共查询到19条相似文献，搜索用时 109 毫秒
1.
提出一种基于数值模式预报产品的气温预报集成学习误差订正方法,通过人工神经网络、长短期记忆网络和线性回归模型组合出新的集成学习模型（ALS模型）,采用2013—2017年的欧洲中期天气预报中心数值天气预报模式2 m气温预报产品和中国部分气象站点数据,利用气象站点气温、风速、气压、相对湿度4个观测要素,挖掘观测数据的时序特征并结合模式2 m气温预报结果训练机器学习模型,对2018年模式2 m气温6~168 h格点预报产品插值到站点后的预报结果进行偏差订正。结果表明:ALS模型可将站点气温预报整体均方根误差由3.11℃降至2.50℃,降幅达0.61℃（19.6%）,而传统的线性回归模型降幅为0.23℃（8.4%）。ALS模型对站点气温预报误差较大的区域和气温峰值预报的订正效果尤为显著,因此,集成学习方法在数值模式预报结果订正中具有较大的应用潜力。 相似文献
2.
ECMWF模式地面气温预报的四种误差订正方法的比较研究 总被引：16，自引：5，他引：11
采用均方根误差对欧洲中期天气预报中心（ECWMF）确定性预报模式2007年1月至2010年12月的地面气温预报结果进行评估,并分别利用一元线性回归、多元线性回归、单时效消除偏差和多时效消除偏差平均的订正方法,对ECMWF模式地面气温预报结果进行订正。结果表明,4种订正方法都能有效地减小地面气温多个时效预报的误差,改进幅度约为1℃。在短期预报中仅考虑最新预报结果的一元线性回归订正方法要优于考虑多个预报结果的多元集成预报订正方法。在中期预报中考虑多个预报结果的多元集成预报订正方法更优,更稳定。在模式预报误差较大的情况下,多时效集成的订正方法能更稳定地减小误差。 相似文献
3.
延伸期温度预报误差订正技术初探 总被引：1，自引：0，他引：1
应用滑动平均误差订正方法和历史偏差订正方法,对欧洲中期天气预报中心的数值模式延伸期2 m温度预报进行误差订正。研究发现,应用滑动平均误差订正方法进行11～15 d逐日温度预报订正时,25～30 d是最优的训练期长度。对2018年订正预报的检验分析显示,应用上述两种误差订正方法均可减小模式预报的系统偏差,有效修正模式温度预报较实况明显偏低的问题,并将预报准确率提高30%以上。在6—10月,订正后的温度预报平均绝对误差基本在2℃以内,具有一定的参考性,其业务化产品可支撑预报员的业务预报需求。在15 d内的延伸期预报时效上,两种订正方法对温度预报的订正效果差异并不明显。随着时效的延长,历史偏差订正方法的优势逐渐显现。 相似文献
4.
基于WRF模式的太阳辐射预报初步试验研究 总被引：1，自引：0，他引：1
采用中尺度气象模式WRF(Weather Research Forecast)对北京地区的太阳辐射进行了4个典型月的逐时预报试验,用南郊观象台的辐射观测数据对预报结果进行了对比分析和初步订正试验。结果表明:在现有模式条件下,5 km分辨率的短波辐射预报结果和1 km分辨率预报结果无明显差别;WRF模式对太阳辐射的预报性能在晴天较好,多云天次之,在满云或阴雨天最差;通过误差分解发现,位相偏差、系统偏差及振幅偏差在各月对均方根误差的贡献有明显差异;针对模式预报结果的系统偏差和振幅偏差。经过简单的线性订正可以较明显地改进模式预报结果;双偏订正(DBC)法比线性回归(LR)法对预报误差的改进效果略明显;仅通过简单的线性订正,位相差很难消除,需要针对位相差研究新的订正方法。 相似文献
5.
