基于GLUE法的多指标水质模型参数率定方法

目前对水质模型进行参数率定通常利用计算机算法来实现,但由于水质模型日趋复杂的非线性结构往往会导致"异参同效"现象,无法通过单个似然度判断参数的取值是否能够取得真值.为避免这一情况,本文提出了一套基于GLUE法的多目标模型参数率定方法,并以WASP水质模型的应用为例,通过Sobol方法确定模型的敏感参数,并利用DO、CBOD、氨氮、硝态氮4项指标的似然函数对参数同时进行率定.结果表明,该方法既可以有效地避免因追求"过拟合"而造成模型参数取值不当,也可以减小模型参数的不确定性,为具有"异参同效"现象的复杂模型的参数率定工作提供了一个更为可靠的方法.     

半干旱半湿润地区HSPF模型水文模拟及参数不确定性研究

本研究选取半干旱半湿润地区北京妫水河流域2005—2007年和2008年月径流数据为率定期和验证期,建立HSPF水文模型进行径流模拟,结合人工率定和PEST自动率定程序进行参数优选,并通过GLUE方法分析模型参数不确定性.通过Monte-Carlo随机采样得到30000组参数组合,分析参数与似然值散点图,把参数分为敏感参数(LZSN、AGWRC)、区间敏感参数(BASETP)和不敏感参数(AGWETP、INFILT、CEPSC、DEEPFR、UZSN、INTFW、IRC).针对比较敏感的参数LZSN、AGWRC和BASETP分析其相关性,发现LZSN和AGWRC相关性较强.模型存在大量"异参同效"现象,表明影响结果的是参数组合而不是单一参数.进一步计算90%置信度下的不确定性范围,发现不确定性范围与径流大小密切相关,径流愈大其不确定性范围愈大,反之亦然.本文对参数不确定的分析研究可为HSPF模型在区域尺度水文预测等提供参考和依据.     

An analysis of the likely success of policy actions under uncertainty: Recovery from acidification across Great Britain

In the context of wider debates about the role of uncertainty in environmental science and the development of environmental policy, we use a Generalised Likelihood Uncertainty Estimate (GLUE) approach to address the uncertainty in both acid deposition model predictions and in the sensitivity of the soils to assess the likely success of policy actions to reduce acid deposition damage across Great Britain. A subset of 11, 699 acid deposition model runs that adequately represented observed deposition data were used to provide acid deposition distributions for 2005 and 2020, following a substantial reduction in SO₂ and NO_x emissions. Uncertain critical loads data for soils were then combined with these deposition data to derive estimates of the accumulated exceedance (AE) of critical loads for 2005 and 2020. For the more sensitive soils, the differences in accumulated exceedance between 2005 and 2020 were such that we could be sure that they were significant and a meaningful environmental improvement would result. For the least sensitive soils, critical loads were largely met by 2020, hence uncertainties in the differences in accumulated exceedance were of little policy relevance. Our approach of combining estimates of uncertainty in both a pollution model and an effects model, shows that even taking these combined uncertainties into account, policy-makers can be sure that the substantial planned reduction in acidic emissions will reduce critical loads exceedances. The use of accumulated exceedance as a relative measure of environmental protection provides additional information to policy makers in tackling this ‘wicked problem’.     

优化、RSA和GLUE方法在非线性环境模型参数识别中的比较

参数识别是数学模型应用的前提,本文对非线性环境模型常用的3种参数识别方法进行了比较分析.优化方法是出现最早、应用最广泛的参数识别方法之一.但在观测误差存在的情况下,采用优化方法识别的最优参数进行模型预测,存在很大的决策风险.考虑到这种不足,RSA方法和GLUE方法摒弃了识别单一最优参数的传统思维,而把识别参数扩大到多点组成的参数集.RSA方法与GLUE方法不同的是,RSA方法把可行的参数点看成是同等接受的,而GLUE方法则根据模拟值与实测值的差别,确定其似然度,代表参数的可信度水平.除参数识别以外,RSA方法和GLUE方法也是全局灵敏度分析的重要工具.     

