Superensemble Forecasts Research Articles

PICTUREE-Aedes: A Web Application for Dengue Data Visualization and Case Prediction.

Dengue fever remains a significant public health concern in many tropical and subtropical countries, and there is still a need for a system that can effectively combine global risk assessment with timely incidence forecasting. This research describes an integrated application called PICTUREE-Aedes, which can collect and analyze dengue-related data, display simulation results, and forecast outbreak incidence. PICTUREE-Aedes automatically updates global temperature and precipitation data and contains historical records of dengue incidence (1960-2012) and Aedes mosquito occurrences (1960-2014) in its database. The application utilizes a mosquito population model to estimate mosquito abundance, dengue reproduction number, and dengue risk. To predict future dengue outbreak incidence, PICTUREE-Aedes applies various forecasting techniques, including the ensemble Kalman filter, recurrent neural network, particle filter, and super ensemble forecast, which are all based on user-entered case data. The PICTUREE-Aedes' risk estimation identifies favorable conditions for potential dengue outbreaks, and its forecasting accuracy is validated by available outbreak data from Cambodia.

Multimodel Ensemble Forecasts of Precipitation Based on an Object-Based Diagnostic Evaluation

Abstract We analyzed 24-h accumulated precipitation forecasts over the 4-month period from 1 May to 31 August 2013 over an area located in East Asia covering the region 15.05°–58.95°N, 70.15°–139.95°E generated with the ensemble prediction systems (EPS) from ECMWF, NCEP, UKMO, JMA, and CMA contained in the TIGGE dataset. The forecasts are first evaluated with the Method for Object-Based Diagnostic Evaluation (MODE). Then a multimodel ensemble (MME) forecast technique that is based on weights derived from object-based scores is investigated and compared with the equally weighted MME and the traditional gridpoint-based MME forecast using weights derived from the point-to-point metric, mean absolute error (MAE). The object-based evaluation revealed that attributes of objects derived from the ensemble members of the five individual EPS forecasts and the observations differ consistently. For instance, their predicted centroid location is more southwestward, their shape is more circular, and their orientation is more meridional than in the observations. The sensitivity of the number of objects and their attributes to methodological parameters is also investigated. An MME prediction technique that is based on weights computed from the object-based scores, median of maximum interest, and object-based threat score is explored and the results are compared with the ensemble forecasts of the individual EPS, the equally weighted MME forecast, and the traditional superensemble forecast. When using MODE statistics for the forecast evaluation, the object-based MME prediction outperforms all other predictions. This is mainly because of a better prediction of the objects’ centroid locations. When using the precipitation-based fractions skill score, which is not used in either of the weighted MME forecasts, the object-based MME forecasts are slightly better than the equally weighted MME forecasts but are inferior to the traditional superensemble forecast that is based on weights derived from the point-to-point metric MAE.

Superensemble forecast of respiratory syncytial virus outbreaks at national, regional, and state levels in the United States

Respiratory syncytial virus (RSV) infections peak during the winter months in the United States, yet the timing, intensity, and onset of these outbreaks vary each year. An RSV vaccine is on the cusp of being released; precise models and accurate forecasts of RSV epidemics may prove vital for planning where and when the vaccine should be deployed. Accurate forecasts with sufficient spatial and temporal resolution could also be used to support the prevention or treatment of RSV infections. Previously, we developed and validated an RSV forecast system at the regional scale in the United States. This model-inference system had considerable forecast skill, relative to the historical expectance, for outbreak peak intensity, total outbreak size, and onset, but only marginal skill for predicting the timing of the outbreak peak. Here, we use a superensemble approach to combine three forecasting methods for RSV prediction in the US at three different spatial resolutions: national, regional, and state. At the regional and state levels, we find a substantial improvement of forecast skill, relative to historical expectance, for peak intensity, timing, and onset outbreak up to two months in advance of the predicted outbreak peak. Moreover, due to the greater variability of RSV outbreaks at finer spatial scales, we find that improvement of forecast skill at the state level exceeds that at the regional and national levels. Such finer scale superensemble forecasts may be more relevant for effecting local-scale interventions, particularly in communities with a high burden of RSV infection.

