台风麦莎渤海转向的可预报性研究

利用MM5模式研究了2005年第9号台风"麦莎"在渤海向东北转向路径的可预报性.试验用不同的积云参数化方案、不同的预报时效分别从确定性和集合预报角度对"麦莎"在渤海向东北方向的转向过程做了模拟.结果表明,"麦莎"在渤海的转向可预报时效为48小时左右.不同的积云参数化方案对台风麦莎路径的48小时预报结果显示台风均转向东北,预示"麦莎 "基本不会直接影响北京.60和72小时的预报结果显示,Kuo和Betts-Miller积云参数化方案的台风模拟路径与实况比较接近,而Grell和KeinFritch积云参数化方案的台风模拟路径却偏向了实际台风位置的西北,台风有可能直接影响北京.研究表明,对于台风麦莎而言,时效超过两天的转向预报可信度较低,Kuo和Betts-Miller积云参数化方案的预报准确性较高.     

长江流域夏季降水的贝叶斯多模式集合概率预测

A Bayesian probabilistic prediction scheme of the Yangtze River Valley (YRV) summer rainfall is proposed to combine forecast information from multi-model ensemble dataset provided by ENSEMBLES project.Due to the low forecast skill of rainfall in dynamic models,the time series of regressed YRV summer rainfall are selected as ensemble members in the new scheme,instead of commonly-used YRV summer rainfall simulated by models.Each time series of regressed YRV summer rainfall is derived from a simple linear regression.The predictor in each simple linear regression is the skillfully simulated circulation or surface temperature factor which is highly linear with the observed YRV summer rainfall in the training set.The high correlation between the ensemble mean of these regressed YRV summer rainfall and observation benefit extracting more sample information from the ensemble system.The results show that the cross-validated skill of the new scheme over the period of 1960 to 2002 is much higher than equally-weighted ensemble,multiple linear regression,and Bayesian ensemble with simulated YRV summer rainfall as ensemble members.In addition,the new scheme is also more skillful than reference forecasts (random forecast at a 0.01 significance level for ensemble mean and climatology forecast for probability density function).     

联合卷积神经网络与集成学习的遥感影像场景分类

针对人工设计的中、低层特征难以实现复杂场景影像的高精度分类以及卷积神经网络依赖大量训练数据等问题，结合迁移学习与集成学习，提出了一种联合卷积神经网络与集成学习的遥感影像场景分类算法。首先基于迁移学习的思想，利用在自然影像数据集上训练好的多个深层卷积神经网络模型作为特征提取器，提取图像多个高度抽象的语义特征；然后构建由Logistic回归和支持向量机组成的Stacking集成模型，对同一图像的多个特征分别训练Logistic模型，将预测概率结果融合构建概率特征；最后利用支持向量机对概率特征训练和预测，得到场景影像的分类结果。利用UCMerced_LandUse和NWPU-RESISC 45两种不同规模的遥感影像数据集进行试验，即使在只有10%的数据作为训练样本情况下，本文方法能够分别达到90.74%和87.21%的分类精度。     

Noah-MP模型中积雪模拟对参数化方案的敏感性评估

针对Noah-MP模型多参数化方案、模拟结果不确定性范围难以确定的特点,选取北疆地区具有代表性的阿勒泰站气象资料作为模型驱动数据,探讨了积雪对多参数化方案的敏感性。在不考虑模型参数和驱动数据不确定性的条件下,设计了集合数为13824的多参数化方案集合模拟试验。选用Natural selection方法对物理过程的敏感性进行分析,并在敏感性分析结果的基础上进一步讨论了模拟结果的不确定性。结果表明:积雪对地表热交换、雨雪分离、土壤温度底层边界条件和第一层积雪或土壤时间方案4个物理过程敏感;在不考虑驱动数据和模型参数不确定性的条件下,多参数化方案集合模拟试验中的不确定性主要来源于敏感物理过程。去除敏感物理过程中能够明显降低模拟性能的参数化方案后,集合模拟结果的不确定性大幅减小。最后,根据分析结果构建了该站雪深和雪水当量模拟的最优参数化方案组合。     

基于集成学习的阶跃型滑坡阶跃点判别分析

针对阶跃型滑坡阶跃点识别和预测难的问题,提出了一种基于聚类分析和集成学习的阶跃型滑坡阶跃点识别和判别模型。以三峡库区八字门滑坡ZG110钻孔2010年4月至2016年12月80个滑坡位移、库水位和降雨数据为例,通过聚类分析方法识别滑坡累积位移-时间曲线中的阶跃点和平稳点,并利用K均值聚类分析检验分类结果的准确性。基于灰色关联确定了滑坡位移的最佳诱发因素,结合随机森林模型建立阶跃型滑坡阶跃点判别模型并利用八字门滑坡ZG111钻孔验证该模型的准确性。模型阶跃点和平稳点的识别准确率均达90%以上,表明该方法在阶跃型滑坡识别中具有较好的适用性,可为阶跃型滑坡的预测提供参考。     

利用综合评分方法评估CMIP5模式对北极海冰的模拟能力

北极海冰冰盖自20世纪以来经历了前所未有的缩减，这使得北极海冰异常对大气环流的反馈作用日益显现。尽管目前的气候模式模拟北极海冰均为减少的趋势，但各模式间仍然存在较大的分散性。为了评估模式对于北极海冰变化及其气候效应的模拟能力，我们将海冰线性趋势和年际异常两者结合起来构造了一种合理的衡量指标。我们还强调巴伦支与卡拉海的重要性，因为前人研究证明此区域海冰异常是近年来影响大尺度大气环流变异的关键因子。根据我们设定的标准，CMIP5模式对海冰的模拟可被归为三种类型。这三组多模式集合平均之间存在巨大的差异，验证了这种分组方法的合理性。此外，我们还进一步探讨了造成模式海冰模拟能力差别的潜在物理因子。结果表明模式所采用的臭氧资料集对海冰模拟能力有显著的影响。     

