基于ConvLSTM的广西短临降水预报

针对传统雷达回波外推算法在快速增长或消散降水过程预报精度较低的问题,利用华南雷达回波拼图资料数据,建立ConvLSTM回波外推模型,对广西区域范围进行短临降水预报研究.采用气象业务中的正确率(POD)、临界成功指数(CSI)及误报率(FAR)评判标准检验预报模型,并将ConvLSTM与光流法的预报结果进行对比分析.结果 表明,ConvLSTM模型的CSI、POD分别比光流法提高0.06和0.059,而FAR下降了0.058.ConvLSTM方法比光流法的回波外推预报准确率高,该方法可为广西短临降水预报提供新的参考.     

BCC第二代气候预测模式系统对2015年一次寒潮过程的预报能力评估

利用BCC第二代气候预测模式系统(BCC_CSM2)的回报试验结果,评估了BCC_CSM2对2015年1月27—31日一次强寒潮过程的次季节预报能力,结果表明:(1)此次寒潮过程主要由新地岛以西的短波槽不断东移发展而形成的,模式能够提前10 d较好地预报过程期间降温以及高空环流形势,相关系数、距平符号一致率以及均方根误差都定量表明模式在10 d左右具有较好的预报能力,但是对降温程度的预报能力随起报时间的延长逐渐降低;(2)为了探讨随起报时间延长模式预报能力降低的原因,从位势倾向方程出发,分析相对涡度平流和温度平流随高度变化发现,在模式提前10 d之内的预报时段内,模式预报的相对涡度平流和温度平流随高度变化与再分析资料的诊断结果基本一致,能够合理预测短波槽的东移和槽脊的强度变化,当预报超过10 d后,模式中相对涡度平流的配置不利于短波槽的东移,模式预报的低层出现暖平流,并随高度增加而减小,不利于槽的加深,使模式预报的环流形势产生偏差,导致模式预报能力降低。     

基于ECMWF细网格产品的一种优化BP-MOS气温预报方法

基于ECMWF细网格模式输出产品,以一种优化的BP-MOS模型预测1~7 d日最高和最低气温,并对比该方法和ECMWF细网格的2 m温度输出产品以及线性MOS方法的预报效果。结果表明:在预报因子处理时,考虑云量、风、湿度等对气温变化的"过程"影响能有效提高预报准确率;ECMWF细网格2 m温度产品在短期3 d内均方根误差均在2℃以内,但中期时段预报效果明显低于MOS方法;由于线性MOS模型预报存在不稳定现象,而BP神经网络的非线性映射关系使其在容错性方面优势明显,因此优化的BP-MOS模型预测效果良好。     

风廓线雷达反演温度平流的应用

利用北京延庆站风廓线雷达水平风廓线资料进行大气温度平流的反演,详细分析2014年11月15日夜间冷空气入侵过程,并统计分析2015年9—11月6次冷空气入侵过程,同T639L60模式的预报风场及温度平流预报产品进行对比。结果表明:在一定预报时效内 (约6~12 h),风廓线雷达获取的水平风廓线与模式给出的预报风场有较好的一致性;由风廓线雷达反演的温度平流与模式给出的温度平流量级相同,温度平流属性一致;风廓线雷达6 min完成1次垂直高度分辨率为120 m的探测,高时空分辨率使风廓线雷达获取的温度平流较T639L60模式更能反映大气温度平流的细节。     