Superensemble Statistical Forecasting of Monthly Precipitation over the Contiguous United States, with Improvements from Ocean-Area Precipitation Predictors

AbstractExtended precipitation forecasts, with leads of weeks to seasons, are valuable for planning water use and are produced by the U.S. National Weather Service. Forecast skill tends to be low and any skill improvement could be valuable. Here, methods are discussed for improving statistical precipitation forecasting over the contiguous United States. Monthly precipitation is forecast using predictors from the previous month. Testing shows that improvements are obtained from both improved statistical methods and from the use of satellite-based ocean-area precipitation predictors. The statistical superensemble method gives higher skill compared to traditional statistical forecasting. Ensemble statistical forecasting combines individual forecasts. The proposed superensemble is a weighted mean of many forecasts or of forecasts from different prediction systems and uses the forecast reliability estimate to define weights. The method is tested with different predictors to show its skill and how skill can be improved using additional predictors. Cross validation is used to evaluate the skill. Although predictions are strongly influenced by ENSO, in the superensemble other regions contribute more to the forecast skill. The superensemble optimally combines forecasts based on different predictor regions and predictor types. The contribution from multiple predictor regions improves skill and reduces the ENSO spring barrier. Adding satellite-based ocean-area precipitation predictors noticeably increases forecast skill. The resulting skill is comparable to that from dynamic-model forecasts, but the regions with best forecast skill may be different. This paper shows that the statistical superensemble forecasts may be complementary to dynamic forecasts and that combining them may further increase forecast skill.

Superensemble forecasts of dengue outbreaks.

In recent years, a number of systems capable of predicting future infectious disease incidence have been developed. As more of these systems are operationalized, it is important that the forecasts generated by these different approaches be formally reconciled so that individual forecast error and bias are reduced. Here we present a first example of such multi-system, or superensemble, forecast. We develop three distinct systems for predicting dengue, which are applied retrospectively to forecast outbreak characteristics in San Juan, Puerto Rico. We then use Bayesian averaging methods to combine the predictions from these systems and create superensemble forecasts. We demonstrate that on average, the superensemble approach produces more accurate forecasts than those made from any of the individual forecasting systems.

Improving multimodel medium range forecasts over the Greater Horn of Africa using the FSU superensemble

This study makes use of the WMO’s multimodel data set called THORPEX integrated grand global ensemble (TIGGE) towards the construction of multimodel superensemble forecasts covering a period of 10 days. The goal of this study is to explore the forecast skill for precipitation forecasts over the Greater Horn of Africa (this is a consortium of 11 countries). The multimodels include forecast data set from a suite of models that include: The European Centre for Medium Range Weather Forecasts (ECMWF), the National Centre for Environmental Prediction (NCEP), the Center for Weather Forecast and Climatic Studies (CPTEC) and the United Kingdom Meteorological Office (UKMO). After performing a training phase for the superensemble weights covering the previous 450 days of October, November and December months of 2008–2012, forecasts of precipitation were prepared for the multimodel superensemble. These covered day 1 to day 10 of forecasts over the region. Various skill metrics were prepared to validate the forecast rainfall against the tropical rainfall measuring mission (TRMM) observed rainfall data. This study shows that the construction of the multimodel superensemble was a worthwhile effort since it provided the best overall skills for the RMS errors, the spatial correlations and the equitable threat scores and their bias errors for precipitation forecasts from day 1 to day 10 over all of the countries covered by the Greater Horn of Africa. The best among the member model was the UKMO model. This study strongly suggests the usefulness of a product such as the multimodel superensemble for improved precipitation forecasts over East Africa.

Applicability of the superensemble to the tropical cyclone track forecasts in the western North Pacific

In this study a superensemble was constructed and assessed to examine its applicability to the tropical cyclone track forecasts in the western North Pacific. The data used for this study were outputs of 20 tropical cyclone forecast models and analyzed tropical cyclone tracks by the Korea Meteorological Administration from 2011 to 2013. The annual mean track errors were analyzed at the 24-, 48-, 72-, 96- and 120-h periods for 2012 and 2013, and the superensemble forecasts showed lower annual errors than the simple mean consensus (using 20 numerical models), ECMWF_TIGG, and GFS. The superensemble track errors for individual tropical cyclone cases were lower than the simple mean consensus over 60% of the total cases, and lower than the best-performing model over 50% of the total cases for the 24-, 48-, and 72-h forecast periods. In the track error distribution, the superensemble had lower density for relatively large errors than the simple mean consensus, and higher density for smaller errors than single models. When the results are combined, the probability of the superensemble yielding lower errors than the simple mean consensus and single models becomes high, which indicates that the superensemble can serve as an objective reference for the tropical cyclone track forecasts.