GRAPES_RAFS系统2 m温度偏差订正方法研究 总被引：7，自引：5，他引：2
本文通过对2013年6月20日至7月20日GRAPES(Global and Regional Assimilation and Prediction System)_RAFS(Rapid Analysis and Forecast System)系统每天8个时次每3h的2 m温度预报进行分析,发现各时次的预报均能较好地表征2 m温度日变化特征,但预报与实况存在一定的偏差,其中西藏东部川西高原、云贵高原、江南武夷山脉偏低于实况可达3℃,而华北地区偏高于实况3℃以上.为了减小GRAPES_RAFS系统偏差对2m温度预报的影响,本文采用平均法、双权重平均法、滑动平均法和滑动双权重平均法分别对GRAPES_RAFS系统2 m温度预报产品进行偏差订正,并对订正前后的结果进行检验分析和对比.结果表明:2 m温度订正后的平均误差大部地区减小到(一1～1℃),而均方根误差大部地区降低到2.5℃内.对于偏差较大地区,订正效果更为明显,如西藏东部川西高原,经过订正,平均误差绝对值由订正前3℃以上降低到1℃内,而RMSE由订正前4℃以上控制到3℃内.对比四种订正方法,双权重订正方法与平均法订正整体效果接近,但对个别站点,双权重订正法要优于平均法,经过滑动的订正方法比无滑动的订正方法订正效果更好,订正效果最好的是滑动双权重平均法,全国平均误差大部分在(-0.5～0.5℃)内,不超过(-1～1℃)的范围. 相似文献
6.
几种格点化温度滚动订正预报方案对比研究 总被引：1，自引：1，他引：0
为了快速获得更为精准的格点温度预报产品,使用国家信息中心高分辨率、高频次的温度格点多元融合产品和欧洲中期天气预报中心全球模式2 m温度预报场资料,采用8种误差订正方案进行滚动订正预报试验。选择2017年1月1日至2月28日和6月1日至7月31日两个时间段进行两次回报模拟试验,并对订正前后的预报结果进行格点和站点检验分析,结果表明:8种方案对模式直接输出的预报场有正技巧订作用,全格点滑动误差回归模型订正和全格点滑动双因子回归模型订正效果最优,两种方案都能使订正场的格点平均绝对误差在2℃以下,3、6和9 h的格点准确率均在0.9以上。全格点滑动误差回归模型的检验评分略微好于全格点滑动双因子回归模型,表明作为预报模型因子的起报时刻误差场比数值模式因子在短期订正中扮演着更为重要的角色。 相似文献
7.
本文通过分析2017年9~12月四川地区ECMWF(European Centre for Medium-Range Weather Forecasting)细网格模式、GRAPES_GFS(Global and Regional Assimilation and Prediction System)全球模式和西南区域模式(South West Center-WRF ADAS Real-time Modeling System, SWCWARMS)2m温度168h预报时效内的系统性偏差特征,采用滑动双权重平均法分别对三种模式温度预报产品进行偏差订正,并集成得到各时效2m温度的订正场,结果表明:(1)三种模式的预报存在明显的日变化,整体上EC模式的预报最优。(2)三种模式对于低温和高温的预报,在全省均大致呈现负的系统性误差,特别在高原及过渡区表现的尤为明显。(3)订正后三种模式的预报准确率显著提高,均方根误差减小1.4~2.5℃,大部分地区平均误差维持在±0.5℃之间,在系统性偏差较大的地区,订正效果更好。(4)两种集成方案预报结果接近,且均优于三种模式的订正预报。 相似文献
8.
9.
基于欧洲中期天气预报中心(European Centre for Medium-range weather Forecasts,ECMWF)模式的预报数据和北京地区气象站点的观测数据,使用两种机器学习算法(线性回归和梯度提升回归树)对站点的体感温度进行误差订正,并采用均方根误差(Root Mean Square Error,RMSE)对预报效果进行评估,进一步与传统订正方法模式输出统计(Model Output Statistics,MOS)得到的订正结果进行对比。结果表明:线性回归、梯度提升回归树、MOS和ECMWF预报得到的平均RMSE分别为3.12、3.06、3.45、4.06℃,即机器学习算法明显优于MOS和ECMWF模式原始预报。机器学习订正方法不仅在平原地区取得了较好的效果,在高海拔站点的订正效果更加突出,为北京冬奥会复杂山地条件下赛事正常运行提供了一定的技术保障。 相似文献
10.
