Multi-model calibration and combination of tropical seasonal sea surface temperature forecasts

Different combination methods based on multiple linear regression are explored to identify the conditions that lead to an improvement of seasonal forecast quality when individual operational dynamical systems and a statistical–empirical system are combined. A calibration of the post-processed output is included. The combination methods have been used to merge the ECMWF System 4, the NCEP CFSv2, the Météo-France System 3, and a simple statistical model based on SST lagged regression. The forecast quality was assessed from a deterministic and probabilistic point of view. SSTs averaged over three different tropical regions have been considered: the Niño3.4, the Subtropical Northern Atlantic and Western Tropical Indian SST indices. The forecast quality of these combinations is compared to the forecast quality of a simple multi-model (SMM) where all single models are equally weighted. The results show a large range of behaviours depending on the start date, target month and the index considered. Outperforming the SMM predictions is a difficult task for linear combination methods with the samples currently available in an operational context. The difficulty in the robust estimation of the weights due to the small samples available is one of the reasons that limit the potential benefit of the combination methods that assign unequal weights. However, these combination methods showed the capability to improve the forecast reliability and accuracy in a large proportion of cases. For example, the Forecast Assimilation method proved to be competitive against the SMM while the other combination methods outperformed the SMM when only a small number of forecast systems have skill. Therefore, the weighting does not outperform the SMM when the SMM is very skilful, but it reduces the risk of low skill situations that are found when several single forecast systems have a low skill.     

The reluctance of resource-users to adopt seasonal climate forecasts to enhance resilience to climate variability on the rangelands

As the impacts of climate-change on resource-dependent industries manifest, there is a commensurate effort to identify and implement strategies to reduce them. Yet, even when useful knowledge and tools exist, there can be poor adoption of adaptation strategies. We examine the reasons behind sub-optimal adoption of seasonal climate forecasts by graziers for managing climate variability. We surveyed 100 graziers in north-east Queensland, Australia and examined the influence of adaptive capacity, resource-dependency and forecast-perception on uptake. Technical perceptions were not important. Strategic skills, environmental awareness and social capital were. Results suggest that social factors (but not technical factors) are significant. These insights are important for adaptation planning and for maximising the resilience of communities and industries dependent on climate-sensitive resources.     

Sensitivity of seasonal climate forecasts to persisted SST anomalies

Most estimates of the skill of atmospheric general circulation models (AGCMs) for forecasting seasonal climate anomalies have been based on simulations with actual observed sea surface temperatures (SSTs) as lower boundary forcing. Similarly estimates of the climatological response characteristics of AGCMs used for seasonal-to-interannual climate prediction generally rest on historical simulations using "perfect" SST forecasts. This work examines the errors and biases introduced into the seasonal precipitation response of an AGCM forced with persisted SST anomalies, which are generally considered to constitute a good prediction of SST in the first three-month season. The added uncertainty introduced by the persisted SST anomalies weakens, and in some cases nullifies, the skill of atmospheric predictions that is possible given perfect SST forcing. The use of persisted SST anomalies also leads to changes in local signal-to-noise characteristics. Thus, it is argued that seasonal-to-interannual forecasts using AGCMs should be interpreted relative to historical runs that were subject to the same strategy of boundary forcing used in the current forecast in order to properly account for errors and biases introduced by the particular SST prediction strategy. Two case studies are examined to illustrate how the sensitivity of the climate response to predicted SSTs may be used as a diagnostic to suggest improvements to the predicted SSTs.     

Downscaling of the global seasonal forecasts of hydrometcenter of Russia for North Eurasia

Proposed is a method of downscaling of the global ensemble seasonal forecasts of air temperature computed using the SLAV model of the Hydrometcenter of Russia. The method is based on the regression and suggests a probabilistic interpretation of forecasts based on the assessment of uncertainty associated with the regression and model forecast ensemble spread. The verification of the method for 70 weather stations of North Eurasia using the rank probability skill score RPSS showed a significant advantage of downscaled forecasts over the forecasts interpolated from the model grid points. It is concluded that the use of the downscaling method is reasonable for the long-range forecasting of the station air temperature for North Eurasia.     

