Statistical Downscaling Prediction of Summer Precipitation in Southeastern China

A statistical downscaling approach based on multiple-linear-regression (MLR) for the prediction of summer precipitation anomaly in southeastern China was established, which was based on the outputs of seven operational dynamical models of Development of a European Multi-model Ensemble System for Seasonal to Interannual Prediction (DEMETER) and observed data. It was found that the anomaly correlation coefficients (ACCs) spatial pattern of June-July-August (JJA) precipitation over southeastern China between the seven models and the observation were increased significantly; especially in the central and the northeastern areas, the ACCs were all larger than 0.42 (above 95% level) and 0.53 (above 99% level). Meanwhile, the root-mean-square errors (RMSE) were reduced in each model along with the multi-model ensemble (MME) for some of the stations in the northeastern area; additionally, the value of RMSE difference between before and after downscaling at some stations were larger than 1 mm d-1. Regionally averaged JJA rainfall anomaly temporal series of the downscaling scheme can capture the main characteristics of observation, while the correlation coefficients (CCs) between the temporal variations of the observation and downscaling results varied from 0.52 to 0.69 with corresponding variations from -0.27 to 0.22 for CCs between the observation and outputs of the models.     

西北偏湿背景下青海省夏季降水的统计降尺度方法研究

本文研制建立了一个预测青海省夏季降水的动力—统计相结合的组合降尺度预测方法（Hybrid Statistical Downscaling Prediction，HSDP），该方法综合利用了气候模式Climate Forecast System 2.0版本（CFSv2）实时预测的高可预报性环流信息及前期观测的与青海夏季降水具有高相关性的气候因子，采用年际增量方法，基于气候变量的年际增量规律建立统计模型，从而实现对青海夏季降水进行动力—统计相结合的气候预测。根据全球气候因子的年际增量与青海省夏季降水年际增量的相关系数，以及CFSv2预测产品对实况模拟能力的评估，选取以下关键区气候变量的年际增量作为预测因子：（1） CFSv2模式预测当年夏季包含贝加尔湖脊、乌拉尔山脊和新疆脊区域的500 hPa高度场；（2） CFSv2模式预测青藏高原以西200 hPa纬向风场；（3）观测资料中前1 a秋、冬季热带太平洋地区海表面温度场；（4）观测资料中前1 a秋、冬季西伯利亚地区的海平面气压场，对青海省夏季降水进行统计降尺度预测。统计降尺度模型利用1983—2011年进行建模，回报2012—2018年夏季青海省降水的空间分布和时间变化，并对该模型对1983—2011年的夏季青海省降水的回报能力进行了交叉检验。回报结果表明该统计降尺度模型对CFSv2的青海省夏季降水预测能力有显著的提高，能够很好地再现青海省夏季降水西北部的高原地区偏少，而在东南部偏多的特点。该模型预测所得2012—2018年夏季青海省降水的时间变化也与实况有着较高的相关系数（0.76），对于降水显著偏少的年份（如2015年）和显著偏多的年份（如2012、2018年）的降水预测都有很好的表现。对于建模时段的交叉检验结果（相关系数为0.46，比模型回报结果与实况的相关系数0.48略低）表明，该模型具有较高的稳定性和可靠性。     

长江流域年降水预测动力-统计降尺度方法及其应用

基于站点资料、再分析数据和动力气候模式回报数据,利用经验正交函数分解（EOF,Empirical Orthogonal Function）迭代和年际增量方法,探讨了长江流域年尺度降水异常的动力-统计降尺度预测方法及其应用效果。结果表明,基于再分析数据的年尺度环流场,建立的长江流域年尺度降水异常增量的统计降尺度预测方案,其26 a回报检验的距平相关系数（ACC）平均达0.6,证明该方案具有较高的可预报性。进一步利用模式预测的年尺度环流场,建立了年降水异常增量的动力-统计降尺度预测方案,其ACC平均为0.42,显示了较高的回报技巧,远优于模式直接输出的年降水动力预报结果。通过分析调制年降水预报技巧高低的因素发现,赤道中东太平洋年平均海温距平为负值时,预报技巧更高,ACC平均达0.5以上。在拉尼娜发展年或拉尼娜持续年的冷水背景下,利用EOF迭代选取的特征向量偏多时,多尺度的大气环流信息被纳入预测模型中作为预测信号,预测技巧得到了提高。     