基于2008—2017年全国自动气象观测站逐旬土壤相对湿度观测数据,综合评估中国气象局陆面数据同化系统（CMA Land Data Assimilation System,CLDAS）0～20 cm层融合土壤相对湿度产品在中国地区的适用性,评估表明CLDAS土壤相对湿度产品在中国东北、西北、江南大部及华南等地区存在较大系统性误差,总体上适用性较差。为消除CLDAS土壤相对湿度产品的系统性误差,采用回归订正法、7旬滑动平均订正法和临近加权前旬订正法对CLDAS土壤相对湿度产品进行误差订正处理,对订正结果评估发现：订正处理后CLDAS土壤相对湿度产品与站点观测的相关性显著增加,系统偏差基本消除,适用性明显提高,3种订正方法中临近加权前旬订正法的订正效果最优。最后,采用经不同方法订正后的CLDAS土壤相对湿度产品对2017年5月东北—华北地区一次气象干旱个例进行重现,对比验证表明：相对其他两种订正方法,经临近加权前旬订正法处理后的CLDAS土壤相对湿度产品能更为精准地重现2017年5月东北—华北地区气象干旱的落区和强度。〖JP〗 相似文献
11.
该文介绍了卫星观测海表温度 (SST) 算法的发展历程,给出了所用SST算法的回归模型,并在FY-3B/VIRR业务SST算法的基础上进行了改进。基于NOAA-19/AVHRR匹配数据集,进行多算法建模分析及精度评估,白天最优算法为非线性SST (NL) 算法,夜间最优算法为三通道SST (TC) 算法,最优算法的确定与NESDIS/STAR一致。建立2012年8月—2013年3月FY-3B/VIRR匹配数据集,并在此基础上进行多算法回归建模及精度评估,白天和夜间的最优均为NL算法,分析发现夜间TC算法采用匹配数据集版本2(MDB_V2) 时,3.7 μm通道存在类似百叶窗的条带现象。以2012年10—12月FY-3B/VIRR匹配数据集计算回归系数,以2013年1—3月独立样本进行精度评估,与浮标SST相比,NL算法白天和夜间的均方根误差分别为0.41℃和0.43℃。与日平均最优插值海温 (OISST) 相比,NL算法白天和夜间的均方根误差分别为1.45℃和1.5℃; 选择与OISST偏差在2℃以内的样本,NL算法白天和夜间均方根误差分别为0.82℃和0.84℃。 相似文献
12.
Influence of SST biases on future climate change projections 总被引：1，自引：0，他引：1
We use a quantile-based bias correction technique and a multi-member ensemble of the atmospheric component of NCAR CCSM3 (CAM3) simulations to investigate the influence of sea surface temperature (SST) biases on future climate change projections. The simulations, which cover 1977?C1999 in the historical period and 2077?C2099 in the future (A1B) period, use the CCSM3-generated SSTs as prescribed boundary conditions. Bias correction is applied to the monthly time-series of SSTs so that the simulated changes in SST mean and variability are preserved. Our comparison of CAM3 simulations with and without SST correction shows that the SST biases affect the precipitation distribution in CAM3 over many regions by introducing errors in atmospheric moisture content and upper-level (lower-level) divergence (convergence). Also, bias correction leads to significantly different precipitation and surface temperature changes over many oceanic and terrestrial regions (predominantly in the tropics) in response to the future anthropogenic increases in greenhouse forcing. The differences in the precipitation response from SST bias correction occur both in the mean and the percent change, and are independent of the ocean?Catmosphere coupling. Many of these differences are comparable to or larger than the spread of future precipitation changes across the CMIP3 ensemble. Such biases can affect the simulated terrestrial feedbacks and thermohaline circulations in coupled climate model integrations through changes in the hydrological cycle and ocean salinity. Moreover, biases in CCSM3-generated SSTs are generally similar to the biases in CMIP3 ensemble mean SSTs, suggesting that other GCMs may display a similar sensitivity of projected climate change to SST errors. These results help to quantify the influence of climate model biases on the simulated climate change, and therefore should inform the effort to further develop approaches for reliable climate change projection. 相似文献
13.
