Climate Forecast System Version Research Articles

Comparison of intra-seasonal forecast of Indian summer monsoon between two versions of NCEP coupled models

The real-time forecasting of monsoon activity over India on extended range time scale (about 3 weeks) is analyzed for the monsoon season of 2012 during June to September (JJAS) by using the outputs from latest (CFSv2 [Climate Forecast System version 2]) and previous version (CFSv1 [Climate Forecast System version 1]) of NCEP coupled modeling system. The skill of monsoon rainfall forecast is found to be much better in CFSv2 than CFSv1. For the country as a whole the correlation coefficient (CC) between weekly observed and forecast rainfall departure was found to be statistically significant (99 % level) at least for 2 weeks (up to 18 days) and also having positive CC during week 3 (days 19–25) in CFSv2. The other skill scores like the mean absolute error (MAE) and the root mean square error (RMSE) also had better performance in CFSv2 compared to that of CFSv1. Over the four homogeneous regions of India the forecast skill is found to be better in CFSv2 with almost all four regions with CC significant at 95 % level up to 2 weeks, whereas the CFSv1 forecast had significant CC only over northwest India during week 1 (days 5–11) forecast. The improvement in CFSv2 was very prominent over central India and northwest India compared to other two regions. On the meteorological subdivision level (India is divided into 36 meteorological subdivisions) the percentage of correct category forecast was found to be much higher than the climatology normal forecast in CFSv2 as well as in CFSv1, with CFSv2 being 8–10 % higher in the category of correct to partially correct (one category out) forecast compared to that in CFSv1. Thus, it is concluded that the latest version of CFS coupled model has higher skill in predicting Indian monsoon rainfall on extended range time scale up to about 25 days.

Does bias correction in the forecasted SST improve the extended range prediction skill of active‐break spells of Indian summer monsoon rainfall?

AbstractImpact of bias correction of sea surface temperature (SST) forecast on extended range (ER, ∼3–4 weeks) prediction skill is studied using the bias‐corrected forecasted SST from Climate Forecast System version 2 (CFSv2) as the boundary condition for running the Global Forecast System version 2 (GFSv2) model. Potential predictability limit is comparable (∼16 days) for both bias‐corrected GFSv2 (GFSv2bc) and CFSv2. Prediction skills of active and break spells and of low‐frequency monsoon intraseasonal oscillations is higher for GFSv2bc at all lead pentads. Although initially same, predictability error after 14 days grows slightly faster for GFSv2bc compared to CFSv2. Bias correction in SST has minimal impact in short‐to‐medium range, while substantial influence is felt in ER between 12–18 days.

Does An ENSO-Conditional Skill Mask Improve Seasonal Predictions?

Abstract The National Centers for Environmental Prediction Climate Prediction Center uses statistical tools together with the Climate Forecast System (CFS) to produce forecasts for seasonal outlooks of U.S. temperature and precipitation. They are combined using an optimal weighting procedure that depends on a skill mask consisting of the average historical forecast skill of each tool. However, it is likely that skill during El Niño–Southern Oscillation events is higher and the use of this information in developing forecasts could lead to improved seasonal predictions. This study explores the potential to improve the skill of seasonal predictions by developing an ENSO-conditional skill mask. The conditional masks are developed in a perfect-model framework using the CFS version 2 hindcasts and two indices of ENSO. The skill of the indices in forecasting variations in conditional skill is evaluated. The ENSO-conditional skill masks provide improvements in correlation skill over the unconditional mask when averaged over the globe. The masks are applied to tercile forecasts of seasonal temperature and precipitation during the spring and forecasts are verified in a perfect-model context. Application of the conditional masks to tercile forecasts results in modified Heidke skill scores of more than 10% less than using the average mask for temperature and little difference in skill for precipitation. This is attributed to the larger number of equal chances forecasts when using the conditional masks, particularly for temperature. For precipitation, the skill predicted by the average and conditional masks is frequently below 0.3, leading to low skill regardless of which mask is used.