Rainfall anomaly prediction using statistical downscaling in a multimodel superensemble over tropical South America

This study addresses the predictability of rainfall variations over South America and the Amazon basin. A primary factor leading to model inaccuracy in precipitation forecasts is the coarse resolution data utilized by coupled models during the training phase. By using MERRA reanalysis and statistical downscaling along with the superensemble methodology, it is possible to obtain more precise forecast of rainfall anomalies over tropical South America during austral fall. Selective inclusion (and exclusion) of member models also allows for increased accuracy of superensemble forecasts. The use of coupled atmospheric–ocean numerical models to predict the rainfall anomalies has had mixed results. Improvement in individual member models is also possible on smaller spatial scales and in regions where substantial topographical changes were not handled well under original model initial conditions. The combination of downscaling and superensemble methodologies with other research methods presents the potential opportunity for increased accuracy not only in seasonal forecasts but on shorter temporal scales as well.

Lead time for medium range prediction of the dry spell of monsoon using multi-models

The purpose of this study is to address prediction of the start date and the duration of breaks in the summer monsoon rains using multi-model superensemble. The availability of datasets from the ‘observing system research and predictability experiment (THORPEX)’ initiated a forecast data archive, called THORPEX interactive grand global ensemble (TIGGE), makes it possible to use forecasts from a suite of individual ensemble prediction systems (member models) and to construct multi-model superensemble forecasts that are designed to remove the collective bias errors of the suite of models. Precipitation datasets are important for this study, we have used high resolution daily gridded rainfall dataset of India Meteorological Department (IMD), in addition to rainfall estimates from tropical rainfall microwave mission (TRMM) satellite and the CPC morphing technique (CMORPH). The scientific approach of this study entails the use of a multi-model superensemble for forecast and to verify against the rainfall information during a training phase, as well as during a forecast phase. We examine the results of forecasts out to day-10 and ask how well do forecast strings of day-1 through day-10 handle the prediction of the onset and duration of the breaks in the summer monsoon rains. Our results confirm that it is possible to predict the onset of a dry spell, around week in advance from the use of the multi-model superensemble and a suite of TIGGE models. We also examine trajectories of the parcels arriving in India in such forecasts from member models and from the multi-model superensemble to validate the arrival of descending dry desert air from the Arabian region during the dry spells and its mode of transition from wet spell. Some phenological features such as a shift in the latitude of the tropical easterly jet and changes in its intensity during break periods are additional observed features that are validated from the history of multi-model superensemble forecasts. Invariably this multi-model superensemble performs better than any single model in proving the better forecasts during our experiment period.

Improved Seasonal Precipitation Forecasts for the Asian Monsoon Using 16 Atmosphere–Ocean Coupled Models. Part II: Anomaly

Abstract This is the second part of a paper on the improved seasonal precipitation forecasts for the Asian monsoon using 16 atmosphere–ocean coupled models. This study utilizes a large suite of coupled atmosphere–ocean models; this second part largely addresses the skill of rainfall anomaly forecasts. These include both deterministic and probabilistic skill measures such as the RMS errors, anomaly correlations, equitable threat scores, and the Brier skill score. It was possible to improve the skills of rainfall climatology from the use of a downscaled multimodel superensemble to very high levels, and it is of interest to ask how far this methodology would go toward improving the skills of seasonal rainfall anomaly forecasts. It is possible to go through a sequence of multimodel post processing to improve upon these skills by using a dense rain gauge network over Asia, downscaling forecasts for each member model, and constructing a multimodel superensemble that benefits from the persistence of errors of the member models. This paper addresses the spinup issues of the downscaling and the superensemble results where the number of years of model data needed for training phase, for the downscaling, and for the construction of the superensemble, is addressed. In the context of cross validation, the training phase includes 14 seasons of monsoon data. The forecast phase is only one season; it is this season that was not included in the training phase each time. The relationship between data length and the number of models needed for enhanced skills is another issue that is addressed. Seasonal climate forecasts over the larger monsoon Asia domain and over the regional belts are evaluated. The superensemble forecasts invariably have the highest skill compared to the member models globally and regionally. This is largely due to the presence of large systematic errors in models that carry low seasonal prediction skills. Such models carry persistent signatures of systematic errors, and their errors are recognized by the multimodel superensemble. The probabilistic skills show that the superensemble-based forecasts carry a much higher reliability score compared to the member models. This implies that the superensemble-based forecasts are the most reliable among all the member models. It is possible to examine the performance of models and of the superensemble during periods of heavy monsoon rainfall versus those for deficient monsoon rainfall seasons. One of the conclusions of this study is that given the uncertainties in current modeling for seasonal rainfall forecasts, post processing of multimodel forecasts, using the superensemble methodology, seems to provide the most promising results for the rainfall anomaly forecasts. These results are confirmed by an additional skill metric where the RMS errors and the correlations of forecast skills are evaluated using a normalized precipitation anomaly for the forecasts and the observed estimates.