利用我国541个测站1960—2010年气温资料以及国家气候中心参加第5次耦合模式比较计划 (CMIP5) 的气候系统模式BCC_CSM1.1的历史试验和年代际试验结果,评估了该模式对我国近50年气温变化特征的模拟能力, 对模式的年代际试验结果进行了误差订正,并给出未来10~20年我国气温变化的预估。结果表明:历史试验和年代际试验均模拟出了与观测较为一致的增暖趋势,但均没有观测资料的增暖幅度大。其中,历史试验比年代际试验更接近于观测。年代际尺度上,模式对我国东部的模拟要好于西部;年际尺度上,模式的高预报技巧区在我国西北地区西南部和东部、西南地区北部。历史试验和年代际试验对我国气温空间场整体分布模拟较好,误差订正后的年代际试验结果对空间气温场的模拟有更好把握。相对于观测资料得到的1960—2010年0.27℃/10 a的增温速率,模式预估我国2011—2030年平均气温变化速率达到0.48℃/10 a, 上升趋势更加明显。 相似文献
14.
Tropical cyclone (TC) annual frequency forecasting is significant for disaster prevention and mitigation in Guangdong Province. Based on the NCEP-NCAR reanalysis and NOAA Extended Reconstructed global sea surface temperature (SST) V5 data in winter, the TC frequency climatic features and prediction models have been studied. During 1951-2019, 353 TCs directly affected Guangdong with an annual average of about 5.1. TCs have experienced an abrupt change from abundance to deficiency in the mid to late 1980 with a slightly decreasing trend and a normal distribution. 338 primary precursors are obtained from statistically significant correlation regions of SST, sea level pressure, 1000hPa air temperature, 850hPa specific humidity, 500hPa geopotential height and zonal wind shear in winter. Then those 338 primary factors are reduced into 19 independent predictors by principal component analysis (PCA). Furthermore, the Multiple Linear Regression (MLR), the Gaussian Process Regression (GPR) and the Long Short-term Memory Networks and Fully Connected Layers (LSTM-FC) models are constructed relying on the above 19 factors. For three different kinds of test sets from 2010 to 2019, 2011 to 2019 and 2010 to 2019, the root mean square errors (RMSEs) of MLR, GPR and LSTM-FC between prediction and observations fluctuate within the range of 1.05-2.45, 1.00-1.93 and 0.71-0.95 as well as the average absolute errors (AAEs) 0.88-1.0, 0.75-1.36 and 0.50-0.70, respectively. As for the 2010-2019 experiment, the mean deviations of the three model outputs from the observation are 0.89, 0.78 and 0.56, together with the average evaluation scores 82.22, 84.44 and 88.89, separately. The prediction skill comparisons unveil that LSTM-FC model has a better performance than MLR and GPR. In conclusion, the deep learning model of LSTM-FC may shed light on improving the accuracy of short-term climate prediction about TC frequency. The current research can provide experience on the development of deep learning in this field and help to achieve further progress of TC disaster prevention and mitigation in Guangdong Province. 相似文献
15.