基于数值预报及上级指导产品的本地气温MOS预报方法

利用恩施基准站2008—2011年地面实测资料、数值预报产品、上级台站指导产品,采用M(数值模式预报)、E(天气学经验)、D(诊断分析)相结合的方法(简称MED),从气温变化的影响因素如大气稳定度、温度平流、水汽条件等设计具有物理意义的预报因子。或根据需要进行因子的组合叠加利用,并将指导产品直接作为预报因子。利用常规统计预报方法(逐步回归),将高、低温实况作为预报对象来建立地方气温预报模式。结果表明:将天空状况进行分型后建立地方气温MOS预报模型,并对应高低温一般出现时间段来选取数值预报产品因子进行预报,对本地气温预报质量的提高有积极的意义;模型建立过程中,综合采用了MED相结合的方法,并应用了数值预报再加工因子,考虑了天气系统变化对气温的影响,增强了数值预报的解释应用能力;参考客观数值模式产品、上级业务部门指导产品相结合的综合MOS预报方法,建立地方气温MOS预报是提高本地天气预报准确率的有效尝试。检验结果也表明,本地气温MOS预报效果较好,明显高于指导预报,已较好应用于实际业务中。     

基于20—30d振荡的长江下游地区夏季低频降水延伸期预报方法研究

用长江下游降水低频分量和环流低频主成分,构造多变量时滞回归模型(MLR)和主成分复数自回归模型(PC-CAR)的混合预报模型(MLR/PC-CAR),对长江下游降水低频分量进行延伸期逐日变化预报,延长预报时效。通过2011年6—8月预测试验表明,20—30 d时间尺度的长江下游低频降水预测时效可达50 d左右,采用南半球中高纬度地区850 hPa低频经向风的主成分作为预测因子的模型的预测精度明显高于东亚地区低频经向风作为预测因子的模型。这表明在20—30 d时间尺度上,长江下游降水与南半球中纬度绕球遥相关(SCGT)型有关的主分量的时滞相关更加密切。进一步对于较强20—30 d振荡的多年资料构建的MLR/PC-CAR混合模型预测试验表明,SCGT是预测夏季长江下游低频降水未来50 d变化的显著信号。基于SCGT的发展和演变,对于把握类似长江下游地区2011年6月初旱涝急转和7月中旬持续降水和强降水过程异常变化过程很有帮助,SCGT可以作为夏季长江下游20—30 d低频降水和强降水过程进行延伸期预报的主要可预报性来源之一。     

川西高原地区SCMOC温度精细化指导预报质量检验及影响因子分析

利用2014~2015年阿坝州13站共730天08:00和20:00起报的SCMOC温度精细化指导预报资料,对比实况日最高(低)气温,进行预报质量检验。结果表明:日最高(低)气温预报准确率与预报时效成反比,两个时次预报的最低气温准确率高于最高气温,且最低气温预报准确率有明显的季节变化。08:00起报的日最低气温多出现负误差,其余预报最高(低)气温多出现正误差。日最低气温预报绝对误差与海拔高度有关。24h最高(低)气温预报绝对误差>4℃样本分析表明,温度平流、大气稳定度与非绝热过程对温度的影响明显,造成气温偏差的主要原因是降水及冷空气影响范围和强度,冷、暖平流影响偏差,高空槽强度和移动偏差等几方面。     

青藏高原地表特征对短期数值预报效果的影响

本文用实际天气过程的数值预报,试验了青藏高原反照率、拖曳系数及土壤温度等因子对短期数值预报质量的影响。结果表明,对于高空环流形势,在48小时预报时效内影响不明显,但青藏高原的热力作用可改变高低空流场强度以及降水分布和降水量。     

