汛期分期变点分析方法的原理及验证

 目前常用的汛期分期方法存在一定的主观性,划分汛期也不够精确,为此提出了用于汛期分期计算的3种变点分析模型：基于日最大值取样的均值变点分析模型、基于超定量取样的概率变点分析模型以及基于年最大值取样的概率变点分析模型。应用统计试验方法对这3种汛期分期模型进行效率评价与比较,结果表明：3种方法都可以客观、精确地进行汛期分期计算,在一定程度上避免了分期中的主观随意性;其中以基于超定量取样的概率变点分析模型具有较好的无偏性和有效性,最适宜在汛期分期计算中应用。     

基于SWMM模型的LID措施年SS总量去除率计算

由于SWMM(storm water management model)模型在进行污染物模拟时并未考虑LID(low impact development)措施的净化作用,致使计算年SS总量去除率时存在较大误差,为弥补该误差,本研究提出了一种基于SWMM模型并考虑LID措施净化作用的年SS总量去除率值计算方法。以深圳市玉田河流域为例,构建SWMM模型,结果表明,年SS总量去除率值不考虑LID措施净化作用时为23. 2%,考虑LID措施净化作用时为30. 1%,后者为前者的1. 3倍。同时论述了年径流总量控制率和年SS总量去除率的概念差异,指出两者之间并无确定比例关系。因为LID主要截留初期中小降雨,在考虑初期雨水冲刷效应的情况下,典型地块年SS总量去除率值可能大于年径流总量控制率值。     

基于小波变换的开都河径流量多时间尺度分析

以开都河大山口站1956—2006年的年径流量资料为例,利用Morlet小波变换对其水文序列的多时间尺度演变特性进行了分析,并预测了开都河2006年以后的年径流量变化趋势。结果表明:开都河存在6 a、10 a、15 a和38 a等4个主周期,其中38 a和10 a分别为第一和第二主周期;2006年后的未来几年内开都河将处于偏枯期。     

邯郸地区年降水量统计特性

运用线性回归分析与Mann-Kendall趋势检验法对邯郸地区及各水资源分区的降雨资料进行分析,揭示了该地区及各水资源分区年降水量不断减少的变化趋势及显著性水平,并得出如下结论:各水资源分区年降水量变化规律不一致;随着海拔高度的降低,区域年降水量变化趋势越明显;低海拔区域年平均降水量减少是海河流域年径流量减少的重要原因。     

基于GM和BP网络的年均流量组合预测模型

以岷江紫坪铺测站1937～2004年的实测流量资料为依据,运用灰色拓扑预测方法,在建模过程中将GM(1,1)模型、无偏GM(1,1)模型以及GM(1,1)改进模型与BP神经网络进行优化组合,建立年均流量组合预测模型.本文采用1995～2004年的年均流量资料对模型进行检验,结果表明,预测相对误差值较为合理,可以作为年均流量预测的有效方法.     

基于LBA-PP模型的年径流丰枯分类

针对年径流丰枯特性同时取决于径流本身大小和年内时程分配的特点,利用一种基于Lévy飞行策略改进的蝙蝠算法(Lévy Bat Algorithm,LBA)搜索投影寻踪模型(Projection Pursuit,PP)最佳投影方向a,提出LBA-PP年径流丰枯分类模型,并构建粒子群优化(Particle Swarm Optimization,PSO)算法-PP模型,与LBA-PP年径流丰枯分类模型对比,以云南省西洋站为例进行实例研究。结果表明LBA算法寻优能力优于PSO算法,具有较高的收敛精度、较好的稳健性能和全局寻优能力。利用LBA算法寻优PP模型最佳投影方向a,不但提高了PP模型的分类精度,而且为PP模型最佳投影方向的选取提供了新的途径和方法。LBA-PP模型同时考虑了年径流大小及年内时程分配信息,其分类结果较常规方法更科学、客观。     

黄河中下游天然年径流变化趋势预测

基于神经网络理论与传统分析方法,采用改进的BP网络算法,利用黄河陕县水文站531 a天然年径流时间序列,建立了黄河中下游年径流长期变化的BP网络预测模型,并以此模型对2001—2050年黄河中下游天然年径流变化趋势进行了预测分析,结果表明:①中下游天然年径流在未来50 a变化的大趋势是丰水时段比枯水时段略占优势,要经历2001—2020年相对丰水时段、2021—2026年相对枯水时段、2027—2037年相对丰水时段及2038—2050年相对枯水时段;②黄河中下游与上游的天然年径流变化的大趋势基本一致,但中下游出现连续枯水时段的时间较上游长。     

Impact assessment of climate change and human activities on annual highest water level of Taihu Lake

The annual highest water level of Taihu Lake (Zm) is very significant for flood management in the Taihu Basin. This paper first describes the inter-annual and intra-annual traits of Zm from 1956 to 2000. Then, using the Mann-Kenall (MK) and Spearman (SP) nonparametric tests, the long-term change trends of area precipitation and pan evaporation in the Taihu Basin are determined. Meanwhile, using the Morlet wavelet transformation, the fluctuation patterns and change points of precipitation and pan evaporation are analyzed. Also, human activities in the Taihu Basin are described, including land use change and hydraulic project construction. Finally, the relationship between Zm, the water level of Taihu Lake 30 days prior to the day of Zm (Z0), and the 30-day total precipitation and pan evaporation prior to the day of Zm (P and E0, respectively) is described based on multi-linear regression equations. The relative influence of climate change and human activities on the change of Zm is quantitatively ascertained. The results demonstrate that: (1) Zm was distinctly higher during the 1980-2000 period than during the 1956-1979 period, and the 30 days prior to the day of Zm are the key phase influencing Zm every year; (2) P increased significantly at a confidence level of 95% during the 1956-2000 period, while the reverse was true for E0; (3) The relationship between Zm, P and E0 distinctly changed after 1980; (4) Climate change and human activities together caused frequent occurrences of high Zm after 1980; (5) Climate change caused a substantially greater Zm difference between the 1956-1979 and 1980-2000 periods than human activities. Climate change, as represented by P and E0, was the dominant factor raising Zm, with a relative influence ratio of 83.6%, while human activities had a smaller influence ratio of 16.4%.     

Monitoring Method for Blast Furnace Wall With Copper Staves

 A monitoring method that has been designed for the first time for blast furnace wall with copper staves manufactured in China was introduced. Combining the method of “inverse problem” and the concept “non inverse problem”, the monitoring program for blast furnace wall with copper staves has been realized, which can be used to calculate online the accretion thickness and temperature of hot surface of copper staves after obtaining the values of thermocouples of copper staves. The accretion state obtained in the actual investigation has proved that the result of the program is correct. The monitoring program shows that the accretion would easily fluctuate when the accretion layer is extremely thick or thin, thereby the stable and smooth operation of the blast furnace is hindered. By maintaining appropriate accretion thickness, both long campaigns and high productivity of the blast furnace can be achieved; furthermore, it can also optimize the operation of blast furnace and maximize its production. Approximately 30-50 mm in thickness of accretion layer is maintained on the wall of Shougang blast furnace 2, which can meet the requirement for obtaining both long campaign and high productivity.