adequate monitoring.     

水稻土有机碳密度的空间预测分析——以浙江省长兴县为例

准确预测未采样区域SOC密度,是研究SOC演变趋势和探索土壤固碳作用对缓解全球气候变化的基础。采用泛克里格法(Universal Kriging,UK)和土壤类型法(pedological professional knowledge-based method,PKB),分别对长兴县水稻土有机碳密度进行了预测,其中,UK直接以长兴水稻土剖面资料为源数据、PKB以长兴水稻土剖面数据和长兴1∶5万数字土壤图为源数据进行预测。根据平均绝对误差(MAE)及均方根误差(RMSE)大小,评价了两种方法在县域尺度土壤有机碳密度空间预测效果。结果表明:UK的MAE(31.2)、RMSE(52.5)均大于PKB的MAE(24.7)、RMSE(43.1),说明PKB法的预测效果较好,UK法相对较差。研究表明,对土壤类型、土壤母质,以及剖面点位置等信息的综合考虑能使PKB法更好地表达土壤属性的空间特征,也更适于县域尺度土壤有机碳密度的空间预测。     

A New Method to Predict Erythrocyte Sedimentation Rate with Natural Geographical Factors and Location by Case-based Reasoning: A Case Study of China

Reference values of erythrocyte sedimentation rate(ESR)are the key to interpret ESR blood test in clinic.The common local reference ESR values are more accuracy in blood test that are established with natural geographical factors by using the multiple linear regression(MLR)model and the artificial neural network(ANN).These knowledge-based methods have limitations since the knowledge domains of ESR and natural geographical factors are limited.This paper presents a new cases-depended model to establish reference ESR values with natural geographical factors and location using case-based reasoning(CBR)since knowledge domain of ESR and geographical factors is weak.Overall 224 local normal ESR values of China that calculated from 13623 samples were obtained,and the corresponding natural geographical factors and location that include altitude,sunshine hours,relative humidity,temperature,precipitation,annual temperature range and annual average wind speed were obtained from the National Geomatics Center of China.CBR was used to predict the unseen local reference ESR values with cases.The average absolute deviation(AAD),mean square error(MSE),prediction accuracy(PA),and Pearson correlation coefficient(r)between the observed and estimated data of proposed model is 33.07%,9.02,66.93% and 0.78,which are better than those of ANN and MLR model.The results show that the proposed model provides higher prediction accuracy than those of the artificial neural network and multiple linear regression models.The predicted values are very close to the observed values.Model results show significant agreement of cases data.Consequently,the model is used to predict the unseen local reference ESR with natural geographical factors and location.In spatial,the highest ESR reference areas are distributed in the southern-western district of China that includes Sichuan,Chongqing,Guangxi and Guizhou provinces,and the reference ESR values are greater than 23 mm/60 min.The higher ESR reference values are distributed in the middle part and northern-eastern of China which include Hubei,Henan,Shaanxi,Shanxi,Jilin and Heilongjiang provinces,and the reference ESR values are greater than 18 mm/60min.The lowest ESR reference values are distributed in the northern-western of China that includes Tibet and Xinjiang,and the reference ESR values are lower than 5 mm/60min.     

An artificial neural network-based snow cover predictive modeling in the higher Himalayas

With trends indicating increase in temperature and decrease in winter precipitation, a significant negative trend in snow-covered areas has been identified in the last decade in the Himalayas. This requires a quantitative analysis of the snow cover in the higher Himalayas. In this study, a nonlinear autoregressive exogenous model, an artificial neural network （ANN）, was deployed to predict the snow cover in the Kaligandaki river basin for the next 30 years. Observed climatic data, and snow covered area was used to train and test the model that captures the gross features of snow under the current climate scenario. The range of the likely effects of climate change on seasonal snow was assessed in the Himalayas using downscaled temperature and precipitation change projection from - HadCM3, a global circulation model to project future climate scenario, under the AIB emission scenario, which describes a future world of very rapid economic growth with balance use between fossil and non-fossil energy sources. The results show that there is a reduction of 9% to 46% of snow cover in different elevation zones during the considered time period, i.e., 2Oll to 2040. The 4700 m to 52oo m elevation zone is the most affected area and the area higher than 5200 m is the least affected. Overall, however, it is clear from the analysis that seasonal snow in the Kaligandaki basin is likely to be subject to substantialchanges due to the impact of climate change.     