Probabilistic Seasonal Forecasting of African Drought by Dynamical Models

Abstract As a natural phenomenon, drought can have devastating impacts on local populations through food insecurity and famine in the developing world, such as in Africa. In this study, the authors have established a seasonal hydrologic forecasting system for Africa. The system is based on the Climate Forecast System, version 2 (CFSv2), and the Variable Infiltration Capacity (VIC) land surface model. With a set of 26-yr (1982–2007) seasonal hydrologic hindcasts run at 0.25°, the probabilistic drought forecasts are validated using the 6-month Standard Precipitation Index (SPI6) and soil moisture percentile as indices. In terms of Brier skill score (BSS), the system is more skillful than climatology out to 3–5 months, except for the forecast of soil moisture drought over central Africa. The spatial distribution of BSS, which is similar to the pattern of persistency, shows more heterogeneity for soil moisture than the SPI6. Drought forecasts based on SPI6 are generally more skillful than for soil moisture, and their differences originate from the skill attribute of resolution rather than reliability. However, the soil moisture drought forecast can be more skillful than SPI6 at the beginning of the rainy season over western and southern Africa because of the strong annual cycle. Singular value decomposition (SVD) analysis of African precipitation and global SSTs indicates that CFSv2 reproduces the ENSO dominance on rainy season drought forecasts quite well, but the corresponding SVD mode from observations and CFSv2 only account for less than 24% and 31% of the covariance, respectively, suggesting that further understanding of drought drivers, including regional atmospheric dynamics and land–atmosphere coupling, is necessary.

A dynamical‐statistical forecast model for the annual frequency of western Pacific tropical cyclones based on the NCEP Climate Forecast System version 2

A dynamical‐statistical forecast model for the annual tropical cyclones over the western North Pacific is developed based on the empirical relationship between the actual annual number of tropical cyclones (ANTCs) and the dynamical predictions of large‐scale variables by the Climate Forecast System version 2 of the National Centers for Environmental Prediction (NCEP). On interannual time scales, the ANTCs are significantly and negatively correlated with the July–October tropical North Atlantic sea surface temperature, tropical western Pacific vertical zonal wind shear (WPVZWS), and subtropical Pacific geopotential height at 500 hPa (HGT500). They are also positively correlated with the zonal wind at 850 hPa over the tropical Pacific Ocean. Skillful forecasts of the above four potential predictors are made with the 24‐member ensemble predictions by the NCEP model. The two‐predictor model with the HGT500 and the WPVZWS shows the most skillful hindcasts at 0–2 month leads assessed by the leave‐one‐out cross validation for the ANTCs over the 31 year record between 1982 and 2012. The corresponding correlation coefficients and the root‐mean‐square errors (RMSEs) between the observed and hindcast ANTCs are in the ranges from 0.73 to 0.79 and from 3.11 to 2.75, respectively. Observed ANTCs during El Niño–Southern Oscillation events are generally well captured with RMSEs ranging from 3.12 to 3.04 during El Niño years and from 3.62 to 2.44 during La Niña years. The forecast skill of the model for the past 10 years (2003–2012) is competitive with the current forecast schemes. The forecast model initialized in March, May, and June 2013 suggests an inactive season for 2013, with about 22 tropical cyclones.

Dominant Modes of Wintertime Upper-Tropospheric Temperature Variations over Asia and Links to Surface Climate

AbstractIn this study, the authors investigate the variations and predictability of wintertime upper-tropospheric temperature (UTT) over Asia, which are often linked to severe climate anomalies, and the associated features of large-scale circulation and surface climate. The ECMWF Interim Re-Analysis (ERA-Interim) and hindcast of the NCEP Climate Forecast System, version 2 (CFSv2), are mainly analyzed.The first empirical orthogonal function mode of UTT shows a dipole structure, with a strong positive center over southern China and a weak negative center over Mongolia. The second mode is featured by a monopole variation, with a positive center appearing from the northwestern Tibetan Plateau (TP) to Japan. The third mode exhibits a tripole pattern, with two positive centers over Pakistan and the Sea of Japan and a negative center over central Asia. The first mode is linked to El Niño–Southern Oscillation, accompanied by surface warming over the southeastern TP and deficient precipitation over southern China, the Korean Peninsula, and from equatorial East Africa to the east of the TP. The second mode is associated with circulation anomalies similar to those associated with the Arctic Oscillation, with significant warming over East Asia. The third mode features two wave trains and is linked to the Middle East jet stream, and is associated with excessive precipitation from the eastern TP to southern Japan. The CFSv2 can predict the first mode skillfully by several months in advance, but it shows little skill in predicting the second and third modes.