Impact of improved assimilation of temperature and salinity for coupled model seasonal forecasts

We assess the impact of improved ocean initial conditions for predicting El Niño-Southern Oscillation (ENSO) and Indian Ocean dipole (IOD) using the Bureau of Meteorology’s Predictive Ocean Atmosphere Model for Australia (POAMA) coupled seasonal prediction model for the period 1982–2006. The new ocean initial conditions are provided by an ensemble-based analysis system that assimilates subsurface temperatures and salinity and which is a clear improvement over the previous optimal interpolation system which used static error covariances and was univariate (temperature only). Hindcasts using the new ocean initial conditions have better skill at predicting sea surface temperature (SST) variations associated with ENSO than do the hindcasts initialized with the old ocean analyses. The improvement derives from better prediction of subsurface temperatures and the largest improvements come during ENSO–IOD neutral years. We show that improved prediction of the Niño3.4 SST index derives from improved initial depiction of the thermocline and halocline in the equatorial Pacific but as lead time increases the improved depiction of the initial salinity field in the western Pacific become more important. Improved ocean initial conditions do not translate into improved skill for predicting the IOD but we do see an improvement in the prediction of subsurface temperatures in the Indian Ocean (IO). This result reflects that the coupling between subsurface and surface temperature variations is weaker in the IO than in the Pacific, but coupled model errors may also be limiting predictive skill in the IO.     

Prediction of the Caspian Sea level using ECMWF seasonal forecasts and reanalysis

The hydrological budget of the Caspian Sea (CS) is investigated using the European Centre for Medium-Range Weather Forecasts interim reanalysis (ERAi) and seasonal forecast (FCST) data with the aim of predicting the Caspian Sea Level (CSL) some months ahead. Precipitation and evaporation are used. After precipitation events over the Volga River, the discharge (Volga River discharge (VRD)) follows with delays, which are parameterized. The components of the water budget from ERAi and FCSTs are integrated to obtain time series of the CSL. Observations of the CSL and the VRD are used for comparison and tuning. The quality of ERAi data is sufficiently good to calculate the time variability of the CSL with a satisfactory accuracy. Already the storage of water within the Volga Basin allows forecasts of the CSL a few months ahead, and using the FCSTs of precipitation improves the CSL forecasts. The evaporation in the seasonal forecasts is deficient due to unrealistic sea surface temperatures over the CS. Impacts of different water budget terms on the CSL variability are shown by a variety of validation tools. The importance of precipitation anomalies over the catchment of the Volga River is confirmed, but also impacts from the two southern rivers (Sefidrud and Kura River) and the evaporation over the CS become obvious for some periods. When pushing the FCSTs beyond the limits of the seasonal FCSTs to 1 year, considerable forecast skill can still be found. Validating only FCSTs by the present approach, which show the same trend as one based on a statistical method, significantly enhances the skill scores.     

ECMWF全风速场集合预报结果的偏差订正与预报不一致性分析

采用卡尔曼滤波类型自适应误差订正法和滑动自适应权重法,对2012年夏季ECMWF 10 m全风速场集合预报结果进行偏差订正,对订正前后的预报结果进行评估,并通过Jumpiness指数对预报结果订正前后的预报不一致性特征进行分析。结果表明,卡尔曼滤波类型自适应误差订正法能有效降低集合预报的均方根误差,且当起报时刻为00时对中低纬度地区的订正效果更显著,当起报时刻为12时对中高纬度地区的订正效果更明显;卡尔曼滤波类型自适应误差订正法能有效改善Talagrand的U型或L型分布;由均方根误差分析结果知道,ECMWF 10 m全风速场集合预报本身存在较大的预报不一致性,经过卡尔曼滤波类型自适应误差订正后,集合预报的预报不一致性明显降低,偏差订正可有效改善集合预报的预报不一致性,且随着预报时效的延长,卡尔曼滤波误差法对预报不一致性的改善效果更加明显;从预报不一致性的发生次数特征来看,单点跳跃出现的次数最多,异号三点跳跃的次数最少;经过卡尔曼滤波类型自适应误差订正后,单点跳跃、异号两点跳跃、异号三点跳跃次数都有所下降。     