Northern-Hemisphere snow cover patterns and formation conditions in winter 2007 and 2012

The Arctic sea ice minimum records appeared in the Septembers of 2007 and 2012, followed by high snow cover areas in the Northern Hemisphere winters. The snow cover distributions show different spatial patterns in these two years: increased snow cover in Central Asia and Central North America in 2007, while increased snow cover in East Asia and northwestern Europe in 2012. The high snow cover anomaly shifted to higher latitudes in winter of 2012 compared to 2007. It is noticed that the snow cover had positive anomaly in 2007 and 2012 with the following conditions: the negative geopotential height and the related cyclonic wind anomaly were favorable for upwelling, and, with the above conditions, the low troposphere and surface air temperature anomaly and water vapor anomaly were favorable for the formation and maintenance of snowfalls. The negative geopotential height, cyclonic wind and low air temperature conditions were satisfied in different locations in 2007 and 2012, resulting in different spatial snow cover patterns. The cross section of lower air temperature move to higher latitudes in winter of 2012 compared to 2007.     

Identifying sensitive areas of adaptive observations for prediction of the Kuroshio large meander using a shallowwater model

Sensitive areas for prediction of the Kuroshio large meander using a 1.5-layer,shallowwater ocean model were investigated using the conditional nonlinear optimal perturbation(CNOP) and first singular vector(FSV) methods.A series of sensitivity experiments were designed to test the sensitivity of sensitive areas within the numerical model.The following results were obtained:(1) the effect of initial CNOP and FSV patterns in their sensitive areas is greater than that of the same patterns in randomly selected areas,with the effect of the initial CNOP patterns in CNOP sensitive areas being the greatest;(2) both CNOP- and FSV-type initial errors grow more quickly than random errors;(3) the effect of random errors superimposed on the sensitive areas is greater than that of random errors introduced into randomly selected areas,and initial errors in the CNOP sensitive areas have greater effects on final forecasts.These results reveal that the sensitive areas determined using the CNOP are more sensitive than those of FSV and other randomly selected areas.In addition,ideal hindcasting experiments were conducted to examine the validity of the sensitive areas.The results indicate that reduction(or elimination) of CNOP-type errors in CNOP sensitive areas at the initial time has a greater forecast benefit than the reduction(or elimination) of FSVtype errors in FSV sensitive areas.These results suggest that the CNOP method is suitable for determining sensitive areas in the prediction of the Kuroshio large-meander path.     

新疆阿克苏河流域降水空间变异特征分析

根据阿克苏河流域降水空间观测数据,其降水稀疏且分布不均匀的特点,选取不同模型对降水空间变化规律进行研究,其结果精确性差异很大。通常应用地统计理论研究降水空间变异性,一般只涉及单个变量,传统的多元回归分析虽然涉及多个变量的影响,但缺乏区域化的空间结构特征。揭示具有协同区域化特征的降水空间变异现象及建立其空间分布模型,既要考虑多元信息的空间位置关系,即同一变量在不同地理位置上的相关性,又要考虑多元信息由于空间重复性引起的协同关系,即同一地理位置上不同变量的相关性。本文用阿克苏河流域范围内的降水观测数据建立析取-协克立格模型,考虑高程变量对降水量空间分布的影响,定量地揭示降水区域化变量的空间变异规律,并将其结果用于降水量的空间最优插值。     

Response of snow hydrological processes to a changing climate during 1961 to 2016 in the headwater of Irtysh River Basin,Chinese Altai Mountains