Hurricane forecasts using a suite of large-scale models

This paper provides an account of the performance of a multimodel ensemble for real time forecasts of Atlantic tropical cyclones during 2004, 2005 and 2006. The Florida State University (FSU) superensemble is based on a suite of model forecasts and the interpolated official forecast that were received in real time at the National Hurricane Center. The FSU superensemble is a multimodel ensemble that utilizes forecasts from the member models by removing their individual biases based on a recent past history of their performances. This superensemble carries separate statistical weights for track and intensity forecasts for every 6 h of the member model forecasts. The real time results from 2004 show an improvement up to 15% for track forecasts and up to 11% for intensity forecasts for the superensemble compared to other models and consensus aids. During 2005, the superensemble intensity performance was best for most lead times. The consistency of the superensemble forecasts of track are also illustrated for several storms of 2004 season. The superensemble methodology produced impressive intensity forecasts for Rita and Wilma during 2005. The study shows the capability of the superensemble in predicting rapidly intensifying storms when most member models failed to capture their strengthening.

Hurricane forecasts using a suite of large-scale models

This paper provides an account of the performance of a multimodel ensemble for real time forecasts of Atlantic tropical cyclones during 2004, 2005 and 2006. The Florida State University (FSU) superensemble is based on a suite of model forecasts and the interpolated official forecast that were received in real time at the National Hurricane Center. The FSU superensemble is a multimodel ensemble that utilizes forecasts from the member models by removing their individual biases based on a recent past history of their performances. This superensemble carries separate statistical weights for track and intensity forecasts for every 6 h of themembermodel forecasts.The real time results from 2004 show an improvement up to 15% for track forecasts and up to 11% for intensity forecasts for the superensemble compared to other models and consensus aids. During 2005, the superensemble intensity performance was best for most lead times. The consistency of the superensemble forecasts of track are also illustrated for several storms of 2004 season. The superensemble methodology produced impressive intensity forecasts for Rita andWilma during 2005. The study shows the capability of the superensemble in predicting rapidly intensifying storms when most member models failed to capture their strengthening

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.

An analysis of the difference between the multiple linear regression approach and the multimodel ensemble mean

An investigation of the difference in seasonal precipitation forecast skills between the multiple linear regression (MLR) ensemble and the simple multimodel ensemble mean (EM) was based on the forecast quality of individual models. The possible causes of difference in previous studies were analyzed. In order to make the simulation capability of studied regions relatively uniform, three regions with different temporal correlation coefficients were chosen for this study. Results show the causes resulting in the incapability of the MLR approach vary among different regions. In the Nino3.4 region, strong co-linearity within individual models is generally the main reason. However, in the high latitude region, no significant co-linearity can be found in individual models, but the abilities of single models are so poor that it makes the MLR approach inappropriate for superensemble forecasts in this region. In addition, it is important to note that the use of various score measurements could result in some discrepancies when we compare the results derived from different multimodel ensemble approaches.

Improving Global Model Precipitation Forecasts over India Using Downscaling and the FSU Superensemble. Part II: Seasonal Climate

AbstractThis study addresses seasonal forecasts of rains over India using the following components: high-resolution rain gauge–based rainfall data covering the years 1987–2001, rain-rate initialization, four global atmosphere–ocean coupled models, a regional downscaling of the multimodel forecasts, and a multimodel superensemble that includes a training and a forecast phase at the high resolution over the internal India domain. The results of monthly and seasonal forecasts of rains for the member models and for the superensemble are presented here. The main findings, assessed via the use of RMS error, anomaly correlation, equitable threat score, and ranked probability skill score, are (i) high forecast skills for the downscaled superensemble-based seasonal forecasts compared to the forecasts from the direct use of large-scale model forecasts were possible; (ii) very high scores for rainfall forecasts have been noted separately for dry and wet years, for different regions over India and especially for heavier rains in excess of 15 mm day−1; and (iii) the superensemble forecast skills exceed that of the benchmark observed climatology. The availability of reliable measures of high-resolution rain gauge–based rainfall was central for this study. Overall, the proposed algorithms, added together, show very promising results for the prediction of monsoon rains on the seasonal time scale.