基于CFS模式的中国站点夏季降水统计降尺度预测

本研究针对中国夏季站点降水,研制建立了基于Climate Forecast System(CFS)实时预测数值产品及观测资料的统计降尺度预测系统。此预测系统选取了CFS模式中当年夏季500 hPa高度场和观测资料中前一年秋、冬季海表面温度场作为预测因子,两因子的关键区分别为泛东亚地区和热带太平洋地区。统计降尺度模型对1982~2011年中国夏季降水的回报效果较CFS模式原始结果显著提高,空间距平相关系数由0.03提高到0.31,时间相关系数在中国大部分地区显著提高,最大可达0.6。均方根误差较CFS模式原始结果明显降低,同时,此降尺度模型较好的回报出2011年汛期降水的距平百分率的空间分布型。     

基于海气耦合模式的山西省夏季降水统计降尺度预测研究

基于中国气象局国国家气候中心海气耦合模式(CGCM/NCC)预测产品和山西省50站夏季降水资料,利用典型因子回归的方法(CCA),建立了山西省夏季降水的统计降尺度预测模型。该预测模型选取了CGCM/NCC模式夏季500 h Pa高度场和海平面气压作为预测因子,分别选取了长江中下游地区和热带中东太平洋作为预报关键区。统计降尺度模型对2007~2014年山西省夏季降水的回算较模式原始结果有显著提高,除2008年外,空间距平相似系数(ACC)均通过了0.01的显著性检验,时间相关系数(TCC)在山西省大部分地区都有显著提高,最大可达0.6,降水预测(PS)评分在70分以上。检验结果显示,基于CCA降尺度方法建立的预测模型对山西省夏季降水模态预测的准确率较高且比较稳定,其预测效果远高于CGCM/NCC直接输出降水结果。     

Prediction of spring precipitation in China using a downscaling approach

The aim of this paper is to use a statistical downscaling model to predict spring precipitation over China based on a large-scale circulation simulation using Development of a European Multi-model Ensemble System for Seasonal to Interannual Prediction (DEMETER) General Circulation Models (GCMs) from 1960 to 2001. A singular value decomposition regression analysis was performed to establish the link between the spring precipitation and the large-scale variables, particularly for the geopotential height at 500?hPa and the sea-level pressure. The DEMETER GCM predictors were determined on the basis of their agreement with the reanalysis data for specific domains. This downscaling scheme significantly improved the predictability compared with the raw DEMETER GCM output for both the independent hindcast test and the cross-validation test. For the independent hindcast test, multi-year average spatial correlation coefficients (CCs) increased by at least ~30?% compared with the DEMETER GCMs’ precipitation output. In addition, the root mean-square errors (RMSEs) decreased more than 35?% compared with the raw DEMETER GCM output. For the cross-validation test, the spatial CCs increased to greater than 0.9 for most of the individual years, and the temporal CCs increased to greater than 0.3 (95?% confidence level) for most regions in China from 1960 to 2001. The RMSEs decreased significantly compared with the raw output. Furthermore, the preceding predictor, the Arctic Oscillation, increased the predicted skill of the downscaling scheme during the spring of 1963.     

