融合AR模型和MCMC方法的水文模拟不确定性分析

为提高水文模型参数识别的可靠性,融合自回归模型与马尔可夫链-蒙特卡洛方法(auto regressive model based modified Markov Chain-Monte Carlo,AR-MCMC),利用自回归模型刻画残差序列的自相关性,修正MCMC方法中的残差协方差矩阵。通过新疆提孜那甫河流域融雪径流模型(SRM)的案例分析发现:融雪径流模拟的残差序列具有显著的自相关性;修正残差协方差矩阵后,边缘似然值更大;综合考虑多项评价指标,AR-MCMC方法在识别期与验证期推求的预测区间均优于MCMC方法;对比2种方法在识别期与验证期的纳什系数,采用AR-MCMC方法依次为0.86、0.89,而采用MCMC方法依次为0.84、0.87,即AR-MCMC方法获取的模型拟合效果更好。分析结果表明,相对于传统的MCMC方法,AR-MCMC方法能够更好地对研究区融雪径流过程进行模拟预测。

基于陆面水文耦合模式CLHMS的淮河流域水文过程的模拟评估及其不确定性分析

利用最新的CFSR（Climate Forecast System Reanalysis）再分析及观测的降水和地表气温资料驱动陆面水文耦合模式CLHMS（Coupled Landsurface and Hydrologic Model System），对淮河流域1980-2003年共24年的水文水循环过程进行了模拟，系统评估了CLHMS对淮河流域水文过程的模拟能力及其不确定性。分析结果表明，CLHMS模式对淮河流域水文过程具有良好的模拟能力，模式尤其对湿润年份流域的水量平衡以及河道流量的季节、年际变化具有很强的模拟能力，而对降水偏少的干旱年份，模式模拟的河道流量通常会高于观测实况，与实况间存在着一定的偏差，而这也是导致CLHMS对流域水文过程模拟能力存在显著年代际差异的主要原因。基于三组不同降水强迫的流域水文过程模拟结果比较表明，降水驱动资料准确与否是陆面水文模拟最主要的不确定性来源之一，正是由于CFSR再分析降水与观测降水之间存在较大的差异，从而导致CFSR降水驱动下模式模拟的淮河流域河道流量与观测存在较大的偏差，其模拟性能相对较差。进一步分析还表明，可以保持较强降水日变化的时间解集方法，也是保证合理模拟流域水文过程的重要因素。

草原生态系统模式中参数不确定性导致的模拟结果不确定性研究

基于五变量草原生态系统理论模式,应用与参数有关的条件非线性最优扰动(CNOP-P)方法,探讨了由参数不确定性导致的草原生态系统模式模拟结果的不确定性问题。参数的不确定性可能来源于观测和(或)对物理过程描述等的不确定性。选取了五变量草原生态系统模式中具有物理意义的32个模式参数进行数值试验。试验结果表明,对所考察的32个模式参数,在一定的不确定性和给定的优化时刻范围内,单独优化每个参数所得CNOP-Ps的联合模态与同时优化32个参数所得CNOP-P的模态并不相同。比较了上述两类参数误差以及随机参数误差对草原生态系统模拟的差异。随机参数误差与上述优化方法所得参数误差的不确定性范围大小相同。数值结果表明,同时优化32个参数所得CNOP-P类型参数误差使得草原生态系统模拟的不确定性程度最大。这种影响表现在使得草原生态系统转变为沙漠生态系统,或者使得草原生态系统转变为具有更多生草量的草原生态系统。上述数值结果不依赖于优化时间和参数不确定性程度的大小。这些数值结果建议我们应当考虑多参数的非线性相互作用来研究草原生态系统模式模拟的不确定性问题,并且揭示出CNOP-P方法是讨论上述问题的一个有用的工具。

CMIP6火山强迫的气候响应模拟比较计划（VolMIP）概况与评述

为揭示造成火山强迫气候响应模拟不确定性的原因,第六次国际耦合模式比较计划(CMIP6)设立了火山强迫的气候响应模拟比较计划(VolMIP)。该计划由基于历史火山爆发的理想火山扰动试验组成,包括三组主要的试验:第一组关注短期(季节至年际)大气动力响应;第二组关注海气耦合系统的长期(年际至年代际)响应;第三组关注气候系统对火山群的响应。VolMIP旨在通过给定相同的辐射强迫并进行多成员集合模拟,揭示模式对外强迫响应的不确定性,通过设定不同的背景气候态,阐明内部变率和外强迫对气候响应的相对贡献。