Impact of land-atmospheric coupling in CFSv2 on drought prediction

Recent summers in the United States have been plagued by intense droughts that have caused significant damage to crops and have had a large impact on society. The ability to forecasts such events would allow for preparations that could help reduce the impact on society. Coupled land–atmosphere–ocean models were created to provide such forecasts but there are large uncertainties associated with their predictions. The predictive skill of these models is particularly low during the convective season due to the weaker connections with the oceans and an increase in the land–atmosphere interactions. To better understand the degradation of forecasts skill during the summer months and its connection to the land–atmosphere interactions we analyze National Centers for Environmental Prediction’s Climate Forecast System Version 2 (CFSv2) in terms of its climatological land–atmosphere interactions. To do this we use a recently developed classification of land–atmosphere interactions and other diagnostic variables to compare the reanalysis from the Climate Forecast System (CFSR) with CFSv2 re-forecasts (CFSRR) over the period 1982–2009. Coupling in the CFSRR tends toward the wet coupling regime for most areas east of the Rocky Mountains. Although the specific mechanism driving CFSRR to wet coupling state varies by region, the overall cause is enhanced vegetation rooting depth, originally implemented to address a near-surface warm bias in CFSR. The long-term tendency to wet coupling precludes the forecast model from consistently predicting and maintaining drought over the continental US.

Lagged Ensembles, Forecast Configuration, and Seasonal Predictions

Abstract An analysis of lagged ensemble seasonal forecasts from the National Centers for Environmental Prediction (NCEP) Climate Forecast System, version 2 (CFSv2), is presented. The focus of the analysis is on the construction of lagged ensemble forecasts with increasing lead time (thus allowing use of larger ensemble sizes) and its influence on seasonal prediction skill. Predictions of seasonal means of sea surface temperature (SST), 200-hPa height (z200), precipitation, and 2-m air temperature (T2m) over land are analyzed. Measures of prediction skill include deterministic (anomaly correlation and mean square error) and probabilistic [rank probability skill score (RPSS)]. The results show that for a fixed lead time, and as one would expect, the skill of seasonal forecast improves as the ensemble size increases, while for a fixed ensemble size the forecast skill decreases as the lead time becomes longer. However, when a forecast is based on a lagged ensemble, there exists an optimal lagged ensemble time (OLET) when positive influence of increasing ensemble size and negative influence due to an increasing lead time result in a maximum in seasonal prediction skill. The OLET is shown to depend on the geographical location and variable. For precipitation and T2m, OLET is relatively longer and skill gain is larger than that for SST and tropical z200. OLET is also dependent on the skill measure with RPSS having the longest OLET. Results of this analysis will be useful in providing guidelines on the design and understanding relative merits for different configuration of seasonal prediction systems.

Variability of the Indian Ocean SST and its possible impact on summer western North Pacific anticyclone in the NCEP Climate Forecast System

The NCEP Climate Forecast System version 2 (CFSv2) provides important source of information about the seasonal prediction of climate over the Indo-Pacific oceans. In this study, the authors provide a comprehensive assessment of the prediction of sea surface temperature (SST) in the tropical Indian Ocean (IO). They also investigate the impact of tropical IO SST on the summer anomalous anticyclonic circulation over the western North Pacific (WNPAC), focusing on the relative contributions of local SST and remote forcing of tropical IO SST to WNPAC variations. The CFSv2 captures the two most dominant modes of summer tropical IO SST: the IO basin warming (IOBW) mode and the IO dipole (IOD) mode, as well as their relationship with El Nino-Southern Oscillation (ENSO). However, it produces a cold SST bias in IO, which may be attributed to deeper-than-observed mixed layer and smaller-than-observed total downward heat flux in the tropical IO. It also overestimates the correlations of ENSO with IOBW and IOD, but underestimates the magnitude of IOD and summer IOBW. The CFSv2 captures the climate anomalies related to IOBW but not those related to IOD. It depicts the impact of summer IOBW on WNPAC via the equatorial Kelvin wave, which contributes to the maintenance of WNPAC in July and August. The WNPAC in June is mostly forced by local cold SST, which is better predicted by the CFSv2 compared to July and August. The mechanism for WNPAC maintenance may vary with lead time in the CFSv2.