全球海洋资料同化系统ZFL_GODAS的研制和初步评估试验

本研究发展了一个全球海洋资料同化系统ZFL_GODAS。该系统是一个短期气候数值预测业务系统的子系统,为短期气候预测海气耦合模式提供全球海洋初始场。系统能够同化的观测资料包括卫星高度计资料、卫星海表温度(SST)资料,以及Argo、XBT、TAO等各种不同来源的现场温盐廓线资料。系统使用的海洋模式为中国科学院大气物理研究所大气科学和地球流体力学数值模拟国家重点实验室开发的气候系统海洋模式LICOM1.0,同化方案为集合最优插值(EnOI)方案。系统使用一个由海洋模式自由积分得到的静态样本来估计背景场误差协方差。这样的基于集合样本的背景场误差协方差具有多变量协变、各向异性的特征,且能反映海洋物理过程固有的空间尺度特征。针对EnOI同化程序的特点,开发了一套特色鲜明、负载均衡、高效的并行化同化程序。本文通过与不同类型观测资料的比较,对同化系统的性能进行了评估。通过比较海表温度和海面高度的年际变率,海表温度异常随时间的变化,SST、海面高度异常(SLA)以及次表层温盐预报产品的均方根误差,5年平均温度偏差廓线、平均盐度廓线、平均纬向流速廓线等发现:系统工作正常、同化效果较好;经过同化以后,各变量都更加接近观测,误差更小,与观测场的相关性更好,可以为短期气候预测系统提供较好的海洋初始场,也可以为物理海洋学的研究提供有效的再分析资料。     

基于集合预报和支持向量机的中期强降雨集成预报试验

本文基于欧洲中期天气预报中心(ECMWF)和美国国家环境预报中心(NCEP)集合预报资料和支持向量机(SVM)回归方法建立了多模式集成的动力-统计客观预报模型(SVM-多模式集成预报),继而选用2012年5—9月(共计153 d)发生在淮河流域及其以南地区的大雨和暴雨开展了回报试验,并将所得预报结果与ECMWF的控制预报和集合平均预报进行了多角度比对评估。结果表明:在中期预报时效(4~7 d),SVM-多模式集成预报方法对2012年5—9月大雨和暴雨的预报效果最优,尤其对暴雨预报准确率明显提高,其优势主要体现在对强降雨中心分布范围和强度的预报更接近实况。     

The role of atmospheric internal variability on the prediction skill of interannual North Pacific sea-surface temperatures

The sensitivity of the sea-surface temperature (SST) prediction skill to the atmospheric internal variability (weather noise) in the North Pacific (20^°–60^°N;120^°E–80^°W) on decadal timescales is examined using state-of-the-art Climate Forecasting System model version 2 (CFS) and a variation of CFS in an Interactive Ensemble approach (CFSIE), wherein six copies of atmospheric components with different perturbed initial states of CFS are coupled with the same ocean model by exchanging heat, momentum and fresh water fluxes dynamically at the air-sea interface throughout the model integrations. The CFSIE experiments are designed to reduce weather noise and using a few ten-year long forecasts this study shows that reduction in weather noise leads to lower SST forecast skill. To understand the pathways that cause the reduced SST prediction skill, two twenty-year long forecasts produced with CFS and CFSIE for 1980-2000 are analyzed for the ocean subsurface characteristics that influence SST due to the reduction in weather noise in the North Pacific. The heat budget analysis in the oceanic mixed layer across the North Pacific reveals that weather noise significantly impacts the heat transport in the oceanic mixed layer. In the CFSIE forecasts, the reduced weather noise leads to increased variations in heat content due to shallower mixed layer, diminished heat storage and enhanced horizontal heat advection. The enhancement of the heat advection spans from the active Kuroshio regions of the east coast of Japan to the west coast of continental United States and significantly diffuses the basin-wide SST anomaly (SSTA) contrasts and leads to reduction in the SST prediction skill in decadal forecasts.     

Global seasonal climate predictability in a two tiered forecast system: part I: boreal summer and fall seasons

This paper shows demonstrable improvement in the global seasonal climate predictability of boreal summer (at zero lead) and fall (at one season lead) seasonal mean precipitation and surface temperature from a two-tiered seasonal hindcast forced with forecasted SST relative to two other contemporary operational coupled ocean–atmosphere climate models. The results from an extensive set of seasonal hindcasts are analyzed to come to this conclusion. This improvement is attributed to: (1) The multi-model bias corrected SST used to force the atmospheric model. (2) The global atmospheric model which is run at a relatively high resolution of 50 km grid resolution compared to the two other coupled ocean–atmosphere models. (3) The physics of the atmospheric model, especially that related to the convective parameterization scheme. The results of the seasonal hindcast are analyzed for both deterministic and probabilistic skill. The probabilistic skill analysis shows that significant forecast skill can be harvested from these seasonal hindcasts relative to the deterministic skill analysis. The paper concludes that the coupled ocean–atmosphere seasonal hindcasts have reached a reasonable fidelity to exploit their SST anomaly forecasts to force such relatively higher resolution two tier prediction experiments to glean further boreal summer and fall seasonal prediction skill.     