With changing climatic conditions and snow cover regime, regional hydrological cycle for a snowy basin will change and further available surface water resources will be redistributed. Assessing snow meltwater effect on runoff is the key to water safety, under climate warming and fast social-economic developing status. In this study, stable isotopic technology was utilized to analyze the snow meltwater effect on regional hydrological processes, and to declare the response of snow hydrology to climate change and snow cover regime, together with longterm meteorological and hydrological observations, in the headwater of Irtysh River, Chinese Altai Mountains during 1961-2015. The average δ~(18) O values of rainfall, snowfall, meltwater, groundwater and river water for 2014–2015 hydrological year were-10.9‰,-22.3‰,-21.7‰,-15.7‰ and-16.0‰, respectively.The results from stable isotopes, snow melting observation and remote sensing indicated that the meltwater effect on hydrological processes in Kayiertesi River Basin mainly occurred during snowmelt supplying period from April to June. The contribution of meltwater to runoff reached 58.1% during this period, but rainfall, meltwater and groundwater supplied 49.1%, 36.9% and 14.0% of water resource to annual runoff, respectively. With rising air temperature and increasing snowfall in cold season, the snow water equivalent(SWE) had an increasing trend but the snow cover duration declined by about one month including 13-day delay of the first day and 17-day advancement of the end day during 1961–2016. Increase in SWE provided more available water resource. However, variations in snow cover timing had resulted in redistribution of surface water resource, represented by an increase of discharge percentage in April and May, and a decline in Juneand July. This trend of snow hydrology will render a deficit of water resource in June and July when the water resource demand is high for agricultural irrigation and industrial manufacture.     

基于小波分解与GA-LSSVM的GPS可降水量短临预报

针对GPS可降水量时间序列具有非线性、非平稳性的特征,研究一种基于小波分解（WD）、遗传算法（GA）和最小二乘支持向量机（LSSVM）的GPS可降水量短临预报方法。先采用小波分解将GPS可降水量时间序列分解成便于预报的低频分量和高频分量;然后利用遗传算法优化LSSVM参数,进而对各分量建立预报模型;再将各分量预报结果进行叠加重构得到最终预报结果。选取两组数据进行实验,并将预报结果分别与LSSVM和遗传小波神经网络（GA-WNN）预报结果进行对比。结果表明,该组合模型具有良好的泛化能力,可有效解决神经网络易陷于局部极小的问题,提高了全局预报精度。     

基于关联规则与随机森林的地震多属性砂体厚度预测

地震属性技术是砂体厚度预测的重要手段,由于目前可从地震数据中提取的地震属性种类较多,在利用地震属性技术前,必须优化出对砂体厚度最敏感的地震属性组合,以减少地震属性信息的重复与冗余。为此提出了一种联合关联规则与随机森林回归算法的地震多属性砂体厚度预测方法。随机森林回归算法能够建立地震多属性与砂体厚度之间的非线性关系,并能进行属性选择,但是该算法无法识别地震多种属性中的冗余特征。关联规则能够发现地震属性之间的非线性关联,并能借助卡方检验消除地震属性间的冗余性。分别采用了随机森林回归算法（RFR）、联合关联规则与随机森林回归（AR-RFR）及BP神经网络回归的算法（AR-BP）对滩坝砂岩合成模型和某实际工区进行了砂体厚度预测。对比结果表明,基于关联规则的属性优选得到的属性间相关性低,关联规则与随机森林算法的结合提高了砂体厚度的预测精度。数值实验证明了该方法的有效性。      

Spatial distribution of snow depth based on geographically weighted regression kriging in the Bayanbulak Basin of the Tianshan Mountains,China

