Uncertainty assessment and optimization of hydrological model with the Shuffled Complex Evolution Metropolis algorithm: an application to artificial neural network rainfall-runoff model

Assessment of parameter and predictive uncertainty of hydrologic models is an essential part in the field of hydrology. However, during the past decades, research related to hydrologic model uncertainty is mostly done with conceptual models. As is accepted that uncertainty in model predictions arises from measurement errors associated with the system input and output, from model structural errors and from problems with parameter estimation. Unfortunately, non-conceptual models, such as black-box models, also suffer from these problems. In this paper, we take the artificial neural network (ANN) rainfall-runoff model as an example, and the Shuffled Complex Evolution Metropolis algorithm (SCEM-UA) is employed to analysis the parameter and predictive uncertainty of this model. Furthermore, based on the results of uncertainty assessment, we finally arrive at a simpler incomplete-connection artificial neural network (ICANN) model as well as with better performance compared to original ANN rainfall-runoff model. These results not only indicate that SCEM-UA can be a useful tool for uncertainty analysis of ANN model, but also prove that uncertainty does exist in ANN rainfall-runoff model. Additionally, in some way, it presents that the ICANN model is with smaller uncertainty than the original ANN model.     

On the growth of disturbances to forced and dissipated barotropic flows

Abstract

We consider the growth of disturbances to large-scale zonally-asymmetric steady states in a truncated spectral model for forced and dissipated barotropic flow. A variant of the energy method is developed to optimize the instantaneous disturbance energy growth rate. The method involves solving a matrix eigenvalue problem amenable to standard numerical techniques. Two applications are discussed. (1) The global stability of a family of steady states is assessed in terms of the Ekman damping coefficient r. It is shown that monotonic global stability (i.e., every disturbances energy monotonically decays to zero) prevails when r≥r_c . (2) Initially fastest-growing disturbances are constructed in the r<r_c regime. Particular attention is paid to a subregion of the r<r_c regime where initially-growing disturbances exist despite stability with respect to normal modes. Nonlinear time-dependent simulations are performed in order to appraise the time evolution of various disturbances.     

芦山地震有限断层混合震源模型模拟

建立了基于有限断层的芦山地震的混合震源模型,突出破裂面上滑动量分布中凹凸体的主导作用,同时借助k平方模型表达对震源复杂性认识尚不够深入所带来的随机性.混合震源模型中全局震源参数均值与局部震源参数均值的确定,主要依据统计的半经验标定公式,进一步借助截断正态分布将区域地震、地质构造、地震活动性提供的参数限值结合起来,生成最终的震源参数值.基于上述思路,本文共建立了30组芦山地震混合震源模型,进而以芦山地震8个近场观测台站为试算点,借助地震动反应谱残差评价的方法,从30组模型中选定了表达“平均”特征的混合震源模型.     

Mass Balance Study on Phosphorus Removal in Constructed Wetland Microcosms Treating Polluted River Water

A number of studies have showed that the mass removal rates of phosphorus (P) in different constructed wetlands (CWs) varied significantly, and it is essential to quantify the contributions of major P removal processes in order to improve system design. The objective of this study was to investigate the effects of vegetation, hydraulic retention time (HRT), and water temperature on P removal from polluted river water and to quantify the contributions of different P removal pathways in surface CWs. Results showed that the average total P removal rates ranged between 2.69 and 20.84 mg/(m² day) in different seasons and were influenced significantly by vegetation, HRT, and water temperature. According to the mass balance approach, plant uptake removed 4.81–22.33% of P input, while media storage contributed 36.16–49.66%. Other P removal processes such as microbiota uptake removed around 0.26–4.13%. Media storage and plant uptake were identified as the main P removal processes in surface CWs treating polluted river water. This illustrated the importance of selecting media and plants in CWs for future practical application.     

Accuracy analysis for SST gravity field model in China

Taking China as the region for test the potential of the new satellite gravity technique, satelliteto-satellite tracking for improving the accuracy of regional gravity field model is studied. With WDM94 as reference, the gravity anomaly residuals of three models, the latest two GRACE global gravity field model (EIGEN_GRACE02S, GGM02S) and EGM96, are computed and compared. The causes for the differences among the residuals of the three models are discussed. The comparison between the residuals shows that in the selected region, EIGEN_GRACE02S or GGM02S is better than EGM96 in lower degree part (less than 110 degree). Additionally, through the analysis of the model gravity anomaly residuals, it is found that some systematic errors with periodical properties exist in the higher degree part of EIGEN and GGM models, the results can also be taken as references in the validation of the SST gravity data.     