Wave forecast in the Atlantic Ocean using a double-stage ConvLSTM network

Accurate forecasting of ocean waves is of great importance to the safety of marine transportation. Despite wave forecasts having been improved, the current level of prediction skill is still far from satisfactory. Here, the authors propose a new physically informed deep learning model, named Double-stage ConvLSTM (D-ConvLSTM), to improve wave forecasts in the Atlantic Ocean. The waves in the next three consecutive days are predicted by feeding the deep learning model with the observed wave conditions in the preceding two days and the simultaneous ECMWF Reanalysis v5 (ERA5) wind forcing during the forecast period. The prediction skill of the d-ConvLSTM model was compared with that of two other forecasting methods—namely, the wave persistence forecast and the original ConvLSTM model. The results showed an increasing prediction error with the forecast lead time when the forecasts were evaluated using ERA5 reanalysis data. The d-ConvLSTM model outperformed the other two models in terms of wave prediction accuracy, with a root-mean-square error of lower than 0.4 m and an anomaly correlation coefficient skill of ∼0.80 at lead times of up to three days. In addition, a similar prediction was generated when the wind forcing was replaced by the IFS forecasted wind, suggesting that the d-ConvLSTM model is comparable to the Wave Model of European Centre for Medium-Range Weather Forecasts (ECMWF-WAM), but more economical and time-saving.摘要海浪预报对海上运输安全至关重要. 本研究提出了一种涵盖物理信息的深度学习模型Double-stage ConvLSTM (D-ConvLSTM) 以改进大西洋的海浪预报. 将D-ConvLSTM模型与海浪持续性预测和原始ConvLSTM模型的预测技巧进行对比. 结果表明, 预测误差随着预测时长的增加而增加. D-ConvLSTM模型在预测准确度方面优于前二者, 且第三天预测的均方根误差低于0.4 m, 距平相关系数约在0.8. 此外, 当使用IFS预测风替代再分析风时, 能够产生相似的预测效果. 这表明D-ConvLSTM模型的预测能力能够与ECMWF-WAM模式相当, 且更节省计算资源和时间.     

Impact of Subsurface Entrainment on ENSO Prediction: 1997-98 El Nino

Twenty-one-year hindcasts of sea surface temperature (SST) anomalies in the tropical Pacific were performed to validate the influence of ocean subsurface entrainment on SST prediction.A new hybrid coupled model was used that considered the entrainment of subsurface temperature anomalies into the sea surface.The results showed that predictions were improved significantly in the new coupled model.The predictive correlation skill increased by about 0.2 at a lead time of 9 months,and the root-mean-square (RMS) errors were decreased by nearly 0.2°C in general.A detailed analysis of the 1997-98 El Nio hindcast showed that the new model was able to predict the onset,peak (both time and amplitude),and decay of the 1997-98 strong El Nio event up to a lead time of one year,factors that are not represented well by many other forecast systems.This implies,in terms of prediction,that subsurface anomalies and their impact on the SST are one of the controlling factors in ENSO cycles.Improving the presentation of such effects in models would increase the forecast skill.     

