Superensemble Technique Research Articles

Superensemble Approach to S2S Model for Predicting Surface Air Temperature in Summer in East Asia from 2016 to 2020

A multimodel superensemble technique was applied to improve the accuracy of a model predicting the subseasonal surface air temperatures in East Asia. Superensemble forecast data based on seven subseasonal-to-seasonal (S2S) forecast models were produced for the initial times from May to August in 2016–2020 and forecast days in weeks 1–4. Superensemble techniques based on the root-mean-square error ensemble mean (SUPR) and the linear regression coefficient ensemble mean (SUPL) were evaluated and compared with the forecast bias-removed ensemble mean (ENS) data. The magnitudes of the forecasting root-mean-square errors (RMSE) of the SUPR and SUPL methods were 4.2% and 13.8% lower, respectively, than that of the ENS method, and the most significant effect was in 2020. The RMSE improvements of SUPR and SUPL were high when the initial date was in May, June, and August, and the forecast date was within weeks 1–4. Applying multimodel superensemble methods effectively reduced the forecasting errors of the S2S models. If weight coefficients based on the characteristics of the models are appropriately applied, the forecasting performance can be significantly improved.

Application of Multimodel Superensemble Technique on the TIGGE Suite of Operational Models

One widely recognized portal which provides numerical weather prediction forecasts is “The Observing System Research and Predictability Experiment” (THORPEX) Interactive Grand Global Ensemble (TIGGE), an initiative of WMO project. This data portal provides forecasts from 1 to 16 days (2 weeks in advance) for many variables such as rainfall, winds, geopotential height, temperature, and relative humidity. These weather forecasting centers have delivered near-real-time (with a delay of 48 hours) ensemble prediction system data to three TIGGE data archives since October 2006. This study is based on six years (2008–2013) of daily rainfall data by utilizing output from six centers, namely the European Centre for Medium-Range Weather Forecasts, the National Centre for Environmental Prediction, the Center for Weather Forecast and Climatic Studies, the China Meteorological Agency, the Canadian Meteorological Centre, and the United Kingdom Meteorological Office, and make consensus forecasts of up to 10 days lead time by utilizing the multimodal multilinear regression technique. The prediction is made over the Indian subcontinent, including the Indian Ocean. TRMM3B42 daily rainfall is used as the benchmark to construct the multimodel superensemble (SE) rainfall forecasts. Based on statistical ability ratings, the SE offers a better near-real-time forecast than any single model. On the one hand, the model from the European Centre for Medium-Range Weather Forecasting and the UK Met Office does this more reliably over the Indian domain. In a case of Indian monsoon onset, 05 June 2014, SE carries the lowest RMSE of 8.5 mm and highest correlation of 0.49 among six member models. Overall, the performance of SE remains better than any individual member model from day 1 to day 10.

Subseasonal Forecast of Surface Air Temperature Using Superensemble Approaches: Experiments over Northeast Asia for 2018

AbstractBridging the gap between weather forecasting and climate prediction, subseasonal to seasonal (S2S) forecasts are of great importance yet currently of relatively poor quality. Using the S2S Prediction Project database, the study evaluates products derived from four operational centers of CMA, KMA, NCEP, and UKMO, and superensemble experiments including the straightforward ensemble mean (EMN), bias-removed ensemble mean (BREM), error-based superensemble (ESUP), and Kalman filter superensemble (KF), in forecasts of surface air temperature with lead times of 6–30 days over northeast Asia in 2018. Validations after the preprocessing of a 5-day running mean suggest that the KMA model shows the highest skill for either the control run or the ensemble mean. The nonequal weighted ESUP is slightly superior to BREM, whereas they both show larger biases than EMN after a lead time of 22 days. The KF forecast constantly outperforms the others, decreasing mean absolute errors by 0.2°–0.5°C relative to EMN. Forecast experiments of the 2018 northeast Asia heat wave reveal that the superensembles remarkably improve the raw forecasts featuring biases of >4°C. The prominent advancement of KF is further confirmed, showing the regionally averaged bias of ≤2°C and the hit rate of 2°C reaching up to 60% at a lead time of 22 days. The superensemble techniques, particularly the KF method of dynamically adjusting the weights in accordance with the latest information available, are capable of improving forecasts of spatiotemporal patterns of surface air temperature on the subseasonal time scale, which could extend the skillful prediction lead time of extreme events such as heat waves to about 3 weeks.