降尺度方法在中国不同区域夏季降水预测中的应用

在中国降水气候分区的基础上,利用降尺度方法进行区域夏季降水预测(RSPP),预测模型建立的基础是寻找影响区域气候的关键因子。降尺度预测模型中使用的资料有国家气候中心海-气耦合模式(CGCM/NCC)回报资料、NCEP/NCAR再分析资料和台站观测资料。为了避免年代际变化特征对季节尺度降水预测的影响,首先对CGCM/NCC模式输出资料、NCEP/NCAR再分析资料、区域平均降水资料去除年代际线性变化趋势,即去除所有预报因子场和预报对象场的长期变化趋势。然后分别计算预报对象和模式资料的预报因子场以及再分析资料的预报因子场的相关系数,把相关系数值同时达到0.05显著性检验水平的区域平均环流特征作为预测因子,保证挑选出的预测因子既能反映实际大气中预测因子与预报对象的关系,同时又是海-气耦合模式预测的高技巧信息。利用最优子集回归作为转换函数的降尺度方法建立区域夏季降水预测模型。交叉检验和独立样本检验结果表明,文中设计的区域夏季降水预测模型对中国大部分地区的夏季降水趋势预测的准确率较高且比较稳定,其预测效果远高于CGCM/NCC直接输出降水结果。进一步对具有较高预测技巧的代表性区域的可预报性来源分析发现,物理意义明确且独立性强的预测因子有助于提高预测准确率。     

基于NCEP CFSv2模式的黑龙江省夏季降水预测研究

基于1982-2017年NCEP_CFSv2（NCEP Climate Forecast System version 2）模式预测资料对黑龙江省夏季降水进行降尺度预测。通过分析黑龙江省夏季降水与同期环流因子的关系、模式对关键区环流因子的预测,选取模式模拟与再分析资料相关较好、黑龙江降水实况与再分析资料关系较好的环流因子作为预测因子,结合最优子集回归法筛选因子,建立降尺度预测模型,最后采用交叉检验法进行预测效果检验和独立样本预测。结果表明：模式降尺度预测与实况的距平符号-致率为69%,6 a独立样本预测中有5 a预测正确,优于目前的业务预测效果。进-步研究发现,在模式能够准确预测环流因子的情况下,模式降尺度可以较好地预测黑龙江省夏季降水的趋势。此外,模式降尺度在拉尼娜年预测效果较好。     

云南夏季降水异常的影响因子及物理统计预测方法

云南夏季降水年际变化较大,影响因子众多,夏季降水的预测较为困难。使用1965—2017年云南省122个气象观测站的逐日降水资料和NCEP大气环流资料,采用年际增量的方法来预测云南夏季降水。文中基于云南夏季降水年际增量变化规律和影响夏季降水的环流形势及物理过程,选取了6个具有物理意义的预测因子,包括:前期2月南太平洋海温异常、前期2月东亚北部海平面气压异常、前期4月北美500hPa位势高度异常、前期5月太平洋北部海平面气压异常、前期1月印度半岛北部500hPa位势高度异常及前期2月澳洲以南地区200hPa高度场偶极子异常,来建立云南夏季降水预测模型。并对预测模型进行逐年交叉检验和1998—2017年逐年独立样本检验。交叉检验中夏季降水年际增量预测值和观测值的相关系数为0.85,相对均方根误差为8.0%。回报检验中夏季降水年际增量的相对均方根误差为9.1%,63.0%的异常年份预测值能够准确地预报出夏季降水异常。该预测模型有较好的预测能力。     

A statistical downscaling scheme to improve global precipitation forecasting

Based on hindcasts obtained from the “Development of a European Multimodel Ensemble system for seasonal to inTERannual prediction” (DEMETER) project, this study proposes a statistical downscaling (SD) scheme suitable for global precipitation forecasting. The key idea of this SD scheme is to select the optimal predictors that are best forecast by coupled general circulation models (CGCMs) and that have the most stable relationships with observed precipitation. Developing the prediction model and further making predictions using these predictors can extract useful information from the CGCMs. Cross-validation and independent sample tests indicate that this SD scheme can significantly improve the prediction capability of CGCMs during the boreal summer (June–August), even over polar regions. The predicted and observed precipitations are significantly correlated, and the root-mean-square-error of the SD scheme-predicted precipitation is largely decreased compared with the raw CGCM predictions. An inter-model comparison shows that the multi-model ensemble provides the best prediction performance. This study suggests that combining a multi-model ensemble with the SD scheme can improve the prediction skill for precipitation globally, which is valuable for current operational precipitation prediction.     