The seasonal cycle and interannual variability in the tropical oceans simulated by three versions of the Flexible Ocean-Atmosphere-Land System (FGOALS) model (FGOALS-g1.0, FGOALS-g2 and FGOALSs2), which have participated in phases 3 and 5 of the Coupled Model Intercomparison Project (CMIP3 and CMIP5), are presented in this paper. The seasonal cycle of SST in the tropical Pacific is realistically reproduced by FGOALS-g2 and FGOALSs2, while it is poorly simulated in FGOALS-g1.0. Three feedback mechanisms responsible for the SST annual cycle in the eastern Pacific are evaluated. The ocean-atmosphere dynamic feedback, which is successfully reproduced by both FGOALS-g2 and FGOALS-s2, plays a key role in determining the SST annual cycle, while the overestimated stratus cloud-SST feedback amplifies the annual cycle in FGOALS-s2. Because of the serious warm bias existing in FGOALS-g1.0, the ocean-atmosphere dynamic feedback is greatly underestimated in FGOALS-g1.0, in which the SST annual cycle is mainly driven by surface solar radiation. FGOALS-g1.0 simulates much stronger ENSO events than observed, whereas FGOALS-g2 and FGOALSs2 successfully simulate the observed ENSO amplitude and period and positive asymmetry, but with less strength. Further ENSO feedback analyses suggest that surface solar radiation feedback is principally responsible for the overestimated ENSO amplitude in FGOALS-g1.0. Both FGOALS-g1.0 and FGOALS-s2 can simulate two different types of El Ni-no events — with maximum SST anomalies in the eastern Pacific (EP) or in the central Pacific (CP) — but FGOALS-g2 is only able to simulate EP El Ni-no, because the negative cloud shortwave forcing feedback by FGOALS-g2 is much stronger than observed in the central Pacific. 相似文献
16.
Systematic model error remains a difficult problem for seasonal forecasting and climate predictions. An error in the mean state could affect the variability of the system. In this paper, we investigate the impact of the mean state on the properties of ENSO. A set of coupled decadal integrations have been conducted, where the mean state and its seasonal cycle have been modified by applying flux correction to the momentum-flux and a combination of heat and momentum fluxes. It is shown that correcting the mean state and the seasonal cycle improves the amplitude of SST inter-annual variability and also the penetration of the ENSO signal into the troposphere and the spatial distribution of the ENSO teleconnections are improved. An analysis of a multivariate PDF of ENSO shows clearly that the flux correction affects the mean, variance, skewness and tails of the distribution. The changes in the tails of the distribution are particularly noticeable in the case of precipitation, showing that without the flux correction the model is unable to reproduce the frequency of large events. For the inter-annual variability the momentum-flux correction alone has a large impact, while the additional heat-flux correction is important for the teleconnections. These results suggest that the current forecasts practices of removing the forecast bias a-posteriori or anomaly initialisation are by no means optimal, since they can not deal with the strong nonlinear interactions. A consequence of the results presented here is that the predictability on annual time-ranges could be higher than currently achieved. Whether or not the correction of the model mean state by some sort of flux correction leads to better forecasts needs to be addressed. In any case, flux correction may be a powerful tool for diagnosing coupled model errors and predictability studies. 相似文献
17.
国家卫星气象中心FY-3C/VIRR（visible and infrared radiometer,可见光红外扫描辐射计）海表温度产品在云检测产品的基础上,采用多通道MCSST（multichannel SST）算法进行晴空区海温反演。该文详细介绍了海表温度产品算法、产品设计、质量控制及质量检验方法。FY-3C/VIRR海表温度产品包括5 min段原始投影海温和5 km全球等经纬度投影海温。设计逐像元的海温质量标识,将海温像元分为优、良、差3个等级,用户可根据应用目标选择海温的质量等级。与日最优插值海温OISST（optimum interpolation SST）相比,FY-3C/VIRR 2015年1月—2019年12月的5 min段海温质量检验结果表明:质量等级为优的海温,白天和夜间的偏差分别为-0.18℃和-0.06℃,均方根误差分别为0.85℃和0.8℃;白天海温均方根误差有季节性波动,夏季有的月份均方根误差大于1℃（如2015年7月、2016年7月和2019年7月）;在海温回归系数不变的条件下,夜间海温偏差的季节性波动与星上黑体温度相关显著。从一级数据质量、定位、业务运行状况等方面讨论引起海表温度产品异常的原因,为FY-3C/VIRR历史数据定位、定标和产品重处理及用户应用提供重要的参考信息。 相似文献
18.