The missing teleconnection between the North Atlantic and the Sahel precipitation in CFSv2

AbstractThis study presents new findings on the link between interannual variabilities of atmospheric circulations over the North Atlantic and precipitation over the African Sahel (PS). Our analysis shows a meridionally stratified circulation wave train resembling the East Atlantic (EA) mode, apparently connected to PS through Rossby wave dispersion in the middle troposphere originating from the North Atlantic. However, the Climate Forecast System version 2 fails to depict the EA and its PS impact. Because the EA explains 29% of variance of PS, this portion of PS variability is either missing in the seasonal forecast, or being made up by an alternative process.

CFSv2 prediction skill of stratospheric temperature anomalies

This study evaluates the prediction skill of stratospheric temperature anomalies by the Climate Forecast System version 2 (CFSv2) reforecasts for the 12-year period from January 1, 1999 to December 2010. The goal is to explore if the CFSv2 forecasts for the stratosphere would remain skillful beyond the inherent tropospheric predictability time scale of at most 2 weeks. The anomaly correlation between observations and forecasts for temperature field at 50 hPa (T50) in winter seasons remains above 0.3 over the polar stratosphere out to a lead time of 28 days whereas its counterpart in the troposphere at 500 hPa drops more quickly and falls below the 0.3 level after 12 days. We further show that the CFSv2 has a high prediction skill in the stratosphere both in an absolute sense and in terms of gain over persistence except in the equatorial region where the skill would mainly come from persistence of the quasi-biennial oscillation signal. We present evidence showing that the CFSv2 forecasts can capture both timing and amplitude of wave activities in the extratropical stratosphere at a lead time longer than 30 days. Based on the mass circulation theory, we conjecture that as long as the westward tilting of planetary waves in the stratosphere and their overall amplitude can be captured, the CFSv2 forecasts is still very skillful in predicting zonal mean anomalies even though it cannot predict the exact locations of planetary waves and their spatial scales. This explains why the CFSv2 has a high skill for the first EOF mode of T50, the intraseasonal variability of the annular mode while its skill degrades rapidly for higher EOF modes associated with stationary waves. This also explains why the CFSv2’s skill closely follows the seasonality and its interannual variability of the meridional mass circulation and stratosphere polar vortex. In particular, the CFSv2 is capable of predicting mid-winter polar stratosphere warming events in the Northern Hemisphere and the timing of the final polar stratosphere warming in spring in both hemispheres 3–4 weeks in advance.

The seasonal footprinting mechanism in CFSv2: simulation and impact on ENSO prediction

The seasonal footprinting mechanism (SFM) is thought to be a pre-cursor to the El Nino Southern Oscillation (ENSO). Fluctuations in the North Pacific Oscillation (NPO) impact the ocean via surface heat fluxes during winter, leaving a sea-surface temperature (SST) “footprint” in the subtropics. This footprint persists through the spring, impacting the tropical Pacific atmosphere–ocean circulation throughout the following year. The simulation of the SFM in the National Centers for Environmental Prediction (NCEP)/Climate Forecast System, version 2 (CFSv2) is likely to have an impact on operational predictions of ENSO and potentially seasonal predictions in the United States associated with ENSO teleconnection patterns. The ability of the CFSv2 to simulate the SFM and the relationship between the SFM and ENSO prediction skill in the NCEP/CFSv2 are investigated. Results indicate that the CFSv2 is able to simulate the basic characteristics of the SFM and its relationship with ENSO, including extratropical sea level pressure anomalies associated with the NPO in the winter, corresponding wind and SST anomalies that impact the tropics, and the development of ENSO-related SST anomalies the following winter. Although the model is able to predict the correct sign of ENSO associated with the SFM in a composite sense, probabilistic predictions of ENSO following a positive or negative NPO event are generally less reliable than when the NPO is not active.

