ECMWF Forecasting System Research Articles

Development of a hybrid wind field for modeling the tropical cyclone wave field

Due to the increasing trend of tropical cyclone occurrence in the recent decades, the necessity of developing models for forecasting and hindcasting these atmospheric-oceanic hazards, becomes more apparent. One of the uncertainties about the tropical cyclones is the wind field accuracy of modeling oceanic waves and flows. In this paper, six different wind fields have been examined for hindcasting the wave field of cyclone Gonu within the areas of the Arabian Sea and the Gulf of Oman. The wind field of ECMWF (ERA-Interim) and Climate Forecast System Reanalysis (CFSR) data, that obtained via weather research forecasting (WRF) mesoscale modeling and three wind fields achieved via Holland, Rankin, Young and Sobey (S&Y) parametric models have been investigated as well. The wave field modeling results have been compared with the third-generation SWAN model in two stations within the area of Chabahar Bay. The wind field results obtained via the reanalysis models and WRF one indicate an appropriate estimation of the wave characteristics and the maximum wave height at the measurement stations. In order to better evaluate the results of different wind fields, the maximum wave height maps have been studied in the simulation time period. The results indicated that the maximum wave height values obtained from the reanalysis data do not match the track and development direction of the cyclone Gonu. The mismatch between storm development track and maximum wave height in WRF data can also be observed. The maximum significant wave height obtained via the parametric wind fields is very compatible with the cyclone Gonu track. Examining the maximum wave heights obtained via different wind fields, along with comparing the wave characteristics at the measurement stations, highlights the effect of different wind fields in hindcasting the wave characteristics, especially on the northern coasts of the Gulf of Oman. Examination of the maximum wave heights obtained from different wind fields has been investigated in this study along with comparison of wave characteristics at measurement stations. Results indicated that the optimal approach for simulating the tropical cyclone's wave field is to use a hybrid one. Therefore, a hybrid wind field has been developed which includes the wind characteristics of the parametric model in the areas near the storm source and the WRF meso scale model in the outer zone. The hybrid wind field has led to an improvement in the characteristics of waves within deep and coastal waters.

Prediction error growth in a more realistic atmospheric toy model with three spatiotemporal scales

Abstract. This article studies the growth of the prediction error over lead time in a schematic model of atmospheric transport. Inspired by the Lorenz (2005) system, we mimic an atmospheric variable in one dimension, which can be decomposed into three spatiotemporal scales. We identify parameter values that provide spatiotemporal scaling and chaotic behavior. Instead of exponential growth of the forecast error over time, we observe a more complex behavior. We test a power law and the quadratic hypothesis for the scale-dependent error growth. The power law is valid for the first days of the growth, and with an included saturation effect, we extend its validity to the entire period of growth. The theory explaining the parameters of the power law is confirmed. Although the quadratic hypothesis cannot be completely rejected and could serve as a first guess, the hypothesis's parameters are not theoretically justifiable in the model. In addition, we study the initial error growth for the ECMWF forecast system (500 hPa geopotential height) over the 1986 to 2011 period. For these data, it is impossible to assess which of the error growth descriptions is more appropriate, but the extended power law, which is theoretically substantiated and valid for the Lorenz system, provides an excellent fit to the average initial error growth of the ECMWF forecast system. Fitting the parameters, we conclude that there is an intrinsic limit of predictability after 22 d.

On the quality of RS41 radiosonde descent data

Abstract. Radiosonde descent profiles have been available from tens of stations for several years now – mainly from Vaisala RS41 radiosondes. They have been compared with the ascent profiles, with ECMWF short-range forecasts and with co-located radio occultation retrievals. Over this time, our understanding of the data has grown, and the comparison has also shed some light on radiosonde ascent data. The fall rate is very variable and is an important factor, with high fall rates being associated with temperature biases, especially at higher altitudes. Ascent winds are affected by pendulum motion; on average, descent winds are less affected by pendulum motion and are smoother. It is plausible that the true wind variability in the vertical lies between that shown by ascent and descent profiles. This discrepancy indicates the need for reference wind measurements. With current processing, the best results are for radiosondes with parachutes and pressure sensors. Some of the wind, temperature and humidity data are now assimilated in the ECMWF forecast system.