典型相关分析(CCA)对我国冬季气温的短期气候预测试验

利用典型相关分析(CCA)方法建立统计气候预测模型，对我国冬季气温进行了预测试验，采用历史资料独立样本检验的方法，对预报技巧给出合理的评定。结果表明，使用CCA方法对我国冬季气温进行短期气候预测，有一定的预报技巧，对于特定地区和特定时期优选的因子场组合，可以取得较为满意的预报效果。大部分地区的季平均预报时效在2个季以内时，最佳预报相关系数在0.5以上。季平均的预报水平明显高于月平均的预报。海温场是所有因子场中最好的预报因子，不仅单独海温场的预报效果较好，而且与其他因子场组合后的预报水平还可以得到进一步提高。     

Impact of different ocean reanalyses on decadal climate prediction

We study the impact of three ocean state estimates (GECCO, SODA, [ECMWF]-ORA-S3) on decadal predictability in one particular forecast system, the Earth system model from the Max Planck Institute for Meteorology in Hamburg. The forecast procedure follows two steps. First, anomalies of temperature and salinity of the observational estimates are assimilated into our coupled model. Second, the assimilation runs are then used to initialize 10-year-long hindcasts/forecasts starting from each year between 1960 and 2001. The impact of the individual ocean state estimates is evaluated both by the extent to which climate variations from the ocean state estimates are adopted by the forecast system (‘fidelity’) and by the prediction skill of the corresponding hindcast experiments. The evaluation focuses on North Atlantic (NA) sea surface temperature (SST), upper-level (0–700?m) NA ocean heat content (OHC) and the Atlantic meridional overturning circulation (MOC). Regarding fidelity, correlations between observations and the assimilation runs are generally high for NA SST and NA OHC, except for NA OHC in the GECCO assimilation. MOC variations experience strong modifications when GECCO and SODA are assimilated, much less so when assimilating ORA-S3. Regarding prediction skill, when initializing with ORA-S3 and SODA, correlations with observations are high for NA OHC and moderate for NA SST. Correlations in case of GECCO, on the other hand, are high for NA SST and moderate for NA OHC. Relatively high MOC correlations between hindcasts and respective assimilation run appear in the first five years in GECCO in the tropics and subtropics and in ORA-S3 north of 50N. Correlations are largely reduced when the MOC signals are detrended. The trends in the assimilation runs are to some extent artifacts of the assimilation procedure. Hence, our potential predictabilities of the MOC are optimistic estimates of the upper limits of predictability. However, the ORA-S3 reanalysis gives the best results for our forecast system as measured by both overall fidelity of the assimilation procedure and predictions of upper-level OHC in the North Atlantic.     

The seasonal predictability of the Asian summer monsoon in a two-tiered forecast system

An extensive set of boreal summer seasonal hindcasts from a two tier system is compared with corresponding seasonal hindcasts from two other coupled ocean–atmosphere models for their seasonal prediction skill (for precipitation and surface temperature) of the Asian summer monsoon. The unique aspect of the two-tier system is that it is at relatively high resolution and the SST forcing is uniquely bias corrected from the multi-model averaged forecasted SST from the two coupled ocean–atmosphere models. Our analysis reveals: (a) The two-tier forecast system has seasonal prediction skill for precipitation that is comparable (over the Southeast Asian monsoon) or even higher (over the South Asian monsoon) than the coupled ocean–atmosphere. For seasonal anomalies of the surface temperature the results are more comparable across models, with all of them showing higher skill than that for precipitation. (b) Despite the improvement from the uncoupled AGCM all models in this study display a deterministic skill for seasonal precipitation anomalies over the Asian summer monsoon region to be weak. But there is useful probabilistic skill for tercile anomalies of precipitation and surface temperature that could be harvested from both the coupled and the uncoupled climate models. (c) Seasonal predictability of the South Asian summer monsoon (rainfall and temperature) does seem to stem from the remote ENSO forcing especially over the Indian monsoon region and the relatively weaker seasonal predictability in the Southeast Asian summer monsoon could be related to the comparatively weaker teleconnection with ENSO. The uncoupled AGCM with the bias corrected SST is able to leverage this teleconnection for improved seasonal prediction skill of the South Asian monsoon relative to the coupled models which display large systematic errors of the tropical SST’s.     