Integrating soil carbon cycling with that of nitrogen and phosphorus in the watershed model SWAT: Theory and model testing

In this paper we describe and test a sub-model that integrates the cycling of carbon (C), nitrogen (N) and phosphorus (P) in the Soil Water Assessment Tool (SWAT) watershed model. The core of the sub-model is a multi-layer, one-pool soil organic carbon (S_C) algorithm, in which the decomposition rate of S_C and input rate to S_C (through decomposition and humification of residues) depend on the current size of S_C. The organic N and P fluxes are coupled to that of C and depend on the available mineral N and P, and the C:N and N:P ratios of the decomposing pools. Tillage explicitly affects the soil organic matter turnover rate through tool-specific coefficients. Unlike most models, the turnover of soil organic matter does not follow first order kinetics. Each soil layer has a specific maximum capacity to accumulate C or C saturation (S_x) that depends on texture and controls the turnover rate. It is shown in an analytical solution that S_x is a parameter with major influence in the model C dynamics. Testing with a 65-yr data set from the dryland wheat growing region in Oregon shows that the model adequately simulates the S_C dynamics in the topsoil (top 0.3 m) for three different treatments. Three key model parameters, the optimal decomposition and humification rates and a factor controlling the effect of soil moisture and temperature on the decomposition rate, showed low uncertainty as determined by generalized likelihood uncertainty estimation. Nonetheless, the parameter set that provided accurate simulations in the topsoil tended to overestimate S_C in the subsoil, suggesting that a mechanism that expresses at depth might not be represented in the current sub-model structure. The explicit integration of C, N, and P fluxes allows for a more cohesive simulation of nutrient cycling in the SWAT model. The sub-model has to be tested in forestland and rangeland in addition to agricultural land, and in diverse soils with extreme properties such high or low pH, an organic horizon, or volcanic soils.     

Multiple sources of predictive uncertainty in modeled estimates of net ecosystem CO2 exchange

Net ecosystem CO₂ exchange (NEE) is typically measured directly by eddy covariance towers or is estimated by ecosystem process models, yet comparisons between the data obtained by these two methods can show poor correspondence. There are three potential explanations for this discrepancy. First, estimates of NEE as measured by the eddy-covariance technique are laden with uncertainty and can potentially provide a poor baseline for models to be tested against. Second, there could be fundamental problems in model structure that prevent an accurate simulation of NEE. Third, ecosystem process models are dependent on ecophysiological parameter sets derived from field measurements in which a single parameter for a given species can vary considerably. The latter problem suggests that with such broad variation among multiple inputs, any ecosystem modeling scheme must account for the possibility that many combinations of apparently feasible parameter values might not allow the model to emulate the observed NEE dynamics of a terrestrial ecosystem, as well as the possibility that there may be many parameter sets within a particular model structure that can successfully reproduce the observed data. We examined the extent to which these three issues influence estimates of NEE in a widely used ecosystem process model, Biome-BGC, by adapting the generalized likelihood uncertainty estimation (GLUE) methodology. This procedure involved 400,000 model runs, each with randomly generated parameter values from a uniform distribution based on published parameter ranges, resulting in estimates of NEE that were compared to daily NEE data from young and mature Ponderosa pine stands at Metolius, Oregon. Of the 400,000 simulations run with different parameter sets for each age class (800,000 total), over 99% of the simulations underestimated the magnitude of net ecosystem CO₂ exchange, with only 4.07% and 0.045% of all simulations providing satisfactory simulations of the field data for the young and mature stands, even when uncertainties in eddy-covariance measurements are accounted for. Results indicate fundamental shortcomings in the ability of this model to produce realistic carbon flux data over the course of forest development, and we suspect that much of the mismatch derives from an inability to realistically model ecosystem respiration. However, difficulties in estimating historic climate data are also a cause for model-data mismatch, particularly in a highly ecotonal region such as central Oregon. This latter difficulty may be less prevalent in other ecosystems, but it nonetheless highlights a challenge in trying to develop a dynamic representation of the terrestrial biosphere.     

似然度函数对GLUE方法的影响

GLUE方法是不确定性条件下参数识别的重要方法,似然度函数的选择是GLUE方法的关键. 为克服观测误差和模型结构误差对研究结果的影响,利用合成的数据序列,以污染物衰减模型为基础,分析了似然度函数对参数识别和模型预测结果的影响. 从识别结果来看,由于似然度函数的变化改变了参数似然度之间的对比,因此对参数识别及灵敏度分析的结果都产生了影响,甚至能够改变参数全局灵敏度的相对排序. 从预测结果来看,似然度函数的变化影响了模型预测的分布,但即使预测结果的模糊性在减少,精度并没有相应地提高. 因此,似然度函数的恰当选择对分析结果具有重要的影响,应结合研究问题的具体特点和对模拟结果的要求,采用概率论等相关方法选择出尽可能体现参数真实重要性的函数.      