Quasi-Biweekly Mode and Its Modulation on the Diurnal Rainfall in Taiwan Forecasted by the CFS

Abstract The occurrence of diurnal afternoon convection in Taiwan undergoes substantial modulation from tropical intraseasonal oscillations in the western North Pacific, including the quasi-biweekly (QBW) mode. By analyzing surface station observations and the Climate Forecast System (CFS) Reanalyses (CFSR), as well as the NCEP CFS version 2 (CFSv2) reforecast data over 18 summers from 1993 to 2010, it was found that the QBW mode plays a significant role in the formation of episodic diurnal convection. When the cyclonic circulation of the QBW mode is located west of Taiwan, followed by an anticyclonic circulation to the east, Taiwan's diurnal convection activity tends to intensify and persists for about 4–7 days. Synoptically, this situation reflects the enhanced subtropical anticyclone leading to fair weather conditions and increased monsoon southwesterly winds moistening the lower troposphere, all of which are conducive to thermally induced diurnal convection in Taiwan. The opposite situation tends to suppress the diurnal convection activity for a sustained period of time. Based upon this synoptic linkage, an empirical relationship between the precipitation diurnal amplitude and low-level circulation fields of the CFSv2 is derived. It was found that the CFSv2 forecast exhibits an effective lead time ranging from 16 to 24 days for the QBW mode and, subsequently, diurnal convection episodes in Taiwan.

Prediction Skill and Bias of Tropical Pacific Sea Surface Temperatures in the NCEP Climate Forecast System Version 2

Abstract The prediction skill and bias of tropical Pacific sea surface temperature (SST) in the retrospective forecasts of the Climate Forecast System, version 2 (CFSv2), of the National Centers for Environmental Prediction were examined. The CFSv2 was initialized from the Climate Forecast System Reanalysis (CFSR) over 1982–2010. There was a systematic cold bias in the central–eastern equatorial Pacific during summer/fall. The cold bias in the Niño-3.4 index was about −2.5°C in summer/fall before 1999 but suddenly changed to −1°C around 1999, related to a sudden shift in the trade winds and equatorial subsurface temperature in the CFSR. The SST anomaly (SSTA) was computed by removing model climatology for the periods 1982–98 and 1999–2010 separately. The standard deviation (STD) of forecast SSTA agreed well with that of observations in 1982–98, but in 1999–2010 it was about 200% too strong in the eastern Pacific and 50% too weak near the date line during winter/spring. The shift in STD bias was partially related to change of ENSO characteristics: central Pacific (CP) El Niños were more frequent than eastern Pacific (EP) El Niños after 2000. The composites analysis shows that the CFSv2 had a tendency to delay the onset phase of the EP El Niños in the 1980s and 1990s but predicted their decay phases well. In contrast, the CFSv2 predicted the onset phase of the CP El Niños well but prolonged their decay phase. The hit rate for both El Niño and La Niña was lower in the later period than in the early period, and the false alarm for La Niña increased appreciably from the early to the later period.

Understanding Prediction Skill of Seasonal Mean Precipitation over the Tropics

Abstract The connection between the local SST and precipitation (SST–P) correlation and the prediction skill of precipitation on a seasonal time scale is investigated based on seasonal hindcasts from the National Centers for Environmental Prediction (NCEP) Climate Forecast System version 2 (CFSv2). The results demonstrate that there is good correspondence between the two: precipitation skill is generally high only over the regions where SST–P correlation is positive and is low where SST–P correlation is small or weakly negative. This result has fundamental implications for understanding the limits of precipitation predictability on seasonal time scale and helps explain spatial variations in the skill of seasonal mean precipitation. Over the regions where atmospheric variability drives the ocean variability (and consequently the local SST–P correlation is weakly negative), the inherently unpredictable nature of atmospheric variability leads to low predictability for seasonal precipitation. On the other hand, over the regions where slow time scale ocean variability drives the atmosphere (and the local SST–P correlation is large positive), the predictability of seasonal mean precipitation is also high.