江淮流域夏季持续性强降水的低频特征分析

基于1986-2015年湖南逐日降水资料、同期美国气象环境预报中心（NCEP）和美国国家大气研究中心（NCAR）再分析资料,通过分析强低频振荡年的汛期强降水特征和低频环流场演变对强降水的影响,建立了湖南省汛期延伸期强降水过程预报指数。结果表明：（1）汛期33%的强降水过程均发生在具有显著30~60 d低频振荡的年份中,且大多位于低频降水峰值阶段。（2）通过对强低频振荡年进行合成发现,在活跃位相,南亚高压偏强偏东,副热带高压偏西偏强,这种环流配置导致中国南方大部分地区的高层环流为辐散,底层环流为辐合,有利于降水的产生。在中断位相,南亚高压呈东西带状分布且其位置偏西、强度偏弱,副热带高压偏东偏弱,使得向湖南地区输送水汽的西南气流减弱,进入降水中断期。（3）基于低频散度场不同位相的变化特征,选取了与低频降水相关的两个关键区,从而建立延伸期预报指数,该指数对低频降水显著年的强降水回报准确率能够达到73%。（4）前期4月黑潮的海温异常（SSTA）可作为湖南省强低频振荡年的预测指标。

     

Prediction skill of the Madden and Julian Oscillation in dynamical extended range forecasts

The Madden and Julian Oscillation (MJO) is the most prominent mode of intraseasonal variations in the tropical region. It plays an important role in climate variability and has a significant influence on medium-to-extended ranges weather forecasting in the tropics. This study examines the forecast skill of the oscillation in a set of recent dynamical extended range forecasts (DERF) experiments performed by the National Centers for Environmental Prediction (NCEP). The present DERF experiments were done with the reanalysis version of the medium range forecast (MRF) model and include 50-day forecasts, initialized once-a-day (0Z) with reanalyses fields, for the period between 1 January, 1985, and 31 December, 1989. The MRF model shows large mean errors in representing intraseasonal variations of the large-scale circulation, especially over the equatorial eastern Pacific Ocean. A diagnostic analysis has considered the different phases of the MJO and the associated forecast skill of the MRF model. Anomaly correlations on the order of 0.3 to 0.4 indicate that skillful forecasts extend out to 5 to 7 days lead-time. Furthermore, the results show a slight increase in the forecast skill for periods when convective anomalies associated with the MJO are intense. By removing the mean errors, the analysis shows systematic errors in the representation of the MJO with weaker than observed upper level zonal circulations. The examination of the climate run of the MRF model shows the existence of an intraseasonal oscillation, although less intense (50–70%) and with faster (nearly twice as fast) eastward propagation than the observed MJO. The results indicate that the MRF model likely has difficulty maintaining the MJO, which impacts its forecast. A discussion of future work to improve the representation of the MJO in dynamical models and assess its prediction is presented. Received: 28 December 1998 / Accepted: 27 September 1999     

Improved seasonal climate forecasts for the Caribbean region using the Florida State University Synthetic Superensemble

Summary Climate variations in the Caribbean, largely manifest in rainfall activity, have important consequences for the large-scale water budget, natural vegetation, and land use in the region. The wet and dry seasons will be defined, and the important roles played by the El Ni?o-Southern Oscillation (ENSO) and the North Atlantic Oscillation (NAO) in modulating the rainfall during these seasons will be discussed. The seasonal climate forecasts in this paper are made by 13 state of the art coupled atmosphere-ocean general circulation models (CGCMs) and by the Florida State University Synthetic Superensemble (FSUSSE), whose forecasts are obtained by a weighted combination of the individual CGCM forecasts based on a training period. The success of the models in simulating the observed 1989–2001 climatology of the various forecast parameters will be examined and linked to the models’ success in predicting the seasonal climate for individual years. Seasonal forecasts will be examined for precipitation, sea-surface temperature (SST), 2-meter air temperature, and 850 hPa u- and v-wind components during the period 1989–2001. Evaluation metrics include root mean square (RMS) error and Brier skill score. It will be shown that the FSUSSE is superior to the individual CGCMs and their ensemble mean both in simulating the 1989–2001 climatology for the various parameters and in predicting the seasonal climate of the various parameters for individual years. The seasonal climate forecasts of the FSUSSE and of the ensemble mean of the 13 state of the art CGCMs will be evaluated for years (during the period 1989–2001) that have particular ENSO and NAO signals that are known to influence Caribbean weather, particularly the rainfall. It will be shown that the FSUSSE provides superior forecasts of rainfall, SST, 2-meter air temperature, and 850 hPa u- and v-wind components during dry summers that are modulated by negative SOI and/or positive NAO indices. Such summers have become a feature of a twenty-year pattern of drought in the Caribbean region. The results presented in this paper will show that the FSUSSE is a valuable tool for forecasting rainfall and other atmospheric and oceanic variables during such periods of drought.     