Development of a Downscaling Method in China Regional Summer Precipitation Prediction

A downscaling method taking into account of precipitation regionalization is developed and used in the regional summer precipitation prediction (RSPP) in China. The downscaling is realized by utilizing the optimal subset regression based on the hindcast data of the Coupled Ocean-Atmosphere General Climate Model of National Climate Center (CGCM/NCC), the historical reanalysis data, and the observations. The data are detrended in order to remove the influence of the interannual variations on the selection of predictors for the RSPP. Optimal predictors are selected through calculation of anomaly correlation coe±cients (ACCs) twice to ensure that the high-skill areas of the CGCM/NCC are also those of observations, with the ACC value reaching the 0.05 significant level. One-year out cross-validation and independent sample tests indicate that the downscaling method is applicable in the prediction of summer precipitation anomaly across most of China with high and stable accuracy, and is much better than the direct CGCM/NCC prediction. The predictors used in the downscaling method for the RSPP are independent and have strong physical meanings, thus leading to the improvements in the prediction of regional precipitation anomalies.     

年际增量方法在西南夏季降水预测中的应用

利用中国西南地区80站逐月降水资料及NCEP/NCAR再分析资料等,采用降水预测新方法——年际增量法,考察影响中国西南地区夏季降水年际增量的前期冬、春季大气环流年际增量状况,并选取5个关键影响因子,采用多元回归法建立中国西南夏季降水年际增量预测模型。对降水年际增量进行预测,在1971—2010年的建模阶段,预测模型的拟合率为0.78,在2011—2017的后报检验7年中,有6年与实况值同位相。后报检验2011—2017年的降水距平百分率,均方根误差为16%。为考察对降水异常分布型的预报效果,逐站建立回归方程,并进行趋势预报检验,近5年的趋势异常综合评分高于发布预测,预报效果较好。因此,该方法的应用及模型的建立对提高西南地区夏季降水预测水平有重要意义。      

淮河流域夏季极端降水事件的统计预测模型研究

采用年际增量预测方法, 通过考察与淮河流域夏季极端降水事件发生频次(HRF)年际增量相关的环流, 确定了5个预测因子:冬季北太平洋涛动、12月南极涛动、春季3～4月南印度洋气压、春季3～4月白令海气压、春季3～4月印尼—澳洲附近经向风垂直切变;然后利用这5个预测因子, 通过多元线性回归方法建立HRF年际增量的预测模型, 进而预测HRF。交叉检验表明, 在1962～2005年的后报中, 这个预测模型对HRF显示了较高的预测技巧, 预测结果与实测间的相关系数为0.67, 表现出较高的预测潜力, 对淮河流域夏季极端降水事件的预测具有较大的应用价值。     

基于年际增量法的卷积神经网络模型对中国夏季降水的预测研究

将前冬的500 hPa位势高度、向外长波辐射和海表温度的年际增量作为预测因子,建立基于卷积神经网络(Convolutional Neural Network, CNN)的非线性预测模型,对中国160个测站夏季降水展开预测研究,并与基于线性奇异值分解(Singular Value Decomposition, SVD)的预测模型进行效果对比。结果表明：CNN在1981—2020年的交叉检验中所回报的降水平均PS评分和距平相关系数(ACC)分别为74.33和0.12,比SVD高2.15和0.06,说明CNN比SVD在整体上对夏季降水具有更好的预测能力。其中,CNN对SVD预测较好年份的预测效果提升较为明显,对SVD预测较差的年份则改进不大。CNN对中国降水预测存在一定的系统性偏差,订正后CNN对拉尼娜年的降水预测改进较大。结果表明,基于年际增量法的CNN预测模型展示出较好的潜在应用价值。     