Effect of Uncertainty of Grid DEM on TOPMODEL:Evaluation and Analysis

TOPMODEL,a semi-distributed hydrological model,has been widely used.In the process of simulation of the model,Digital Elevation Model(DEM) is used to provide the input data,such as topographic index and distance to the drainage outlet;thus DEM plays an important role in TOPMODEL.This study aims at examining the impacts of DEM uncertainty on the simulation results of TOPMODEL.In this paper,the effects were evaluated mainly from quantitative and qualitative aspects.Firstly,DEM uncertainty was simulated by using the Monte Carlo method,and for every DEM realization,the topographic index and distance to the drainage outlet were extracted.Secondly,the obtained topographic index and the distance to the drainage outlet were input to the TOPMODEL to simulate seven rain-storm-flood events,and four evaluation indices,such as Nash and Sutcliffe efficiency criterion(EFF),sum of squared residuals over all time steps(SSE),sum of squared log residuals over all time steps(SLE) and sum of absolute errors over all time steps(SAE) were recorded.Thirdly,these four evaluation indices were analyzed in statistical manner(minimum,maximum,range,standard deviation and mean value),and effect of DEM uncertainty on TOPMODEL was quantitatively analyzed.Finally,the simulated hydrographs from TOPMODEL using the original DEM and realizations of DEM were qualitatively evaluated under each flood cases.Results show that the effect of DEM uncertainty on TOPMODEL is inconsiderable and could be ignored in the model’s application.This can be explained by:1) TOPMODEL is not sensitive to the distribution of topographic index and distance to the drainage outlet;2) the distri-bution of topographic index and distance to the drainage outlet are slightly affected by DEM uncertainty.

3D Simulation of Storm Surge Disaster Based on Scenario Analysis

The occurrence of storm surge disaster is often accompanied with floodplain, overflow, dike breach and other complex phenomena, while current studies on storm surge flooding are more concentrated on the 1D/2D numerical simulation of single disaster scenario(floodplain, overflow or dike breach), ignoring the composite effects of various phenomena. Therefore, considering the uncertainty in the disaster process of storm surge, scenario analysis was firstly proposed to identify the composite disaster scenario including multiple phenomena by analyzing key driving forces, building scenario matrix and deducing situation logic. Secondly, by combining the advantages of k-ω and k-ε models in the wall treatment, a shear stress transmission k-ω model coupled with VOF was proposed to simulate the 3D flood routing for storm surge disaster. Thirdly, risk degree was introduced to make the risk analysis of storm surge disaster. Finally, based on the scenario analysis, four scenarios with different storm surge intensity(100-year and 200-year frequency) were identified in Tianjin Binhai New Area. Then, 3D numerical simulation and risk map were made for the case.

Regional Climate Change and Uncertainty Analysis based on Four Regional Climate Model Simulations over China

Four sets of climate change simulations at grid spacing of 50 km were conducted over East Asia with two regional climate models driven at the lateral bounda- ries by two global models for the period 1981-2050. The locus of the study was on the ensemble projection of cli- mate change in the mid-21st century （2031-50） over China. Validation of each simulation and the ensemble average showed good performances of the models overall, as well as advantages of the ensemble in reproducing present day （1981 2000） December-February （DJF）, June-August （JJA）, and annual （ANN） mean temperature and precipitation. Significant wanning was projected for the mid-21st century, with larger values of temperature increase found in the northern part of China and in the cold seasons. The ensemble average changes of precipitation in DJF, JJA, and ANN were determined, and the uncertainties of the projected changes analyzed based on the consistencies of the simulations. It was concluded that the largest uncertainties in precipitation projection are in eastern China during the summer season （monsoon pre-cipitation）.

基于改进Newmark模型的同震滑坡易发性研究

Newmark模型被广泛应用于同震滑坡快速评估,然而传统模型受限于参数(如黏聚力、内摩擦角)的不确定性.本文从概率建模的角度,基于蒙特卡罗方法提出改进Newmark模型,该模型从特定分布(如高斯分布)中抽取随机样本以解释参数不确定性.以2017年九寨沟地震为例阐述了改进Newmark模型在同震滑坡易发性快速评估中的应用,并与不同采样策略下3种机器学习模型相对比.结果表明:改进Newmark模型极高易发区面积占比为5%,包含滑坡数量占比为59.6%,频率比值为128.4,不同采样策略下模型平均AUC值为0.70,表明改进Newmark模型在同震滑坡易发性评估中取得较好效果.本文所提出的模型由特定参数驱动,与数据驱动方法的区别在于不依赖历史滑坡数据,其成果可应用于震后滑坡快速评估,或在给定地震情境下模拟区域滑坡失稳概率以服务于工程早期选址.