Snow depth is a general input variable in many models of agriculture,hydrology,climate and ecology.This study makes use of observational data of snow depth and explanatory variables to compare the accuracy and effect of geographically weighted regression kriging(GWRK)and regression kriging(RK)in a spatial interpolation of regional snow depth.The auxiliary variables are analyzed using correlation coefficients and the variance inflation factor(VIF).Three variables,Height,topographic ruggedness index(TRI),and land surface temperature(LST),are used as explanatory variables to establish a regression model for snow depth.The estimated spatial distribution of snow depth in the Bayanbulak Basin of the Tianshan Mountains in China with a spatial resolution of 1 km is obtained.The results indicate that 1)the result of GWRK's accuracy is slightly higher than that of RK(R~2=0.55 vs.R~2=0.50,RMSE(root mean square error)=0.102 m vs.RMSE=0.077 m);2)for the subareas,GWRK and RK exhibit similar estimation results of snow depth.Areas in the Bayanbulak Basin with a snow depth greater than 0.15m are mainly distributed in an elevation range of 2632.00–3269.00 m and the snow in this area comprises 45.00–46.00% of the total amount of snow in this basin.However,the GWRK resulted in more detailed information on snow depth distribution than the RK.The final conclusion is that GWRK is better suited for estimating regional snow depth distribution.     

基于手机定位数据的城市人口分布近实时预测

精细时空尺度下城市人口分布的近实时预测可为优化公共资源配置、协助城市交通诱导、制定公共安全应急预案、探索城市居民活动规律等提供重要科学依据。本文采用城市手机定位数据,基于时间序列分析方法,分别建立参数预测模型和非参数预测模型,对精细尺度下的城市人口空间分布开展近实时预测。预测结果表明,基于时间序列分析方法的预测模型可为精细尺度下的城市人口分布近实时预测提供方法支持;在本文实验条件下,从人口规模、时空分布、多时间尺度、特殊事件等多个角度评估模型精度,非参数预测模型其预测误差均小于参数预测模型,且预测结果更为稳定。     

Comparison of different sampling strategies for debris flow susceptibility mapping: A case study using the centroids of the scarp area,flowing area and accumulation area of debris flow watersheds

The quality of debris flow susceptibility mapping varies with sampling strategies. This paper aims at comparing three sampling strategies and determining the optimal one to sample the debris flow watersheds. The three sampling strategies studied were the centroid of the scarp area(COSA), the centroid of the flowing area(COFA), and the centroid of the accumulation area(COAA) of debris flow watersheds. An inventory consisting of 150 debris flow watersheds and 12 conditioning factors were prepared for research. Firstly, the information gain ratio(IGR) method was used to analyze the predictive ability of the conditioning factors. Subsequently, 12 conditioning factors were involved in the modeling of artificial neural network(ANN), random forest(RF) and support vector machine(SVM). Then, the receiver operating characteristic curves(ROC) and the area under curves(AUC) were used to evaluate the model performance. Finally, a scoring system was used to score the quality of the debris flow susceptibility maps. Samples obtained from the accumulation area have the strongest predictive ability and can make the models achieve the best performance. The AUC values corresponding to the best model performance on the validation dataset were 0.861, 0.804 and 0.856 for SVM, ANN and RF respectively. The sampling strategy of the centroid of the scarp area is optimal with the highest quality of debris flow susceptibility maps having scores of 373470, 393241 and 362485 for SVM, ANN and RF respectively.     

确定性系数与随机森林模型在云南芒市滑坡易发性评价中的应用

编制科学的滑坡易发性分区图,可以有效降低灾害带来的损失。以云南省芒市为研究区,利用确定性系数模型（certainty factor,简称CF）方法计算各个因子的敏感值,作为随机森林（random forests,简称RF）的分类数据,选取合适的训练数据和最优化的模型参数进行模型预测,从而对研究区进行滑坡易发性评价分区。采用频率比方法将连续性因子离散化,从而通过确定性系数计算因子不同区间的滑坡易发性,同时利用CF先验模型,对研究区负样本进行选取。通过计算袋外误差得到最优化的RF参数,随后利用RF模型对研究区模型进行训练及预测。绘制ROC曲线和三维遥感影像对预测模型结果分别进行定量和定性评价,结果表明,所得到的模型精度为91%,优于随机抽样得到的结果。最后,采用平均基尼不纯度减少和平均准确度下降两种计算方法计算、评价了研究区各个因子的重要性。基于以上对研究区进行的滑坡易发性评价结果,可以为该区灾害风险评估和管理提供依据。