Evaluation of the ENSEMBLES multi-model seasonal forecasts of Indian summer monsoon variability

The performance of the new multi-model seasonal prediction system developed in the frame work of the ENSEMBLES EU project for the seasonal forecasts of India summer monsoon variability is compared with the results from the previous EU project, DEMETER. We have considered the results of six participating ocean-atmosphere coupled models with 9 ensemble members each for the common period of 1960–2005 with May initial conditions. The ENSEMBLES multi-model ensemble (MME) results show systematic biases in the representation of mean monsoon seasonal rainfall over the Indian region, which are similar to that of DEMETER. The ENSEMBLES coupled models are characterized by an excessive oceanic forcing on the atmosphere over the equatorial Indian Ocean. The skill of the seasonal forecasts of Indian summer monsoon rainfall by the ENSEMBLES MME has however improved significantly compared to the DEMETER MME. Its performance in the drought years like 1972, 1974, 1982 and the excess year of 1961 was in particular better than the DEMETER MME. The ENSEMBLES MME could not capture the recent weakening of the ENSO-Indian monsoon relationship resulting in a decrease in the prediction skill compared to the “perfect model” skill during the recent years. The ENSEMBLES MME however correctly captures the north Atlantic-Indian monsoon teleconnections, which are independent of ENSO.     

Evaluation of forecast strategies for seasonal and decadal forecasts in presence of systematic model errors

This study discusses and compares three different strategies used to deal with model error in seasonal and decadal forecasts. The strategies discussed are the so-called full initialisation, anomaly initialisation and flux correction. In the full initialisation the coupled model is initialised to a state close to the real-world attractor and after initialisation the model drifts towards its own attractor, giving rise to model bias. The anomaly initialisation aims to initialise the model close to its own attractor, by initialising only the anomalies. The flux correction strategy aims to keep the model trajectory close to the real-world attractor by adding empirical corrections. These three strategies have been implemented in the ECMWF coupled model, and are evaluated at seasonal and decadal time-scales. The practical implications of the different strategies are also discussed. Results show that full initialisation results in a clear model drift towards a colder climate. The anomaly initialisation is able to reduce the drift, by initialising around the model mean state. However, the erroneous model mean state results in degraded seasonal forecast skill. The best results on the seasonal time-scale are obtained using momentum-flux correction, mainly because it avoids the positive feedback responsible for a strong cold bias in the tropical Pacific. It is likely that these results are model dependent: the coupled model used here shows a strong cold bias in the Central Pacific, resulting from a positive coupled feedback between winds and SST. At decadal time-scales it is difficult to determine whether any of the strategies is superior to the others.     

Monthly and seasonal numerical forecasts by using the anomaly ocean-atmosphere coupled filtered model

Case experiments of monthly predictions of eight winter months during 1976－1977 and 1982－1983 El Nino events are performed by using a three-layer anomalous filtered model (AFM), in which transient Rossby waves are filtered. The results show that this model predicts successfully the large-scale patterns of the monthly mean surface temperature anomalies. The correlation coefficients between observations and predic-tions are basically higher than those of persistence predictions. By comparison with the anomalous general circulation model (AGCM) the AFM gives almost the identical results, but the computation time required for running the AFM is nearly 100 times less than that required for running the AGCM. It is also shown that the results of the three-layer model are better than those of the one-layer model. In the meanwhile, four seasonal forecasts are also earned out by using the same model. It seems that the AFM possesses potential ability in predicting large-scale circulation anomalies.On the basis of the works done during the last decade and of the results in this paper, the predicting ability of the AFM is summed up in the last section.     