Recalculation of error growth models' parameters for the ECMWF forecast system

Abstract. This article provides a new estimate of error growth models' parameters approximating predictability curves and their differentials, calculated from data of the ECMWF forecast system over the 1986 to 2011 period. Estimates of the largest Lyapunov exponent are also provided, along with model error and the limit value of the predictability curve. The proposed correction is based on the ability of the Lorenz (2005) system to simulate the predictability curve of the ECMWF forecasting system and on comparing the parameters estimated for both these systems, as well as on comparison with the largest Lyapunov exponent (λ=0.35 d−1) and limit value of the predictability curve (E∞=8.2) of the Lorenz system. Parameters are calculated from the quadratic model with and without model error, as well as by the logarithmic, general, and hyperbolic tangent models. The average value of the largest Lyapunov exponent is estimated to be in the < 0.32; 0.41 > d−1 range for the ECMWF forecasting system; limit values of the predictability curves are estimated with lower theoretically derived values, and a new approach for the calculation of model error based on comparison of models is presented.

Seasonal Drought Forecasting for Latin America Using the ECMWF S4 Forecast System

Meaningful seasonal prediction of drought conditions is key information for end-users and water managers, particularly in Latin America where crop and livestock production are key for many regional economies. However, there are still not many studies of the feasibility of such a forecasts at continental level in the region. In this study, precipitation predictions from the European Centre for Medium Range Weather (ECMWF) seasonal forecast system S4 are combined with observed precipitation data to generate forecasts of the standardized precipitation index (SPI) for Latin America, and their skill is evaluated over the hindcast period 1981–2010. The value-added utility in using the ensemble S4 forecast to predict the SPI is identified by comparing the skill of its forecasts with a baseline skill based solely on their climatological characteristics. As expected, skill of the S4-generated SPI forecasts depends on the season, location, and the specific aggregation period considered (the 3- and 6-month SPI were evaluated). Added skill from the S4 for lead times equaling the SPI accumulation periods is primarily present in regions with high intra-annual precipitation variability, and is found mostly for the months at the end of the dry seasons for 3-month SPI, and half-yearly periods for 6-month SPI. The ECMWF forecast system behaves better than the climatology for clustered grid points in the North of South America, the Northeast of Argentina, Uruguay, southern Brazil and Mexico. The skillful regions are similar for the SPI3 and -6, but become reduced in extent for the severest SPI categories. Forecasting different magnitudes of meteorological drought intensity on a seasonal time scale still remains a challenge. However, the ECMWF S4 forecasting system does capture the occurrence of drought events for the aforementioned regions and seasons reasonably well. In the near term, the largest advances in the prediction of meteorological drought for Latin America are obtainable from improvements in near-real-time precipitation observations for the region. In the longer term, improvements in precipitation forecast skill from dynamical models, like the fifth generation of the ECMWF seasonal forecasting system, will be essential in this effort.

Evaluation of Submonthly Precipitation Forecast Skill from Global Ensemble Prediction Systems

Abstract The prediction skill of precipitation at submonthly time scales during the boreal summer season is investigated based on hindcasts from three global ensemble prediction systems (EPSs). The results, analyzed for lead times up to 4 weeks, indicate encouraging correlation skill over some regions, particularly over the Maritime Continent and the equatorial Pacific and Atlantic Oceans. The hindcasts from all three models correspond to high prediction skill over the first week compared to the following three weeks. The ECMWF forecast system tends to yield higher prediction skill than the other two systems, in terms of both correlation and mean squared skill score. However, all three systems are found to exhibit large conditional biases in the tropics, highlighted using the mean squared skill score. The sources of submonthly predictability are examined in the ECMWF hindcasts over the Maritime Continent in three typical years of contrasting ENSO phase, with a focus on the combined impact of the intraseasonal MJO and interannual ENSO. Rainfall variations over Borneo in the ENSO-neutral year are found to correspond well with the dominant MJO phase. The contribution of ENSO becomes substantial in the two ENSO years, but the MJO impact can become dominant when the MJO occurs in phases 2–3 during El Niño or in phases 5–6 during the La Niña year. These results support the concept that “windows of opportunity” of high forecast skill exist as a function of ENSO and the MJO in certain locations and seasons, which may lead to subseasonal-to-seasonal forecasts of substantial societal value in the future.