The assessment of health impact caused by energy use in urban areas of China: an intake fraction–based analysis

The Indian Ocean Tsunami of December 2004 caused inundation of seawater along the Northern coast of Tamil Nadu, India, resulting in loss of 8,000 people with extensive damage to properties. The paper describes the inundation of seawater in two northern districts, namely Kancheepuram and Villupuram districts, which showed distinct patterns of inundation of seawater and run-up levels due to variations in geomorphic features. TUNAMI N2 model was used to predict the seawater inundation for earthquakes occurred in 1881 at Car Nicobar, Sumatra 2004 and a worst-case scenario. The coastal areas with beaches having gentle slope showed more inundation compared with coastal areas having varied slope and habited by sand dunes and coastal vegetation. Appreciable inundation of seawater with tsunami simulated for 1881 Car Nicobar indicated that proximity to the source plays a major role besides earthquake parameters in causing inundation. The worst-case scenario generated from subduction zone of Car Nicobar using Sumatra 2004 earthquake parameters revealed extreme vulnerability of coasts of both the districts to giant tsunamis.     

Risk evaluation of China’s natural disaster systems: an approach based on triangular fuzzy numbers and stochastic simulation

To deal effectively with the evaluation problem of natural disaster risk system affected by many uncertain factors, a multivariate connection number expression is presented. This expression is based on the index samples and evaluation grade criterions of natural disaster risk system and is capable of describing the hierarchy property and fuzziness of membership relationship between index samples and evaluation grade criterions. In this proposed method, the fuzzy evaluation grade criterion problem is resolved by combining triangular fuzzy numbers with multivariate connection number theory, and triangular fuzzy numbers are used to express the discrepancy degree coefficients of connection number and evaluation index weights. Accordingly, a connection number-based evaluation method for the natural disaster system of China (named CN-TFN for short) is established using triangular fuzzy numbers and stochastic simulation. The application results show that the spatial distribution of natural disaster risk grades of China has the trend of aggrandizement from west to east of China. The economically developed and densely populated coastal areas are very likely to have a high level of natural disaster risk grade or above; thus, these areas are the key regions of the natural disaster risk management of China. The results also show that the CN-TFN is able to reflect practical conditions of the evaluation problem of natural disaster system and to provide more reliability information as compared to the existing evaluation methods. This is as a result of its comprehensive usage of various information of subjective and objective uncertainties in the evaluation process of natural disaster risk system and its expression by confidence intervals. Due to the simplicity and generalization, the CN-TFM is applicable to comprehensive risk grade evaluation of various natural disaster systems.     

极化散射特性保持的改进迭代Wishart分类算法

为克服基于极化散射特性保持的迭代Wishart分类算法不适用于城区及对混合散射像素分类欠理想等不足,本文提出一种改进方法.其基本思想是先应用四分量分解算法将像素分成4种基本散射类型和混合散射类型,接着以平均合并度为指导对基本散射类型中的像素自适应聚类,最后对所有像素进行散射特性保持的迭代Wishart分类.试验结果表明...     

Spatial Distribution and Temporal Variation of the Winter Wheat Late Frost Disaster in Henan, China

The winter wheat late frost disaster(WFD) occurs mainly in the Yellow and Huaihe River area,of which Henan Province covers the most part.Henan is the major area of wheat production in China,but it is severely hit by the WFD.In this study,we construct a WFD index based on the minimum temperature and the winter wheat development period(WDP).The WFD degrees and days at 30 agrometeorological stations in Henan Province during the period of 1981-2004 are calculated.For the large-scale temporal variation analysis of WFD,the 24-yr WDP observation series is relatively short,so it is expanded by using the relation between the turning green date of winter wheat and the 5-day running mean temperature and that between the stem elongation phase and the effective cumulative temperature above a critical value of 2.5 ℃.The WFD data are also expanded for the last 50 years and are analyzed by using the empirical orthogonal function(EOF) and the Morlet wavelet methods.Characteristics in the spatial distribution and temporal variation of WFD are revealed.The results show that the frequency of WFD is generally high,exceeding 40% in parts of Henan,and exhibits a rising trend in the period of 1970-1990.The variation trend of WFD degrees is similar to that of WFD days,and the areas with higher WFD degrees coincide the areas with more WFD days.Moreover,the WFD degree has a greater impact on the winter wheat yield than the WFD days.The areas with high WFD degrees lie in the southeast and southwest of Henan,and the areas with low WFD degrees lie in the south of the Huaihe River and parts of western Henan.Temporal variations of the first and second EOF modes of the WFD degree display 16-and quasi-22-yr periodicities,respectively.The areas of high(low) WFD frequency are distributed in the northern Henan and the southwest border of Henan(the northeast Henan and the middle part of southwest Henan).The temporal variation of the first(second) EOF mode of WFD days exhibits a periodicity(periodicities) of quasi-4 yr(quasi-3 and quasi-6-7 yr).