Statistical adjustment,calibration and downscaling of seasonal forecasts: a case-study for Southeast Asia

Climate Dynamics - The present paper is a follow-on of the work presented in Manzanas et al. (Clim Dyn 53(3–4):1287–1305, 2019) which provides a comprehensive intercomparison...     

基于频率匹配法的中国降水多模式预报订正研究

基于TIGGE资料中欧洲中期天气预报中心、日本气象厅、美国国家环境预报中心及英国气象局1～7 d日降水量预报以及中国自动站观测资料与CMORPH降水产品融合的逐时降水量网格数据集,利用频率匹配法(Frequency-Matching Method,FMM)对中国降水预报进行客观订正。首先利用卡尔曼滤波方法对降水频率进行调整,并根据不同区域降水强度差异将全国分为7个子区域分别进行频率匹配。结果表明,FMM可以有效减小降水量预报的误差。经过频率匹配法订正后各模式降水预报的平均绝对误差(Mean Absolute Error,MAE)大幅减小,且订正后各量级降水的雨区面积更加接近实际观测值。FMM对小于5 mm和大于15 mm的降水预报技巧改进明显。此外,FMM降低了模式预报的小雨空报率和大雨漏报率,并且明显提高了晴雨预报的准确率。FMM明显消除了大范围小雨空报区域,但是对强降水预报FMM仅能调整降水量大小,强降水落区预报并不能得到明显改善。     

Forecast skill of the tropical intraseasonal oscillation in the NCEP GFS dynamical extended range forecasts

This study examines the forecast performance of tropical intraseasonal oscillation (ISO) in recent dynamical extended range forecast (DERF) experiments conducted with the National Centers for Environmental Prediction (NCEP) Global Forecasting System (GFS) model. The present study extends earlier work by comparing prediction skill of the northern winter ISO (Madden-Julian Oscillation) between the current and earlier experiments. Prediction skill for the northern summer ISO is also investigated. Since the boreal summer ISO exhibits northward propagation as well as eastward propagation along the equator, forecast skill for both components is computed. For the 5-year period from 1 January, 1998 through 31 December, 2002, 30-day forecasts were made once a day. Compared to the previous DERF experiment, the current model has shown some improvements in forecasting the ISO during winter season so that the skillful forecasts (anomaly correlation>0.6) for upper-level zonal wind anomaly extend from the previous shorter-than 5 days out to 7 days lead-time. A similar level of skill is seen for both northward and eastward propagation components during the summer season as in the winter case. Results also show that forecasts from extreme initial states are more skillful than those from null phases for both seasons, extending the skillful range by 3–6 days. For strong ISO convection phases, the GFS model performs better during the summer season than during the winter season. In summer forecasts, large-scale circulation and convection anomalies exhibit northward propagation during the peak phase. In contrast, the GFS model still has difficulties in sustaining ISO variability during the northern winter as in the previous DERF run. That is, the forecast does not maintain the observed eastward propagating signals associated with large-scale circulation; rather the forecast anomalies appear to be stationary at their initial location and decay with time. The NCEP Coupled Forecast System produces daily operational forecasts and its predication skill of the MJO will be reported in the future.     

Long-range forecasting of forest fire danger based on the SLAV model seasonal ensemble forecasts

Developed is a scheme of the long-range forest fire danger (FD) forecasting based on the computation of the Nesterov??s index and on hydrodynamic forecasts of air temperature and precipitation of the SLAV model. The satellite information on the climatic distribution of the snow cover was taken into account for the forecast issuance. It is proposed to express the FD index not using traditional classes of fire danger but in ??above the norm,?? ??norm,?? and ??below the norm?? gradations. Presented are the results of author??s trials of the method based on the retrospective model data for six years. Discussed are the approaches to the improvement of the method of long-range fire danger forecasting.