The assimilation of horizontal line‐of‐sight wind information into the ECMWF data assimilation and forecasting system. Part II: The impact of degraded wind observations

Single‐component wind observations have been assimilated into the ECMWF data assimilation system. Various Observing System Experiments have been performed in order to estimate the impact of such information in numerical weather prediction. The investigations on the impact of the horizontal line‐of‐sight wind on the ECMWF analysis and forecast system is described in a two‐part article. Part I evaluates the impact of wind observations with respect to mass observations and the impact of assimilating different single wind components versus wind vector information. Part II assesses the forecast system performance in the presence of varying levels of random and systematic error in the line‐of‐sight wind measurements. The aim is to understand the sensitivity of weather forecasts with respect to possible observation error scenarios of the Doppler Wind Lidar (DWL) measurements from the Aeolus mission. It was found that even small wind observation biases (of the order of 1 m s−1) can deteriorate the forecast quality by 3–4% in terms of mean integrated total energy of the 24 h forecast error. This is particularly true for observation biases in the Tropics, in the upper troposphere and lower stratosphere. A 2 m s−1 bias degrades the forecasts by 15%. The increase of the line‐of‐sight random errors (from 25 to 100%) adds only 1–2% forecast skill degradation.

The assimilation of horizontal line‐of‐sight wind information into the ECMWF data assimilation and forecasting system. Part I: The assessment of wind impact

In preparation for the Aeolus Doppler Wind Lidar satellite mission, single‐component wind information from conventional observations was assimilated into the ECMWF data assimilation system. Various Observing System Experiments have been designed and performed in order to estimate the impact of such information in numerical weather prediction. The evaluation used various adjoint diagnostic tools and traditional statistical verification methods. The importance of assimilating wind observations as either single component or full vector wind is evaluated. Comparisons between the assimilation of wind and mass observations were also made. Wind observations can lead to significant improvement in the upper troposphere, lower stratosphere and in the Tropics. Mass data are more valuable in the midlatitudes, particularly in the lower part of the atmosphere. The comparison of additional mass and wind observations in the Global Observing System is useful to understand the importance of future wind observations. The investigation also highlighted that the impact of zonal wind observations is larger than the meridional wind (Aeolus will mostly measure wind components near the zonal direction). Root mean‐squared errors of temperature and wind forecasts when the single wind component (zonal) is assimilated show around 35% degradation up to day 2 forecasts and around 20% after day 2 as compared to assimilating full vector wind. Therefore the single (zonal) wind components can provide a large fraction of the vector wind impact particularly for the medium‐range forecast, which is promising for Aeolus.

Impact of GPS radio occultation measurements in the ECMWF system using adjoint‐based diagnostics

In this article, a comprehensive assessment of the impact of radio occultation (RO) observations in the operational ECMWF assimilation and forecast system is presented using different diagnostic tools. In particular, the observations' influence in the assimilation process and the related contribution to the short‐range forecast error of radio occultation observations is evaluated with recently developed diagnostic tools based on the adjoint version of the assimilation and forecast model. The sensitivity with respect to observation error variances is also evaluated for the assimilated observations. GPS‐RO is found to have the largest mean influence among satellite observations in the analysis. It is the fourth best satellite system for analysis information content and the second largest satellite contributor together with IASI and AIRS to decreasing the 24 h forecast error. For the whole observing system, with the exception of radiosondes and polar atmospheric motion vectors, the forecast error sensitivity to the observation error variance indicates that a deflation of the assumed observation errors would improve the forecast skill. For RO observations at all vertical levels, but predominantly between 10 and 20 km, a deflation of the observation error variance is suggested. Interestingly, the sensitivity computation recommends reducing the assumed errors mostly in layers where the weight given to GPS‐RO data is quite large.