A timescale decomposed threshold regression downscaling approach to forecasting South China early summer rainfall

A timescale decomposed threshold regression(TSDTR) downscaling approach to forecasting South China early summer rainfall(SCESR) is described by using long-term observed station rainfall data and NOAA ERSST data. It makes use of two distinct regression downscaling models corresponding to the interannual and interdecadal rainfall variability of SCESR.The two models are developed based on the partial least squares(PLS) regression technique, linking SCESR to SST modes in preceding months on both interannual and interdecadal timescales. Specifically, using the datasets in the calibration period 1915–84, the variability of SCESR and SST are decomposed into interannual and interdecadal components. On the interannual timescale, a threshold PLS regression model is fitted to interannual components of SCESR and March SST patterns by taking account of the modulation of negative and positive phases of the Pacific Decadal Oscillation(PDO). On the interdecadal timescale, a standard PLS regression model is fitted to the relationship between SCESR and preceding November SST patterns. The total rainfall prediction is obtained by the sum of the outputs from both the interannual and interdecadal models. Results show that the TSDTR downscaling approach achieves reasonable skill in predicting the observed rainfall in the validation period 1985–2006, compared to other simpler approaches. This study suggests that the TSDTR approach,considering different interannual SCESR-SST relationships under the modulation of PDO phases, as well as the interdecadal variability of SCESR associated with SST patterns, may provide a new perspective to improve climate predictions.     

多模式集合优选方案在淮河流域夏季降水预测中的应用

基于国家气候中心提供的1981—2010年4种季节气候预测模式的资料,将两种互为补充的降尺度因子挑选方案应用于淮河流域夏季降水预测,利用距平符号一致率ASCR、等级评定PG、距平相关系数ACC方法,评定了每种模式及其所采用的两种降尺度方法对淮河流域夏季降水的预测效果,并采用了一种优选方案进行多模式集合。结果表明,从4种模式的降水预测效果来看,NCEP_CFSv2和TCC_CPS1模式的评分较高,NCC_CGCM1和ECMWF_SYSTEM4模式相对较低;采用2种基于最优子集回归的降尺度方法后,NCC_CGCM1、TCC_CPS1和ECMWF_SYSTEM4模式的降尺度方法相对于模式降水预测为正订正,NCEP_CFSv2模式为负订正;将模式和降尺度预测方案进行优选,其集合平均的评分不仅高于模式降水预测的集合平均,也优于降尺度方法的集合平均,该方法发挥了不同模式的区域性优势,改进了原始集合平均的效果,为提高多模式解释应用水平提供了一种参考性方案。      

Downscaling large-scale NCEP CFS to resolve fine-scale seasonal precipitation and extremes for the crop growing seasons over the southeastern United States

Seasonally predicted precipitation at a resolution of 2.5° was statistically downscaled to a fine spatial scale of ~20 km over the southeastern United States. The downscaling was conducted for spring and summer, when the fine-scale prediction of precipitation is typically very challenging in this region. We obtained the global model precipitation for downscaling from the National Center for Environmental Prediction/Climate Forecast System (NCEP/CFS) retrospective forecasts. Ten member integration data with time-lagged initial conditions centered on mid- or late February each year were used for downscaling, covering the period from 1987 to 2005. The primary techniques involved in downscaling are Cyclostationary Empirical Orthogonal Function (CSEOF) analysis, multiple regression, and stochastic time series generation. Trained with observations and CFS data, CSEOF and multiple regression facilitated the identification of the statistical relationship between coarse-scale and fine-scale climate variability, leading to improved prediction of climate at a fine resolution. Downscaled precipitation produced seasonal and annual patterns that closely resemble the fine resolution observations. Prediction of long-term variation within two decades was improved by the downscaling in terms of variance, root mean square error, and correlation. Relative to the coarsely resolved unskillful CFS forecasts, the proposed downscaling drove a significant reduction in wet biases, and correlation increased by 0.1–0.5. Categorical predictability of seasonal precipitation and extremes (frequency of heavy rainfall days), measured with the Heidke skill score (HSS), was also improved by the downscaling. For instance, domain averaged HSS for two category predictability by the downscaling are at least 0.20, while the scores by the CFS are near zero and never exceed 0.1. On the other hand, prediction of the frequency of subseasonal dry spells showed limited improvement over half of the Georgia and Alabama region.     