Modification of a linear regression-based multi-model super-ensemble technique and its application in forecasting of wave height during extreme weather conditions

ABSTRACTIn this study, we focus on the improvement of wave forecast of the Indian coastal region using a multi-model ensemble technique. Generally, a number of wave forecast are available for the same region from different wave models. The main objective of this study is to merge the wave forecasts available at Indian National Centre for Ocean Information Services from different wave models to obtain an improved wave forecast using a multi-model super-ensemble method [Krishnamurti et al. 1999. Improved weather and seasonal climate forecasts from multi-model super-ensemble. Science. 285:1548–1550] during extreme weather conditions and to modify Krishnamurthy’s techniques and validate with observations for a better prediction. Here, Multi-grid WAVEWATCH III, Simulating WAves Nearshore and MIKE 21 Spectral Waves are used for the generation of wave forecast. We propose a modification of Krishnamurthy’s linear regression-based ensemble model. By using both of these ensemble techniques, we perform a multi-model ensemble forecasting of significant wave height up to 24-h lead time in the Indian Ocean for three different cyclones (Nilofar, Hudhud and Phailin) and during the southwest monsoon. A comparison of ensemble predictions and individual model predictions with the actual observations showed generally satisfactory performance of the chosen tools. At the time of severe cyclones such as Hudhud and Phailin, our modified technique shows significantly better prediction than the linear regression-based ensemble technique.

Development of super-ensemble techniques for ocean analyses: the Mediterranean Sea case

Abstract. A super-ensemble methodology is proposed to improve the quality of short-term ocean analyses for sea surface temperature (SST) in the Mediterranean Sea. The methodology consists of a multiple linear regression technique applied to a multi-physics multi-model super-ensemble (MMSE) data set. This is a collection of different operational forecasting analyses together with ad hoc simulations, created by modifying selected numerical model parameterizations. A new linear regression algorithm based on empirical orthogonal function filtering techniques is shown to be efficient in preventing overfitting problems, although the best performance is achieved when a simple spatial filter is applied after the linear regression. Our results show that the MMSE methodology improves the ocean analysis SST estimates with respect to the best ensemble member (BEM) and that the performance is dependent on the selection of an unbiased operator and the length of training. The quality of the MMSE data set has the largest impact on the MMSE analysis root mean square error (RMSE) evaluated with respect to observed satellite SST. The MMSE analysis estimates are also affected by training period length, with the longest period leading to the smoothest estimates. Finally, lower RMSE analysis estimates result from the following: a 15-day training period, an overconfident MMSE data set (a subset with the higher-quality ensemble members) and the least-squares algorithm being filtered a posteriori.

Use of APHRODITE Rain Gauge–Based Precipitation and TRMM 3B43 Products for Improving Asian Monsoon Seasonal Precipitation Forecasts by the Superensemble Method

Abstract A multimodel superensemble developed by the Florida State University combines multiple model forecasts based on their past performance (training phase) to make a consensus forecast. Because observed precipitation reflects local characteristics such as orography, quantitative high-resolution precipitation products are useful for downscaling coarse model outputs. The Asian Precipitation–Highly-Resolved Observational Data Integration Toward Evaluation of Water Resources (APHRODITE) and Tropical Rainfall Measuring Mission (TRMM) 3B43 products are used for downscaling and as training data in the superensemble training phase. Seven years (1998–2004) of monthly precipitation (June–August) over the Asian monsoon region (0°–50°N, 60°–150°E) and results of four coupled climate models were used. TRMM 3B43 was adjusted by APHRODITE (m-TRMM). For seasonal climate forecasts, a synthetic superensemble technique was used. A cross-validation technique was adopted, in which the year to be forecast was excluded from the calculations for obtaining the regression coefficients. The principal results are as follows: 1) Seasonal forecasts of Asian monsoon precipitation were considerably improved by use of APHRODITE rain gauge–based data or the m-TRMM product. These forecasts are much superior to those from the best model of the suite and ensemble mean. 2) Use of a statistical downscaling and synthetic superensemble method for multimodel forecasts of seasonal climate significantly improved precipitation prediction at higher resolution. This is confirmed by cross-evaluation of superensemble with using other observation data than the data used in the training phase. 3) Availability of a dense rain gauge network–based analysis was essential for the success of this work.