Factors Influencing Skill Improvements in the ECMWF Forecasting System

AbstractDuring the past 30 years the skill in ECMWF numerical forecasts has steadily improved. There are three major contributing factors: 1) improvements in the forecast model, 2) improvements in the data assimilation, and 3) the increased number of available observations. In this study the authors are investigating the relative contribution from these three components by using the simple error growth model introduced in a previous study by Lorenz and extended in another study by Dalcher and Kalnay, together with the results from the ECMWF Re-Analysis Interim (ERA-Interim) forecasts where the improvement is only due to an increased number of observations. The authors are also applying the growth model on “lagged” forecast differences in order to investigate the usefulness of the forecast jumpiness as a diagnostic tool for improvements in the forecasts. The main finding is that the main contribution to the reduced forecast error comes from significant initial condition error reductions between 1996 and 2001 together with continuous model improvements. The changes in the available observations contributed to a lesser degree, but the authors note that all the ERA-Interim forecasts are from the satellite era and here the focus is on the midtroposphere in the extratropics. Regarding the jumpiness in the forecasts, this is mainly a function of the error in the initial conditions and is therefore an insufficient tool to investigate improvements in the full forecasting system.

Impact of the Northern Hemisphere extratropics on the skill in predicting the Madden Julian Oscillation

Using a series of relaxation hindcast experiments, this paper shows that the northern extratropics have a significant impact on the skill of the ECMWF forecast system in predicting the Madden Julian Oscillation (MJO) during boreal winter. The westerly wind burst associated with the March 1997 MJO event is a high‐profile example of an event that was influenced by extratropical circulation anomalies. The propagation speed of MJO events is overall improved when the northern extratropics are relaxed towards analysis fields. Additional experiments in which the relaxation is confined to different areas of the Northern Hemisphere extratropics suggest that it is atmospheric circulation anomalies in the North Pacific region which most strongly affect the skill in predicting the MJO.

Flow dependence of background errors and their vertical correlations for radiance‐data assimilation

AbstractThis article examines the dependence of background‐error statistics on synoptic conditions and flow patterns. Error variances and vertical correlations of background temperatures as used for variational radiance‐data assimilation are estimated for two different weather regimes over Europe using the NMC method. The results are validated with real observations, i.e. radiosonde data and microwave satellite radiances and generalised with half a year of global data from the ECMWF forecasting system, where weather conditions are distinguished using model fields of wind speed, mean sea level pressure, and relative vorticity. Strong winds, low pressure, and cyclonic flow generally induce larger background errors of 500 hPa temperature than calm winds, high pressure, and anticyclonic flow, and also broader temperature correlations in the vertical with other tropospheric levels. Copyright © 2010 Royal Meteorological Society

Impact of the Madden Julian Oscillation on tropical storms and risk of landfall in the ECMWF forecast system

Using a series of 46‐days hindcasts, this paper shows that the Madden Julian Oscillation (MJO) has a significant impact on the statistics of tropical storms generated by a dynamical model. This impact is generally consistent with observations. In particular the risk of landfall over Australia and North America varies significantly with the phases of the MJO, as observed. This suggests that sub‐seasonal prediction of the risk of tropical storm landfall should be possible with a dynamical model such as the one used in this study, which has skill in predicting the evolution of the MJO up to 20 days.

A Probabilistic Forecast Approach for Daily Precipitation Totals

Abstract Commonly, postprocessing techniques are employed to calibrate a model forecast. Here, a probabilistic postprocessor is presented that provides calibrated probability and quantile forecasts of precipitation on the local scale. The forecasts are based on large-scale circulation patterns of the 12-h forecast from the NCEP high-resolution Global Forecast System (GFS). The censored quantile regression is used to estimate selected quantiles of the precipitation amount and the probability of the occurrence of precipitation. The approach accounts for the mixed discrete-continuous character of daily precipitation totals. The forecasts are verified using a new verification score for quantile forecasts, namely the censored quantile verification (CQV) score. The forecast approach is as follows: first, a canonical correlation is employed to correct systematic deviations in the GFS large-scale patterns compared with the NCEP–NCAR reanalysis or the 40-yr ECMWF Re-Analysis (ERA-40). Second, the statistical quantile model between the large-scale circulation and the local precipitation quantile is derived using NCEP and ERA-40 reanalysis data. Then, the statistical quantile model is applied to 12-h forecasts provided by the GFS forecast system. The probabilistic forecasts are reliable and the relative gain in performance of the quantile as well as the probability forecasts compared to the climatological forecasts range between 20% and 50%. The importance of the various parts of the postprocessing is assessed, and the performance is compared to forecasts based on the direct precipitation output from the ECMWF forecast system.