The Role of Air–Sea Coupling in Seasonal Prediction of Asia–Pacific Summer Monsoon Rainfall

Abstract This study examines the role of the air–sea coupled process in the seasonal predictability of Asia–Pacific summer monsoon rainfall by comparing seasonal predictions from two carefully designed model experiments: tier 1 (fully coupled model) and tier 2 (AGCM with prescribed SSTs). In these experiments, an identical AGCM is used in both tier 1 and tier 2 predictions; the daily mean SSTs from tier 1 coupled predictions are prescribed as a boundary condition in tier 2 predictions. Both predictions start in April from 1982 to 2009, with four ensemble members for each case. The model used is the Climate Forecast System, version 2 (CFSv2), the current operational climate prediction model for seasonal-to-interannual prediction at the National Centers for Environmental Prediction (NCEP). Comparisons indicate that tier 2 predictions produce not only higher rainfall biases but also unrealistically high rainfall variations in the tropical western North Pacific (TWNP) and some coastal regions as well. While the prediction skill in terms of anomaly correlations does not present a significant difference between the two types of predictions, the root-mean-square errors (RMSEs) are clearly larger over the above-mentioned regions in the tier 2 prediction. The reduced RMSE skills in the tier 2 predictions are due to the lack of a coupling process in AGCM-alone simulations, which, particularly, results in an unrealistic SST–rainfall relationship over the TWNP region. It is suggested that for a prediction of summer monsoon rainfall over the Asia–Pacific region, it is necessary to use a coupled atmosphere–ocean (tier 1) prediction system.

Improved simulation of Indian summer monsoon in latest NCEP climate forecast system free run

ABSTRACTSimulation of Indian summer monsoon features by latest coupled model of National Centers for Environmental Prediction (NCEPs) Climate Forecast System version 2 (CFSv2) is attempted in its long run. Improvements in the simulation of Indian summer monsoon as compared with previous version (CFSv1) is accessed and areas which still require considerable refinements are introduced. It is found that, spatial pattern of seasonal mean rainfall and wind circulations are more realistic in CFSv2 as compared with CFSv1. Variance and northward propagation of intraseasonal oscillation (ISO), which also contribute to the seasonal mean rainfall are remarkably improved. However, the central Indian dry bias still persists and amplified. Pervasive cold bias in surface (2 m air temperature) as well as in the whole troposphere is further increased in CFSv2. These cold biases may be partly attributed to the lack of model's ability to realistically simulate the ratio of convective and stratiform rainfall. Sea‐surface temperature (SST) over the Indian Ocean is underestimated in CFSv2. However, CFSv1 shows east–west dipole structure in the bias. The teleconnection of El Nino Southern Oscillation (ENSO) and Indian summer monsoon rainfall (ISMR) in terms of Niño3 SST and monsoon rainfall correlation is more realistic in the latest version of the model. Overall, there are substantial improvements in CFSv2 as compared with CFSv1, but it has to evolve further to realistically simulate the mean and variability of ISMR.

Prediction of global patterns of dominant quasi-biweekly oscillation by the NCEP Climate Forecast System version 2

Daily output from the hindcasts by the National Centers for Environmental Prediction (NCEP) Climate Forecast System version 2 (CFSv2) is analyzed to understand the skill of forecasting atmospheric variability on quasi-biweekly (QBW) time scale. Eight dominant quasi-biweekly oscillation (QBWO) modes identified by the extended empirical orthogonal function analysis are focused. In the CFSv2, QBW variability exhibits a significant weakening tendency with lead time for all seasons. For most QBWO modes, the variance drops to only 50 % of the initial value at lead time of 11–15 days. QBW variability has better prediction skill in the winter hemisphere than in the summer hemisphere. Skillful forecast can reach about 10–15 days for most modes but those in the winter hemisphere have better forecast skills. Among the eight QBWO modes, the North Pacific mode and the South Pacific (SP) mode have the highest forecast skills while the Asia–Pacific mode and the Central American mode have the lowest skills. For the Asia–Pacific and Central American modes, the forecasted QBWO phase shows an obvious eastward shift with increase in lead time compared to observations, indicating a smaller propagating speed. However, the predicted feature for the SP mode is more realistic. Air–sea coupling on the QBW time scale is perhaps responsible for the different prediction skills for different QBWO modes. In addition, most QBWO modes have better forecasting skills in El Nino years than in La Nina years. Different dynamical mechanisms for various QBWO modes may be partially responsible for the differences in prediction skill among different QBWO modes.