Regional climate models downscaling in the Alpine area with multimodel superensemble

Abstract. The climatic scenarios show a strong signal of warming in the Alpine area already for the mid-XXI century. The climate simulations, however, even when obtained with regional climate models (RCMs), are affected by strong errors when compared with observations, due both to their difficulties in representing the complex orography of the Alps and to limitations in their physical parametrization. Therefore, the aim of this work is to reduce these model biases by using a specific post processing statistic technique, in order to obtain a more suitable projection of climate change scenarios in the Alpine area. For our purposes we used a selection of regional climate models (RCMs) runs which were developed in the framework of the ENSEMBLES project. They were carefully chosen with the aim to maximise the variety of leading global climate models and of the RCMs themselves, calculated on the SRES scenario A1B. The reference observations for the greater Alpine area were extracted from the European dataset E-OBS (produced by the ENSEMBLES project), which have an available resolution of 25 km. For the study area of Piedmont daily temperature and precipitation observations (covering the period from 1957 to the present) were carefully gridded on a 14 km grid over Piedmont region through the use of an optimal interpolation technique. Hence, we applied the multimodel superensemble technique to temperature fields, reducing the high biases of RCMs temperature field compared to observations in the control period. We also proposed the application of a brand new probabilistic multimodel superensemble dressing technique, already applied to weather forecast models successfully, to RCMS: the aim was to estimate precipitation fields, with careful description of precipitation probability density functions conditioned to the model outputs. This technique allowed for reducing the strong precipitation overestimation, arising from the use of RCMs, over the Alpine chain and to reproduce well the monthly behaviour of precipitation in the control period.

Projection of fire potential to future climate scenarios in the Alpine area: some methodological considerations

In Europe, wildfires are an issue not only for the Mediterranean area, but also in the Alpine regions in terms of increasing number of events and severity. In this study we evaluate the impact of climate change on the fire potential in the Alps in the past and in future scenarios. The Fine Fuel Moisture Code (FFMC) of the Canadian Forest Fire Danger Rating System, which successfully distinguishes among recorded fire/no fire events, is applied to projections of Regional Climate Models (RCMs) calculated on the SRES scenario A1B. We compare two different techniques: 1) a single model run of the COSMO-CLM RCM at 18 km resolution, and 2) a combination of 25-km resolution RCMs from the ENSEMBLES project, combined with the Multimodel SuperEnsemble technique and a new probabilistic Multimodel SuperEnsemble Dressing. The single-model RCM allows for a greater coherence among the input parameters, while the Multimodel techniques permit to reduce the model biases and to downscale to a higher resolution where long term records of observations are available. The projected changes with the Multimodel in the scenario give an estimation of increasing wildfire potential in the mid XXI century. In particular the frequency of severe wildfire potential days is shown to increase dramatically. The single (independent) COSMO model gives a weaker signal and in some regions of the study area the predicted changes are opposite to the ones by the Multimodel. This is mainly due to increasing precipitation amounts simulated especially in the northern parts of the Alps. However, there are also some individual models included in the Multimodel ensemble that show a similar signal. This confirms the ambiguity of any impact study based on a single climate model due to the uncertainty of the projections of the climate models.

Multiphysics superensemble forecast applied to Mediterranean heavy precipitation situations

Abstract. The high-impact precipitation events that regularly affect the western Mediterranean coastal regions are still difficult to predict with the current prediction systems. Bearing this in mind, this paper focuses on the superensemble technique applied to the precipitation field. Encouraged by the skill shown by a previous multiphysics ensemble prediction system applied to western Mediterranean precipitation events, the superensemble is fed with this ensemble. The training phase of the superensemble contributes to the actual forecast with weights obtained by comparing the past performance of the ensemble members and the corresponding observed states. The non-hydrostatic MM5 mesoscale model is used to run the multiphysics ensemble. Simulations are performed with a 22.5 km resolution domain (Domain 1 in http://mm5forecasts.uib.es) nested in the ECMWF forecast fields. The period between September and December 2001 is used to train the superensemble and a collection of 19~MEDEX cyclones is used to test it. The verification procedure involves testing the superensemble performance and comparing it with that of the poor-man and bias-corrected ensemble mean and the multiphysic EPS control member. The results emphasize the need of a well-behaved training phase to obtain good results with the superensemble technique. A strategy to obtain this improved training phase is already outlined.