Performance of the ECMWF forecasting system in the Arctic during winter

AbstractThe performance of the ECMWF forecasting system in the Arctic during boreal winter is investigated. A comparison between different analysis products suggests that synoptic‐scale features in the Arctic are relatively well represented by state‐of‐the‐art analysis systems. Furthermore, it is shown that the improvement in deterministic forecast error in the Arctic since the early 1980s closely follows that reported in previous studies for the Northern Hemisphere as a whole. One of the biggest ‘jumps’ in deterministic forecast skill, both in the troposphere and in the stratosphere, occurred in autumn 2000 when the high‐resolution deterministic system (TL511, about 40 km) was introduced at ECMWF. Our results suggest that 10‐day forecasts of the stratospheric circulation have been very skilful, particularly since the introduction of the high‐resolution system. Adjoint computations reveal pronounced flow‐dependence of the sensitivity of short‐range forecast error to initial perturbations at high northern latitudes (north of 70°N); this highlights the importance of using ensemble prediction systems (EPSs) for weather forecasting in the Arctic. Verification of the ECMWF EPS in the Arctic reveals substantial improvements in probabilistic predictive skill since the mid‐1990s. The most pronounced increase in probabilistic forecast skill, particularly on synoptic scales, occurred in autumn 2000 when the high‐resolution EPS (TL255, about 80 km) was introduced. Finally, it is shown that the observed stratosphere–troposphere link (‘downward propagation’ of stratospheric anomalies), which constitutes a potential source of extended‐range predictability in high latitudes, is realistically represented in seasonal integrations with the ECMWF model—even at the relatively low horizontal resolution of TL95 (about 180 km). Copyright © 2007 Royal Meteorological Society

Upper tropospheric divergence in tropical convective systems from Meteosat‐8

The upper level wind field divergence in tropical convective cloud systems is directly inferred from satellite tracked Atmospheric Motion Vectors (AMVs) using successive images at 15 minute intervals in the water vapor channel (6.2 μm) of Meteosat‐8. The use of radiances in the strongly absorbing 6.2 μm channel confines the feature tracking to the upper troposphere, enabling the derivation of consistent wind fields. A Lagrangian approach following the convective cloud systems throughout a day shows a pronounced diurnal cycle of the divergence with maximum values of 450 * 10−6 s−1. A comparison with the ECMWF forecast system shows that the forecast does not depict the observed spatial structure and does not reach the maximum values of observed divergence at the scale of tropical cloud systems. Divergence averages over the study area of about 1300 * 1400 km2 agree within a factor of two.

Reanalysis and reforecast of three major European storms of the twentieth century using the ECMWF forecasting system. Part II: Ensemble forecasts

AbstractIn Part II of this study the ECMWF Ensemble Prediction System (EPS) is used to study the probabilistic predictability of three major European storms of the twentieth century. The storms considered are the Dutch storm of 1 February 1953, the Hamburg storm of 17 February 1962, and the British/French storm of October 1987 (Great October storm). Common to all these storms is their severity that caused large loss of life and widespread damage.In Part I of this study it has been found that deterministic predictability of the Dutch and Hamburg storms amount to 48 and 84 hours, respectively. Here, it is shown that the ensemble forecasts supplement the deterministic forecasts. The large number of members in the 48 and 84 hour ensemble forecasts of the Dutch and Hamburg storms, respectively, suggest that at this forecast range and for these storms the sensitivity of the forecasts to analysis and model uncertainties is rather small. From these results, therefore, it is argued that reliable warnings (i.e. low probability for the occurrence of a forecast failure) for the Dutch and Hamburg storms could have been issued 48 and 84 hours, respectively, in advance, had the current ECMWF EPS been available.For the Great October storm it has been found in Part I of this study that short‐range and medium‐range forecasts of the intensity and track of the storm were very skilful with a high‐resolution model of the ECWMF model. The actual timing of the storm, however, was difficult to predict. Here, it is shown that the EPS is capable of predicting large forecast uncertainties associated with the timing of the Great October storm up to 4 days in advance. It is argued that reliable warnings could have been issued at least 96 hours in advance had the ECMWF EPS been available.From the results presented in this study it is concluded that an Ensemble Prediction System is an important component of every early warning system for it allows an a priori quantification of the probability of the occurrence of severe wind storms. Copyright © 2005 Royal Meteorological Society