ECMWF seasonal forecast system 3 and its prediction of sea surface temperature

The latest operational version of the ECMWF seasonal forecasting system is described. It shows noticeably improved skill for sea surface temperature (SST) prediction compared with previous versions, particularly with respect to El Nino related variability. Substantial skill is shown for lead times up to 1?year, although at this range the spread in the ensemble forecast implies a loss of predictability large enough to account for most of the forecast error variance, suggesting only moderate scope for improving long range El Nino forecasts. At shorter ranges, particularly 3?C6?months, skill is still substantially below the model-estimated predictability limit. SST forecast skill is higher for more recent periods than earlier ones. Analysis shows that although various factors can affect scores in particular periods, the improvement from 1994 onwards seems to be robust, and is most plausibly due to improvements in the observing system made at that time. The improvement in forecast skill is most evident for 3-month forecasts starting in February, where predictions of NINO3.4 SST from 1994 to present have been almost without fault. It is argued that in situations where the impact of model error is small, the value of improved observational data can be seen most clearly. Significant skill is also shown in the equatorial Indian Ocean, although predictive skill in parts of the tropical Atlantic are relatively poor. SST forecast errors can be especially high in the Southern Ocean.     

Can local linear stochastic theory explain sea surface temperature and salinity variability?

 Sea surface temperature (SST) and salinity (SSS) time series from four ocean weather stations and data from an integration of the GFDL coupled ocean-atmosphere model are analyzed to test the applicability of local linear stochastic theory to the mixed-layer ocean. According to this theory, mixed-layer variability away from coasts and fronts can be explained as a ‘red noise’ response to the ‘white noise’ forcing by atmospheric disturbances. At one weather station, Papa (northeast Pacific), this stochastic theory can be applied to both salinity and temperature, explaining the relative redness of the SSS spectrum. Similar results hold for a model grid point adjacent to Papa, where the relationships between atmospheric energy and water fluxes and actual changes in SST and SSS are what is expected from local linear stochastic theory. At the other weather stations, this theory cannot adequately explain mixed-layer variability. Two oceanic processes must be taken into account: at Panulirus (near Bermuda), mososcale eddies enhance the observed variability at high frequencies. At Mike and India (North Atlantic), variations in SST and SSS advection, indicated by the coherence and equal persistence of SST and SSS anomalies, contribute to much of the low frequency variability in the model and observations. To achieve a global perspective, TOPEX altimeter data and model results are used to identify regions of the ocean where these mechanisms of variability are important. Where mesoscale eddies are as energetic as at Panulirus, indicated by the TOPEX global distribution of sea level variability, one would expect enhanced variability on short time scales. In regions exhibiting signatures of variability similar to Mike and India, variations in SST and SSS advection should dominate at low frequencies. According to the model, this mode of variability is found in the circumpolar ocean and the northern North Atlantic, where it is associated with the irregular oscillations of the model’s thermohaline circulation. Received: 11 March 1996 / Accepted: 6 September 1996     

An atmosphere–ocean regional climate model for the Mediterranean area: assessment of a present climate simulation

We present an atmosphere–ocean regional climate model for the Mediterranean basin, called the PROTHEUS system, composed by the regional climate model RegCM3 as the atmospheric component and by a regional configuration of the MITgcm model as the oceanic component. The model is applied to an area encompassing the Mediterranean Sea and compared to a stand-alone version of its atmospheric component. An assessment of the model performances is done by using available observational datasets. Despite a persistent bias, the PROTHEUS system is able to capture the inter-annual variability of seasonal sea surface temperature (SST) and also the fine scale spatio-temporal evolution of observed SST anomalies, with spatial correlation as high as 0.7 during summer. The close inspection of a 10-day strong wind event during the summer of 2000 proves the capability of the PROTHEUS system to correctly describe the daily evolution of SST under strong air–sea interaction conditions. As a consequence of the model’s skill in reproducing observed SST and wind fields, we expect a reliable estimation of air–sea fluxes. The model skill in reproducing climatological land surface fields is in line with that of state of the art regional climate models.