An ensemble approach for the analysis of extreme rainfall under climate change in Naples (Italy)

In the present paper, an ensemble approach is proposed to estimate possible modifications caused by climate changes in the extreme precipitation regime, with the rain gauge Napoli Servizio Idrografico (Naples, Italy) chosen as test case. The proposed research, focused on the analysis of extremes on the basis of climate model simulations and rainfall observations, is structured in several consecutive steps. In the first step, all the dynamically downscaled EURO‐CORDEX simulations at about 12 km horizontal resolution are collected for the current period 1971–2000 and the future period 2071–2100, for the RCP4.5 and the RCP8.5 concentration scenarios. In the second step, the significance of climate change effects on extreme precipitation is statistically tested by comparing current and future simulated data and bias‐correction is performed by means of a novel approach based on a combination of simple delta change and quantile delta mapping, in compliance with the storm index method. In the third step, two different ensemble models are proposed, accounting for the variabilities given by the use of different climate models and for their hindcast performances. Finally, the ensemble models are used to build novel intensity–duration–frequency curves, and their effects on the early warning system thresholds for the area of interest are evaluated.     

Assimilation of Hourly Surface Observations with the Canadian High-Resolution Ensemble Kalman Filter

An hourly-cycling ensemble Kalman filter (EnKF) working at 2.5?km horizontal grid spacing is implemented over southern Ontario (Canada) to assimilate Meteorological Terminal Aviation Routine Weather Reports (METARs) in addition to the observations assimilated operationally at the Canadian Meteorological Centre. This high-resolution EnKF (HREnKF) system employs ensemble land analyses and perturbed roughness length to prevent an ensemble spread that is too small near the surface. The HREnKF then performs continuously for a four-day period, from which twelve-hour ensemble forecasts are launched every six hours. The impact on analyses and short-term forecasts of assimilating METAR data is given special attention.

It is shown that using ensemble land surface analyses increases near-surface ensemble spreads for temperature and specific humidity. Perturbing roughness length enlarges the spread for surface wind. Given sufficient ensemble spread, the four-day case study shows that the near-surface model state is brought closer to surface observations during the cycling process. The impact of assimilating surface data can also be seen at higher levels by using aircraft reports for verification. The ensemble forecast verification suggests that METAR data assimilation improves ensemble forecasts of air temperature and dewpoint near the surface up to a lead time of six hours or even longer. However, only minor improvement is found in surface wind forecasts.     

基于动力降尺度的区域集合预报初值扰动构建方法研究

利用全球集合预报系统资料(Global Ensemble Forecast System,GEFS),基于WRF中尺度模式构建了区域集合预报系统,区域集合初值的构建采用两种方案,一种是GEFS全球集合预报初值场直接动力降尺度(称为DOWN集合),另一种是提取GEFS全球集合降尺度后的扰动场,并叠加到区域数值预报系统(北京快速更新循环数值预报系统:Beijing Rapid Update Cycle System,BJ-RUC)分析场上构建集合初值场(称为D-RUC集合)。进行了批量试验,通过对比发现D-RUC集合的中小尺度扰动增长优于DOWN集合,而大尺度扰动分量的增长两者相当,说明与高分辨率分析场叠加可以促进动力降尺度扰动的中小尺度扰动分量的增长。集合预报扰动准确性检验结果显示,短预报时效内DOWN集合扰动明显低估了预报误差,在预报误差较大的位置扰动较小,而D-RUC集合能够更好地识别预报场中哪些位置预报误差较大,而哪些位置预报误差较小。集合预报检验结果表明,D-RUC方法能显著改善短时效预报效果,集合离散度有所增加、均方根误差有所减少,概率预报评分显示D-RUC集合比DOWN集合在短预报时效占优。降水个例分析结果表明D-RUC方法能显著改善短时效内的降水概率预报效果。     

基于时空不确定性的对流尺度集合预报效果评估检验

针对对流尺度天气系统的高度非线性特征和高分辨率模式预报结果存在时、空不确定性现象,以及当前邻域概率法主要考虑高分辨率预报结果的空间位移误差,而不能有效解决预报结果存在时间超前与滞后问题,将时间因素引入到邻域概率法中,结合一次强飑线过程进行对流尺度集合预报试验,并基于改进后的新型邻域概率法与分数技巧评分,对降水预报进行了不同时、空尺度的效果评估检验。结果表明:（1）邻域集合概率法和概率匹配平均法在极端降水的分数技巧评分远高于传统集合平均,弥补了集合平均对极端降水预报能力偏低的缺陷。（2）对于此类飑线过程的对流尺度天气系统而言,邻域半径为15—45 km的空间尺度能够改善降水位移误差的空间不确定性,并使其预报效果达到最优,其中15—30 km的邻域半径对于尺度更小的大量级降水事件预报能力更强。（3）对流尺度降水预报考虑时间尺度与降水强度存在着对应关系,不同时间尺度可以捕获到不同量级降水的时间不确定性。同时,时间尺度与空间尺度对于降水预报效果的影响是相互关联的。（4）改进的邻域概率法能够同时体现高分辨率模式预报结果在对流尺度降水事件上存在的时、空不确定性,实现了对流尺度降水在时、空尺度上的综合评估,并能为不同量级降水提供与其时、空尺度相匹配的概率预报结果。