Systematic errors of the atmospheric circulation in the ECMWF forecasting system

The aim of this study is to document some of the key systematic errors of the atmospheric circulation as simulated by the ECMWF model during the winter season (December–March). Two major topics are addressed. First, key systematic circulation errors are described and it is shown how they grow from the short range all the way to the extended range. Second, it is shown how systematic circulation errors changed during the last two decades. Results are presented for 500 hPa geopotential height fields, eddy kinetic energy, synoptic variability, and atmospheric blocking. A concise summary is somewhat hampered by the fact that systematic error characteristics significantly depend on the parameter, region, and the forecast range being considered. In general, though, it has been found that systematic circulation errors have been reduced considerably in recent years, particularly from the 1980s to the early 1990s. Model improvements are primarily manifest through a reduction of the rate at which systematic errors grow; their spatial structure, however, remains more or less unchanged for most of the parameters. On the other hand, and somewhat surprisingly, it has been found that for some parameters and regions systematic circulation errors have slightly deteriorated since the mid-1990s. Moreover, it turns out that during the first few days of the integration the spatial structure of systematic circulation errors undergoes considerable changes; in the medium range and beyond their spatial structure remains virtually unchanged. The most prominent systematic circulation errors identified in this study are: the development of a large anticyclonic bias in the central North Pacific in the medium-range and extended-range, the underestimation of kinetic energy of the transient eddies, a pronounced underestimation of synoptic activity in high latitudes, and the underestimation of atmospheric blocking in the extended range. In the medium range over Europe, however, it is shown that recent cycles of the ECMWF model are able to produce realistic blocking frequencies. Copyright © 2005 Royal Meteorological Society

Reanalysis and reforecast of three major European storms of the twentieth century using the ECMWF forecasting system. Part I: Analyses and deterministic forecasts

In Part I of this study recent versions of the ECWMF Integrated Forecasting System (IFS) are used together with historical observational data to carry out reanalyses and deterministic reforecasts of three major north-west European wind storms of the twentieth century. The storms considered are the Dutch storm of 1 February 1953, the Hamburg storm of 17 February 1962, and the British October storm of 1987 (Great October Storm). Common to all these storms is their severity, which caused large loss of life and widespread damage. Reanalysis of the storms is based on a 3D-Var and 4D-Var assimilation scheme at a horizontal resolution of TL159(≈ 125 km) and TL511(≈ 50 km), respectively. Similarly, two different horizontal resolutions (TL159 and TL511) are used to investigate the deterministic predictability of these storms. The lower-resolution system is exactly that used in the ERA-40 reanalysis project. The high-resolution system is a more recent version of the ECMWF IFS. It is shown that the basic characteristics of the Dutch and Hamburg storms that gave rise to the storm surge are well predicted by the single deterministic forecasts up to about 48 and 84 hours, respectively, in advance. Our capability to predict the Great October Storm is more difficult to assess. On the one hand, even recent versions of the ECMWF IFS underestimate the severity of the storm in the very short-range (12–24-hour forecasts started at 12 UTC 15 October 1987). On the other hand, the high-resolution version of the ECMWF IFS provides excellent deterministic forecasts of the track and intensity of the storm up to 96 hours in advance. However, there are errors in the timing of the storm (12 hours for the 96 hour forecast). From the results presented in this study it is concluded – bearing in mind the limited number of cases considered – that with the current ECMWF forecasting system reliable deterministic predictions of some European wind storms are possible several days in advance. Copyright © 2004 Royal Meteorological Society.

The use of TOVS clear radiances for numerical weather prediction using an updated forward model

AbstractThe impact of TOVS clear radiances, processed using a new transmittance database, on the ECMWF forecast system, is evaluated. This involved the production of a line‐by‐line transmittance database, of sets of coefficients for the operational fast radiative‐transfer scheme, and the use of the operational bias‐correction procedure to reduce residual long‐term systematic differences. Two case‐studies are used to examine data from TOVS instruments on three NOAA satellites, and the differences in radiances measured by the various platforms are discussed. Measured and simulated data are compared before and after the operational bias‐removal scheme, and a final impact assessment is made using an assimilation and forecast experiment.