空间分析在土壤重金属污染研究中的应用

空间分析技术是土壤重金属污染研究中的一个重要工具。与经典数据统计分析方法学相比,空间分析考虑了数据的空间位置和空间关联性,能挖掘土壤重金属在二维甚至三维空间上的空间分异和时空变异特征,并对这些结果进行可视化使之以更加直观的形式呈现出来。本文概述了探索性空间数据分析(ESDA)、空间回归分析、地统计学及空间插值分析、空间随机模拟等主要空间分析理论和方法。在此基础上,综述了空间分析在土壤重金属污染调查采样设计、空间统计分析、空间变异分析及制图、土壤重金属污染来源解析及污染不确定性研究等方面的应用和研究动态,介绍了一些应用于土壤重金属污染研究的空间分析新方法,并就空间分析方法在土壤重金属研究中的应用前景和发展趋势做了展望。

Projections of land use changes under the plant functional type classification in different SSP-RCP scenarios in China

Land use projections are crucial for climate models to forecast the impacts of land use changes on the Earth’s system.However,the spatial resolution of existing global land use projections(e.g.,0.25°×0.25°in the Land-Use Harmonization(LUH2)datasets)is still too coarse to drive regional climate models and assess mitigation effectiveness at regional and local scales.To generate a high-resolution land use product with the newest integrated scenarios of the shared socioeconomic pathways and the representative concentration pathways(SSPs-RCPs)for various regional climate studies in China,here we first conduct land use simulations with a newly developed Future Land Uses Simulation(FLUS)model based on the trajectories of land use demands extracted from the LUH2 datasets.On this basis,a new set of land use projections under the plant functional type(PFT)classification,with a temporal resolution of 5 years and a spatial resolution of 5 km,in eight SSP-RCP scenarios from 2015 to 2100 in China is produced.The results show that differences in land use dynamics under different SSP-RCP scenarios are jointly affected by global assumptions and national policies.Furthermore,with improved spatial resolution,the data produced in this study can sufficiently describe the details of land use distribution and better capture the spatial heterogeneity of different land use types at the regional scale.We highlight that these new land use projections at the PFT level have a strong potential for reducing uncertainty in the simulation of regional climate models with finer spatial resolutions.

Parametric sensitivity analysis of precipitation and temperature based on multi-uncertainty quantification methods in the Weather Research and Forecasting model

Sensitivity analysis(SA) has been widely used to screen out a small number of sensitive parameters for model outputs from all adjustable parameters in weather and climate models, helping to improve model predictions by tuning the parameters. However, most parametric SA studies have focused on a single SA method and a single model output evaluation function, which makes the screened sensitive parameters less comprehensive. In addition, qualitative SA methods are often used because simulations using complex weather and climate models are time-consuming. Unlike previous SA studies, this research has systematically evaluated the sensitivity of parameters that affect precipitation and temperature simulations in the Weather Research and Forecasting(WRF) model using both qualitative and quantitative global SA methods. In the SA studies, multiple model output evaluation functions were used to conduct various SA experiments for precipitation and temperature. The results showed that five parameters(P3, P5, P7, P10, and P16) had the greatest effect on precipitation simulation results and that two parameters(P7 and P10) had the greatest effect for temperature. Using quantitative SA, the two-way interactive effect between P7 and P10 was also found to be important, especially for precipitation. The microphysics scheme had more sensitive parameters for precipitation, and P10(the multiplier for saturated soil water content) was the most sensitive parameter for both precipitation and temperature. From the ensemble simulations, preliminary results indicated that the precipitation and temperature simulation accuracies could be improved by tuning the respective sensitive parameter values, especially for simulations of moderate and heavy rain.

Nuclear uncertainties in the s-process simulation

The heavy elements in the Universe are formed during the s- and r-processes mainly in AGB stars and supernovae, respectively. Simulation of s- and r-nucleosynthesis critically depends on the neutron capture and weak decay rates for all the nuclei on the reaction chain. The present work analyzes systematically the neutron capture rates (cross sections) for the s-process nuclei, including ～3000 rates on ～200 nuclei. The network calculations for the constant temperature s-process have been performed using the different data sets selected as the nuclear inputs to investigate the uncertainties in the predicted s-abundances. We show that the available cross sections of neutron capture on many s-process nuclei still carry large uncertainties, which lead to low accuracy in the determination of s-process isotope abundances. We analyze the neutron capture cross section data for the same unique isobar nucleus accorded by year from previous work. Such an analysis indicates that the s-process has been studied for more than fifty years and there exist two research stages around 1976 and 2002, respectively. The needs and opportunities for future experiments and theoretical tools are highlighted to remove the existing shortcomings in the neutron capture rates.

Geostatistics for Spatial Uncertainty Characterization

Most geospatial phenomena can be interpreted probabilistically because we are ignorant of the biophysical proc- esses and mechanisms that have jointly created and observed events. This philosophy is important because we are certain about the phenomenon under study at sampled locations, except for measurement errors, but, in between the sampled, we become uncertain about how the phenomenon behaves. Geostatistical uncertainty characterization is to generate random numbers in such a way that they simulate the outcomes of the random processes that created the existing sample data. This set of existing sample is viewed as a partially sampled realization of that random function model. The random function＇s spa- tial variability is described by a variogram or covariance model. The realized surfaces need to honour sample data at their lo- cations, and reflect the spatial structure quantified by the variogram models. They should each reproduce the sample histo- gram representative of the whole sampling area. This paper will review the fundamentals in stochastic simulation by covering univariate and indicator techniques in the hope that their applications in geospatial information science will be wide-spread and fruitful.