大型浅水湖泊磷模型参数不确定性及敏感性分析

以太湖作为大型浅水湖泊的代表,以EFDC模型为基础建立关于磷的沉积成岩模型,运用拉丁超立方抽样方法、普适似然不确定性分析方法和区域敏感性分析方法,对成岩模块中与磷迁移转化有关的16个参数进行敏感性分析.结果表明,太湖全湖区磷酸盐和总磷的时间空间分布都极不均匀,成岩模块参数的不确定性在磷酸盐和总磷的模拟方面都存在强烈影响.以上覆水磷酸盐为输出目标,"沉积物-水"界面遭受的风浪扰动在水深相对更浅的湖区更为直接,湖湾区水的流通性较差,利于沉积物中溶解态磷释放.以上覆水总磷为输出目标,湖心区颗粒态磷和溶解态有机磷的质量浓度受成岩模块参数不确定性的影响较大.敏感参数主要是动力特性与和氧相关的两类参数.对于大型浅水湖泊,底泥沉积成岩模块敏感参数的重要性不弱于水动力、水质模块,针对不同水质、底泥分布的水域,模拟磷元素时应注重敏感参数的取值率定.     

Testing a new model of aphid abundance with sedentary and non-sedentary predators

Aphid population dynamics has been thoroughly investigated, especially in tree-dwelling aphids. Among the controls of the aphid rate of increase are the negative effects of antagonists, the positive effects of mutualists, the density-dependence of the aphid dynamics, and the non-stationary quality of plant tissues. Here we present a mechanistic model of aphid growth that considers most of these governing factors using a simple formulation. What is new in this model is that it considers two kinds of antagonists. The first kind is a guild of aphid predator specialists that includes ladybirds (Coleoptera: Coccinellidae), but also species of some families of Hemiptera, Diptera, and Neuroptera. The second kind of antagonists consists of omnivores or generalist predators and in this particular setting is exemplified by the European earwig Forficula auricularia (Dermaptera: Forficulidae). The model developed here compared the effects of these two different kinds of aphid predators, the second one always at the site (sedentary predators) and the first one that arrives in important numbers only once the aphid population has already developed to some degree (non-sedentary predators). Multiple model parameter sets, representing different hypotheses about controls on aphid populations, were evaluated within the Generalised Likelihood Uncertainty Estimation (GLUE) methodology. The model correctly reproduced the experimental data obtained in an organic citrus grove showing the important effect that sedentary predators as earwigs can have on the aphid populations. Low densities of sedentary predators or even low predation rates can have a disproportionate effect on the final aphid density, as they prey on small populations, when the per capita effect on the aphid population is higher. During the main spring peak of aphids the role of non-sedentary predators is secondary, as they track the aphid density rather than control it. However, these non-sedentary predators are important within the proposed model to keep the second autumn peak of aphids at low values.     

Modelling soil carbon development in Swedish coniferous forest soils—An uncertainty analysis of parameters and model estimates using the GLUE method

Boreal forest soils such as those in Sweden contain a large active carbon stock. Hence, a relatively small change in this stock can have a major impact on the Swedish national CO₂ balance. Understanding of the uncertainties in the estimations of soil carbon pools is critical for accurately assessing changes in carbon stocks in the national reports to UNFCCC and the Kyoto Protocol. Our objective was to analyse the parameter uncertainties of simulated estimates of the soil organic carbon (SOC) development between 1994 and 2002 in Swedish coniferous forests with the Q model. Both the sensitivity of model parameters and the uncertainties in simulations were assessed. Data of forests with Norway spruce, Scots pine and Lodgepole pine, from the Swedish Forest Soil Inventory (SFSI) were used. Data of 12 Swedish counties were used to calibrate parameter settings; and data from another 11 counties to validate. The “limits of acceptability” within GLUE were set at the 95% confidence interval for the annual, mean measured SOC at county scale. The calibration procedure reduced the parameter uncertainties and reshaped the distributions of the parameters county-specific. The average measured and simulated SOC amounts varied from 60 t C ha⁻¹ in northern to 140 t C ha⁻¹ in the southern Sweden. The calibrated model simulated the soil carbon pool within the limits of acceptability for all calibration counties except for one county during one year. The efficiency of the calibrated model varied strongly; for five out of 12 counties the model estimates agreed well with measurements, for two counties agreement was moderate and for five counties the agreement was poor. The lack of agreement can be explained with the high inter-annual variability of the down-scaled measured SOC estimates and changes in forest areas over time. We conclude that, although we succeed in reducing the uncertainty in the model estimates, calibrating of a regional scale process-oriented model using a national scale dataset is a sensitive balance between introducing and reducing uncertainties. Parameter distributions showed to be scale sensitive and county specific. Further analysis of uncertainties in the methods used for reporting SOC changes to the UNFCCC and Kyoto protocol is recommended.