Super-ensemble techniques applied to wave forecast: performance and limitations

Abstract. Nowadays, several operational ocean wave forecasts are available for a same region. These predictions may considerably differ, and to choose the best one is generally a difficult task. The super-ensemble approach, which consists in merging different forecasts and past observations into a single multi-model prediction system, is evaluated in this study. During the DART06 campaigns organized by the NATO Undersea Research Centre, four wave forecasting systems were simultaneously run in the Adriatic Sea, and significant wave height was measured at six stations as well as along the tracks of two remote sensors. This effort provided the necessary data set to compare the skills of various multi-model combination techniques. Our results indicate that a super-ensemble based on the Kalman Filter improves the forecast skills: The bias during both the hindcast and forecast periods is reduced, and the correlation coefficient is similar to that of the best individual model. The spatial extrapolation of local results is not straightforward and requires further investigation to be properly implemented.

Can a Multimodel SuperEnsemble technique be used for precipitation forecasts?

Abstract. The Multimodel SuperEnsemble technique is a postprocessing method for the estimation of weather forecast parameters reducing direct model output errors. It differs from other ensemble analysis techniques by the use of an adequate weighting of the input forecast models in order to obtain a combined estimation of meteorological parameters. Weights are calculated by least-square minimization of the differences between the model and the observed field during a so-called training period. Although it can be applied successfully on continuous parameters like temperature, relative humidity, wind speed and mean sea level pressure, the Multimodel SuperEnsemble also gives good results when applied on the precipitation, a parameter quite difficult to handle with standard post-processing methods. Here we present a methodology for the Multimodel precipitation forecasts with a careful ensemble dressing via the precipitation PDF estimation.

Multimodel SuperEnsemble technique for quantitative precipitation forecasts in Piemonte region

Abstract. The Multimodel SuperEnsemble technique is a powerful post-processing method for the estimation of weather forecast parameters reducing direct model output errors. It has been applied to real time NWP, TRMM-SSM/I based multi-analysis, Seasonal Climate Forecasts and Hurricane Forecasts. The novelty of this approach lies in the methodology, which differs from ensemble analysis techniques used elsewhere. Several model outputs are put together with adequate weights to obtain a combined estimation of meteorological parameters. Weights are calculated by least-square minimization of the difference between the model and the observed field during a so-called training period. Although it can be applied successfully on the continuous parameters like temperature, humidity, wind speed and mean sea level pressure, the Multimodel SuperEnsemble gives good results also when applied on the precipitation, a parameter quite difficult to handle with standard post-processing methods. Here we present our methodology for the Multimodel precipitation forecasts, involving a new accurate statistical method for bias correction and a wide spectrum of results over Piemonte very dense non-GTS weather station network.

Super-ensemble techniques: Application to surface drift prediction

The prediction of surface drift of floating objects is an important task, with applications such as marine transport, pollutant dispersion, and search-and-rescue activities. But forecasting even the drift of surface waters is very challenging, because it depends on complex interactions of currents driven by the wind, the wave field and the general prevailing circulation. Furthermore, although each of those can be forecasted by deterministic models, the latter all suffer from limitations, resulting in imperfect predictions. In the present study, we try and predict the drift of two buoys launched during the DART06 (Dynamics of the Adriatic sea in Real-Time 2006) and MREA07 (Maritime Rapid Environmental Assessment 2007) sea trials, using the so-called hyper-ensemble technique: different models are combined in order to minimize departure from independent observations during a training period; the obtained combination is then used in forecasting mode. We review and try out different hyper-ensemble techniques, such as the simple ensemble mean, least-squares weighted linear combinations, and techniques based on data assimilation, which dynamically update the model’s weights in the combination when new observations become available. We show that the latter methods alleviate the need of fixing the training length a priori, as older information is automatically discarded. When the forecast period is relatively short (12 h), the discussed methods lead to much smaller forecasting errors compared with individual models (at least three times smaller), with the dynamic methods leading to the best results. When many models are available, errors can be further reduced by removing colinearities between them by performing a principal component analysis. At the same time, this reduces the amount of weights to be determined. In complex environments when meso- and smaller scale eddy activity is strong, such as the Ligurian Sea, the skill of individual models may vary over time periods smaller than the forecasting period (e.g. when the latter is 36 h). In these cases, a simpler method such as a fixed linear combination or a simple ensemble mean may lead to the smallest forecast errors. In environments where surface currents have strong mean-kinetic energies (e.g. the Western Adriatic Current), dynamic methods can be particularly successful in predicting the drift of surface waters. In any case, the dynamic hyper-ensemble methods allow to estimate a characteristic time during which the model weights are more or less stable, which allows predicting how long the obtained combination will be valid in forecasting mode, and hence to choose which hyper-ensemble method one should use.