模式预报的订正是决定局地天气预报结果的一个重要步骤,基于机器学习的后处理模型近年来开始崭露头角。本文发展了基于岭回归（Ridge）、随机森林（Random Forest,RF）和深度学习（Deep Learning,DL）的3种后处理模型,基于中国气象局（CMA）的BABJ模式、欧洲中期天气预报中心（ECMWF）的ECMF模式、日本气象厅（JMA）的RJTD模式和NCEP的KWBC模式这4个数值天气预报模式2014年2月至2016年9月（训练期）近地面2 m气温预报和实况资料确定各模型参数,进而对2016年10月至2017年9月（预报期）华北地区（38°N~43°N,113°E~119°E）的逐日地面2 m气温预报进行了多模式集合预报分析。采用均方根误差对预报效果进行评估,这3种后处理模型的预报效果和4个数值天气预报模式以及通常的多模式集合平均（Ensemble Mean,EMN）的预报效果的对比表明:1）随着预报时长增加,4个数值预报模式及各种后处理模型的均方根误差均呈上升趋势;但区域平均而言,Ridge、RF和DL的预报效果在任何预报时长上都明显优于EMN和单个天气预报模式;特别是前几天的短期预报DL的预报效果更好,中后期预报Ridge的预报效果略好。2）华北地区的东南部均方根误差较小,其余格点上均方根误差较高,从空间分布而言,DL的订正预报效果最好,3种机器学习模型的误差在1.24~1.26℃之间,而EMN的误差达1.69℃。3）夏季各种方法的预报效果都较好,冬季预报效果都较差;但是Ridge、RF和DL的预报效果明显优于EMN,这3种模型预报的平均均方根误差在2.15~2.18℃之间,而EMN的平均均方根误差达2.45℃。 相似文献
19.
A fast coupled global climate model (CGCM) is used to study the sensitivity of El Ni?o Southern Oscillation (ENSO) characteristics to a new interactive flux correction scheme. With no flux correction applied our CGCM reveals typical bias in the background state: for instance, the cold tongue in the tropical east Pacific becomes too cold, thus degrading atmospheric sensitivity to variations of sea surface temperature (SST). Sufficient atmospheric sensitivity is essential to ENSO. Our adjustment scheme aims to sustain atmospheric sensitivity by counteracting the SST drift in the model. With reduced bias in the forcing of the atmosphere, the CGCM displays ENSO-type variability that otherwise is absent. The adjustment approach employs a one-way anomaly coupling from the ocean to the atmosphere: heat fluxes seen by the ocean are based on full SST, while heat fluxes seen by the atmosphere are based on anomalies of SST. The latter requires knowledge of the model??s climatological SST field, which is accumulated interactively in the spin-up phase (??training??). Applying the flux correction already during the training period (by utilizing the evolving SST climatology) is necessary for efficiently reducing the bias. The combination of corrected fluxes seen by the atmosphere and uncorrected fluxes seen by the ocean implies a restoring mechanism that counteracts the bias and allows for long stable integrations in our CGCM. A suite of sensitivity runs with varying training periods is utilized to study the effect of different levels of bias in the background state on important ENSO properties. Increased duration of training amplifies the coupled sensitivity in our model and leads to stronger amplitudes and longer periods of the Nino3.4 index, increased emphasis of warm events that is reflected in enhanced skewness, and more pronounced teleconnections in the Pacific. Furthermore, with longer training durations we observe a mode switch of ENSO in our model that closely resembles the observed mode switch related to the mid-1970s ??climate shift??. 相似文献