中国夏季降水的组合统计降尺度模型预测研究

现阶段的动力气候模式尚不能满足东亚区域气候预测的实际需求,这就需要动力和统计相结合的方法,将动力模式中具有较高预测技巧的大尺度环流信息应用到降水等气象要素的统计预测模型当中,以改善后者预测效果。本文中所介绍的组合统计降尺度模型,可将动力气候模式预测的大尺度环流变量和前期观测的外强迫信号作为预测因子来预测中国夏季降水异常。交叉检验结果显示,组合统计降尺度预测模型的距平相关系数较原始模式结果有较大提高。在实时夏季降水预测中,2013～2018年平均的预测技巧相对较高,趋势异常综合检验（PS）评分平均为71.5分,特别是2015～2018年平均的PS评分预测技巧达到72.7分,总体上高于业务模式原始预测和业务发布预测的技巧。该组合统计降尺度模型预测性能稳定,为我国季节预测业务提供了一种有效参考。     

基于土壤湿度和年际增量方法的我国夏季降水预测试验

选取欧亚大陆9个关键区的土壤湿度年际增量作为预测因子,采用变形的典型相关分析(BP-CCA)结合集合典型相关分析(ECC)方法建立集合预测模型,对我国东部夏季降水的年际增量进行预测,进而预测夏季降水。其中,1980~2004年的资料用于历史预测试验,而2005~2014年的资料用于独立样本预测试验。首先利用BP-CCA方法对9个因子分别建立单因子预测模型,然后采用ECC方法对9个预测因子按照不同的组合方式建立集合预测模型,并且对独立样本检验的效果进行了评估。结果表明,不同预测因子的组合对我国夏季降水均表现出一定的预测能力:东欧平原、贝加尔湖以北、我国河套地区及长江以南地区的土壤湿度对华北夏季降水预测效果较好;而巴尔喀什湖以北地区、我国西北地区、河套地区以及长江以南地区的土壤湿度对江淮夏季降水有较好预测效果;东欧平原、巴尔喀什湖以北地区以及我国河套地区的土壤湿度对华南降水预测技巧较高。这三组模型预测出的降水变化趋势与相应区域的观测结果较为一致,且预测评分(PS)均超过70分,距平相关系数(ACC)均为正值。研究表明土壤湿度因子中包含了对我国夏季降水有用的预测信号,可以考虑将土壤湿度应用于夏季降水的预测业务中。     

匹配域投影降尺度方法在黄河流域夏季降水预测中的应用

文章采用黄河流域夏季降水数据和BCC-CGCM模式资料,利用匹配域投影降尺度方法对黄河流域夏季降水进行预测,得到以下结论:(1)交叉验证期,匹配域投影降尺度方法对黄河流域夏季降水的预测效果要好于原始模式预测,且较模式直接输出的要素预测稳定;分月预测比夏季整体预测效果要好。(2)匹配域投影降尺度方法对各个区域的预测能力不同,在夏季(6—8月)预测中,预测较好区域比较分散,而分月预测中,预测较好的区域比较集中。月份不同,降尺度方法对于不同地区的预测能力也不同。(3)2009—2013年的独立样本检验表明,匹配域投影降尺度方法对于黄河流域夏季降水的预测效果要明显好于模式直接输出的要素预测。尤其6和7月的降尺度预测较模式直接输出的要素预测有较大提高。