Multi-model MJO forecasting during DYNAMO/CINDY period

The present study assesses the forecast skill of the Madden–Julian Oscillation (MJO) observed during the period of DYNAMO (Dynamics of the MJO)/CINDY (Cooperative Indian Ocean Experiment on Intraseasonal Variability in Year 2011) field campaign in the GFS (NCEP Global Forecast System), CFSv2 (NCEP Climate Forecast System version 2) and UH (University of Hawaii) models, and revealed their strength and weakness in forecasting initiation and propagation of the MJO. Overall, the models forecast better the successive MJO which follows the preceding event than that with no preceding event (primary MJO). The common modeling problems include too slow eastward propagation, the Maritime Continent barrier and weak intensity. The forecasting skills of MJO major modes reach 13, 25 and 28 days, respectively, in the GFS atmosphere-only model, the CFSv2 and UH coupled models. An equal-weighted multi-model ensemble with the CFSv2 and UH models reaches 36 days. Air–sea coupling plays an important role for initiation and propagation of the MJO and largely accounts for the skill difference between the GFS and CFSv2. A series of forecasting experiments by forcing UH model with persistent, forecasted and observed daily SST further demonstrate that: (1) air–sea coupling extends MJO skill by about 1 week; (2) atmosphere-only forecasts driven by forecasted daily SST have a similar skill as the coupled forecasts, which suggests that if the high-resolution GFS is forced with CFSv2 forecasted daily SST, its forecast skill can be much higher than its current level as forced with persistent SST; (3) atmosphere-only forecasts driven by observed daily SST reaches beyond 40 days. It is also found that the MJO–TC (Tropical Cyclone) interactions have been much better represented in the UH and CFSv2 models than that in the GFS model. Both the CFSv2 and UH coupled models reasonably well capture the development of westerly wind bursts associated with November 2011 MJO and the cyclogenesis of TC05A in the Indian Ocean with a lead time of 2 weeks. However, the high-resolution GFS atmosphere-only model fails to reproduce the November MJO and the genesis of TC05A at 2 weeks’ lead. This result highlights the necessity to get MJO right in order to ensure skillful extended-range TC forecasting.

CFSv2 ensemble prediction of the wintertime Arctic Oscillation

Lagged ensembles from the operational Climate Forecast System version 2 (CFSv2) seasonal hindcast dataset are used to assess skill in forecasting interannual variability of the December–February Arctic Oscillation (AO). We find that a small but statistically significant portion of the interannual variance (>20 %) of the wintertime AO can be predicted at leads up to 2 months using lagged ensemble averages. As far as we are aware, this is the first study to demonstrate that an operational model has discernible skill in predicting AO variability on seasonal timescales. We find that the CFS forecast skill is slightly higher when a weighted ensemble is used that rewards forecast runs with the most accurate representations of October Eurasian snow cover extent (SCE), hinting that a stratospheric pathway linking October Eurasian SCE with the AO may be responsible for the model skill. However, further analysis reveals that the CFS is unable to capture many important aspects of this stratospheric mechanism. Model deficiencies identified include: (1) the CFS significantly underestimates the observed variance in October Eurasian SCE, (2) the CFS fails to translate surface pressure anomalies associated with SCE anomalies into vertically propagating waves, and (3) stratospheric AO patterns in the CFS fail to propagate downward through the tropopause to the surface. Thus, alternate boundary forcings are likely contributing to model skill. Improving model deficiencies identified in this study may lead to even more skillful predictions of wintertime AO variability in future versions of the CFS.