Improved ocean prediction skill and reduced uncertainty in the coastal region from multi-model super-ensembles

The use of Multi-model Super-Ensembles (SE) which optimally combine different models, has been shown to significantly improve atmospheric weather and climate predictions. In the highly dynamic coastal ocean, the presence of small-scales processes, the lack of real-time data, and the limited skill of operational models at the meso-scale have so far limited the application of SE methods. Here, we report results from state-of-the-art super-ensemble techniques in which SEPTR (a trawl-resistant bottom mounted instrument platform transmitting data in near real-time) temperature profile data are combined with outputs from eight ocean models run in a coastal area during the Dynamics of the Adriatic in Real-Time (DART) experiment in 2006. New Kalman filter and particle filter based SE methods, which allow for dynamic evolution of weights and associated uncertainty, are compared to standard SE techniques and numerical models. Results show that dynamic SE are able to significantly improve prediction skill. In particular, the particle filter SE copes with non-Gaussian error statistics and provides robust and reduced uncertainty estimates.

Warm Season Mesoscale Superensemble Precipitation Forecasts in the Southeastern United States

Abstract With current computational limitations, the accuracy of high-resolution precipitation forecasts has limited temporal and spatial resolutions. However, with the recent development of the superensemble technique, the potential to improve precipitation forecasts at the regional resolution exists. The purpose of this study is to apply the superensemble technique to regional precipitation forecasts to generate more accurate forecasts pinpointing exact locations and intensities of strong precipitation systems. This study will determine the skill of a regional superensemble forecast out to 60 h by examining its equitable threat score and its false alarm ratio. The regional superensemble consists of 12–60-h daily quantitative precipitation forecasts from six models. Five are independent operational models, and one comes from the physically initialized Florida State University regional spectral model. The superensemble forecasts are verified during the summer 2003 season over the southeastern United States using merged River Forecast Center stage-IV radar–gauge and satellite analyses. Precipitation forecasts were skillful in outperforming the operational models at all model times. Precipitation results were stratified by time of day to allow detections of the diurnal cycle. As expected, warm season daytime precipitation is commonly forced by convection, which is difficult to accurately model. Major synoptic regimes, including subtropical highs, midlatitude troughs/fronts, and tropical cyclones, were examined to determine the skill of the superensemble under various synoptic conditions.

Multimodel based superensemble forecasts for short and medium range NWP over various regions of Africa

This study examines the predictability of weather over several regions in Africa using a multimodel superensemble technique developed at the Florida State University, which is an objective means of combining daily forecasts from multilevel global models. It is referred to as FSUSE and up to 7 different models are used to construct the superensemble. The benchmark reanalysis fields used are the precipitation data sets from CMORPH and all other global fields from ECMWF daily operational analysis. The FSUSE works by using multiple linear regression to derive weights from a comparison of each member model forecast to the benchmark analysis during a training period of the most recent 120 days, and these weights are passed to the forecast phase. This procedure removes the bias of each model and allows for an optimal linear combination of the individual model forecasts by taking account of the relative skill of each model to give a consensus forecast that is superior to the ensemble mean and all the members.