Super-ensemble forecasts and resulting acoustic sensitivities in shallow waters

Multi-model Super-Ensembles (SE) which optimally combine different models, have 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 for acoustic Rapid Environmental Assessment purposes. In the framework of the BP07 experiment conducted South East of Elba, sound speed prediction skills of various SE techniques combining operational model outputs and in-situ measurements are assessed. Results suggest that SE-based predictions provide more robust 24 h forecasts. A detailed acoustic propagation sensitivity study at different frequencies and ranges also reviews the potential of these predictions for acoustic inversion and tomography efforts.

Use of a Dense Rain-gauge Network over India for Improving Blended TRMM Products and Downscaled Weather Models

This study makes use of a network of nearly 2100 rain gauges over India in order to statistically modify the Tropical Rainfall Measuring Mission (TRMM) algorithm 3B42. The validation and usefulness of the modified product is determined against rain gauge datasets and from the training and forecast phase of the Florida State University (FSU) multimodel superensemble. We use downscaled member model forecasts to construct superensemble forecasts. The member model forecasts are scaled down using the modified high-resolution TRMM rains during the training phase of the multimodel superensemble. We demonstrate that a mesoscale superensemble thus constructed has forecast efficiencies superior to those of all of the member models and to the ensemble mean (ENSM) and bias-corrected ensemble mean (BcorENSM). The forecast procedure includes a high-resolution downscaling of the model forecast rain and the construction of a multimodel superensemble. This procedure clearly provides a much-improved forecast of rain using metrics such as equitable threat scores (ETSs), anomaly correlations, root mean square (RMS) errors, and their bias scores. We also demonstrate that the performance of this modified TRMM algorithm provides results very clearly comparable to those obtained from the direct use of rain gauges throughout India and the TRMM 3B42 covering a domain of 50°S to 50°N.

Investigating the Utility of Using Cross-Oceanic Training Sets for Superensemble Forecasting of Eastern Pacific Tropical Cyclone Track and Intensity

Abstract This paper examines how combining training-set forecasts from two separate oceanic basins affects the resulting tropical cyclone track and intensity forecasts in a particular oceanic basin. Atlantic and eastern Pacific training sets for 2002 and 2003 are combined and used to forecast 2004 eastern Pacific tropical cyclones in a real-time setting. These experiments show that the addition of Atlantic training improves the 2004 eastern Pacific forecasts. Finally, a detailed study of training-set and real-time model biases is completed in an effort to determine why cross-oceanic training may have helped in this instance.

Multimodel ensemble forecasts for precipitations in China in 1998

Different multimodel ensemble methods are used to forecast precipitations in China, 1998, and their forecast skills are compared with those of individual models. Datasets were obtained from monthly simulations of eight models during the period of January 1979 to December 1998 from the “Climate of the 20th Century Experiment” (20C3M) for the Fourth IPCC Assessment Report. Climate Research Unit (CRU) data were chosen for the observation analysis field. Root mean square (RMS) error and correlation coefficients (R) are used to measure the forecast skills. In addition, superensemble forecasts based on different input data and weights are analyzed. Results show that for original data, superensemble forecasting based on multiple linear regression (MLR) performs best. However, for bias-corrected data, the superensemble based on singular value decomposition (SVD) produces a lower RMS error and a higher R than in the MLR superensemble. It is an interesting result that the SVD superensemble based on bias-corrected data performs better than the MLR superensemble, but that the SVD superensemble based on original data is inferior to the corresponding MLR superensemble. In addition, weights calculated by different data formats are shown to affect the forecast skills of the superensembles. In comparison with the MLR superensemble, a slightly significant effect is present in the SVD superensemble. However, both the SVD and MLR superensembles based on different weight formats outperform the ensemble mean of bias-corrected data.

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.