Verification of the motion fields over Southeast Asia by the Florida State University (FSU) superensemble forecast

The Florida State University (FSU) multimodel superensemble forecast is evaluated against several other operational weather models for the Southeast Asia region. The superensemble technique has demonstrated its exceptional skills in forecasting precipitation, motion and mass fields compared to either individual global operational or ensemble mean forecasts. The motion field investigation for the season of 2001 reveals that the superensemble forecasts are closer to the observed data compared to the other global member operational models through its low systematic errors at the 850 hPa level. The FSU multimodel superensemble forecasts exhibit the lowest root mean square errors (RSMEs), the highest correlation against the best observed data and the lowest systematic errors compared to the other operational model members. These forecasts have the potential to provide better daily weather predictions over the Southeast Asia region, particularly during the early northeast monsoon that often causes heavy rainfall in the equatorial part of the Southeast Asia region.

Weather forecasts obtained with a Multimodel SuperEnsemble technique in a complex orography region

The Multimodel SuperEnsemble technique (KRISHNAMURTI et al., 2000a) is a new powerful post-processing method for the estimation of weather forecast parameters. Several model outputs are combined, using weights calculated during a training period. Piedmont region is characterised by complex mountainous orography and direct model outputs, even from high-resolution limited area models, show many strong systematic and random errors in the forecast, compared to the values observed by our high-density non-GTS network. This is one of the first applications of this technique in a narrow mountain area and combines both global and limited-area models. Our results show a good improvement of meteorological parameter forecasts such as temperature, humidity, wind speed and precipitation.

Reduction of forecast error for global numerical weather prediction by The Florida State University (FSU) Superensemble

The skill of the FSU Superensemble technique as applied to global numerical weather prediction is evaluated extensively in this paper. The global mass and motion fields for year 2000 and precipitation over the domain 55 S to 55 N for year 2001, as predicted by the Superensemble, the ensemble member models, and the mean of the ensemble members, are evaluated by standard statistical measures of skill to determine the performance of the Superensemble in relation to the other models. The member models are global forecast models from 5 of the world’s operational forecast centers in addition to the FSU global spectral model. For precipitation 5 additional versions of the FSU global model are utilized in the ensemble, as defined by different initial conditions provided by various physical initialization algorithms. Statistical parameters calculated for the mass and motion fields include root mean square (RMS) error, systematic error (or bias), and anomaly correlation. These are applied to the mean sea level pressure, 500 hPa heights, and the wind fields at 850 hPa and 200 hPa. Statistical parameters that were calculated for precipitation include RMS error, correlation, equitable threat score (ETS), and a special definition of bias appropriate for the precipitation field. For the mass and motion fields the performance of the Superensemble was considered for the annual global case, as well as for each hemisphere (north and south) and for each of the four seasons. For precipitation only the annual case was considered over the domain cited above.

Skill of a superensemble forecast over equatorial Southeast Asia

AbstractThe performance of the multianalysis–multimodel superensemble weather forecast and its comparison with several other operational prediction models is investigated in this study for the Southeast Asian region. The superensemble technique has been proven to exhibit exceptional skills compared with individual or even ensemble average forecasts for parameters such as precipitation, motion and mass fields. Precipitation forecasts from the superensemble technique for the year of 2001 from this study show that its skill scores are better than the other global operational models employed. Its precipitation root‐mean‐square errors (RMSEs) are consistently lower than the other member models for the annual period examined. The equitable threat score, which indicates the association of the common rainfall threshold distribution between the forecast and observed best rainfall analysis from the Tropical Rainfall Measurement Mission and Special Sensor Microwave Imager dataset, is slightly underpredicted for light rain thresholds, but it still outperforms the other individual members or even the ensemble mean forecasts. Correlation analysis of the superensemble forecast against the observed is far higher than either the ensemble mean or the individual forecast member models and is maintained for up to 3‐day forecasts.The motion field comparisons also exhibit that the superensemble forecasts are closer to the observed data than the other member models due to its low systematic errors at the 850 hPa pressure level. Both the precipitation and wind field analyses have shown that the Florida State University multianalysis–multimodel superensemble forecast provides the lowest RMSE, the highest correlation against the best‐observed data and lowest systematic errors compared with the other participating operational models. These forecasts have the potential to provide better daily weather predictions over the Southeast Asian region, particularly during the early northeast monsoon that often causes heavy rainfall in the equatorial part of the Southeast Asian region. Copyright © 2004 Royal Meteorological Society