Abstract
Stochastic representations of model uncertainties are of great importance for the performance of ensemble prediction systems (EPSs). The stochastically perturbed parameterization tendencies (SPPT) scheme with a single-scale random pattern has been used in the operational global EPS of China Meteorological Administration (CMA-GEPS) since 2018. To deal with deficiencies in this operational single-scale SPPT scheme, a combined scheme based on the multiscale SPPT (mSPPT) scheme and the stochastically perturbed parameterization for the planetary boundary layer (SPP-PBL) scheme is developed. In the combined scheme, the mSPPT component aims to expand model uncertainties characterized by SPPT at mesoscale, synoptic scale, and planetary scale. The SPP-PBL component with six vital parameters is used to capture uncertainties in PBL processes, which is underrepresented by SPPT for the tapering treatment within PBL. Comparisons between the operational SPPT scheme and the mSPPT scheme reveal that the mSPPT scheme can generate more improvements in both ensemble reliability and forecast skills mainly in tropics. Besides, additional benefits from SPP-PBL on top of mSPPT are shown to be primarily distributed in tropics at the lower layers below 850 hPa and surface. Furthermore, the combined scheme of mSPPT and SPP-PBL is suggested to yield better spread–error relationships and forecast skills than the operational SPPT scheme in terms of objective verification scores for standard upper-air variables and surface parameters. A case study for the extreme precipitation event on 20 July 2021 in Henan Province of China also demonstrates the better ability of the combined scheme in forecasting the precipitation intensity and location.
Significance Statement
A comprehensive and reasonable representation of model uncertainties helps to improve the performance of ensemble prediction systems (EPSs). Despite the popular usage in simulating model uncertainties, the stochastically perturbed parameterization tendencies (SPPT) scheme possesses several shortcomings. To overcome this, a combined scheme based on the multiscale SPPT (mSPPT) scheme and the stochastically perturbed parameterization for the planetary boundary layer (SPP-PBL) scheme is proposed. Based on the global EPS of China Meteorological Administration (CMA-GEPS), additional benefits from the independent usage of mSPPT and SPP-PBL are disclosed. And the combined scheme can inherit the merits of mSPPT and SPP-PBL and generate more improvements on ensemble performance than the original single-scale SPPT scheme. This research provides a guideline for future upgradation of CMA-GEPS.
Abstract
Stochastic representations of model uncertainties are of great importance for the performance of ensemble prediction systems (EPSs). The stochastically perturbed parameterization tendencies (SPPT) scheme with a single-scale random pattern has been used in the operational global EPS of China Meteorological Administration (CMA-GEPS) since 2018. To deal with deficiencies in this operational single-scale SPPT scheme, a combined scheme based on the multiscale SPPT (mSPPT) scheme and the stochastically perturbed parameterization for the planetary boundary layer (SPP-PBL) scheme is developed. In the combined scheme, the mSPPT component aims to expand model uncertainties characterized by SPPT at mesoscale, synoptic scale, and planetary scale. The SPP-PBL component with six vital parameters is used to capture uncertainties in PBL processes, which is underrepresented by SPPT for the tapering treatment within PBL. Comparisons between the operational SPPT scheme and the mSPPT scheme reveal that the mSPPT scheme can generate more improvements in both ensemble reliability and forecast skills mainly in tropics. Besides, additional benefits from SPP-PBL on top of mSPPT are shown to be primarily distributed in tropics at the lower layers below 850 hPa and surface. Furthermore, the combined scheme of mSPPT and SPP-PBL is suggested to yield better spread–error relationships and forecast skills than the operational SPPT scheme in terms of objective verification scores for standard upper-air variables and surface parameters. A case study for the extreme precipitation event on 20 July 2021 in Henan Province of China also demonstrates the better ability of the combined scheme in forecasting the precipitation intensity and location.
Significance Statement
A comprehensive and reasonable representation of model uncertainties helps to improve the performance of ensemble prediction systems (EPSs). Despite the popular usage in simulating model uncertainties, the stochastically perturbed parameterization tendencies (SPPT) scheme possesses several shortcomings. To overcome this, a combined scheme based on the multiscale SPPT (mSPPT) scheme and the stochastically perturbed parameterization for the planetary boundary layer (SPP-PBL) scheme is proposed. Based on the global EPS of China Meteorological Administration (CMA-GEPS), additional benefits from the independent usage of mSPPT and SPP-PBL are disclosed. And the combined scheme can inherit the merits of mSPPT and SPP-PBL and generate more improvements on ensemble performance than the original single-scale SPPT scheme. This research provides a guideline for future upgradation of CMA-GEPS.
Abstract
This study investigated variations in the Madden–Julian oscillation (MJO) behavior between two types of La Niña winters: mega and equatorial La Niñas. Results of this work show that in contrast to mega conditions, accompanied by more abundant intraseasonal column-integrated moisture anomalies over the planetary boundary layer, intensities of intraseasonal outgoing longwave radiation anomalies are stronger over the tropical western Pacific (WP) at MJO phases 5–6 under equatorial conditions. The occurrence of the moisture anomaly change averaged over the tropical WP is supported by a distinct moisture tendency difference. Moisture budget and multiscale interaction diagnoses underscore the pivotal effect of an area-averaged vertical gradient change in low-frequency background moisture in driving such tendency change. Considering notable MJO convection variations, the teleconnections associated with MJO phases 5–6 over East Asia (EA) were also explored, including warmer surface air temperature anomalies and stronger positive geopotential height anomalies at 200 hPa on the intraseasonal time scale under equatorial conditions. Results from a linear baroclinic model demonstrate that such teleconnection changes could be effectively explained by the linear response to the MJO’s diabatic heating anomaly. Roles of mean state and MJO itself variations in the linear response change are also discussed. These findings provide novel insights into MJO activity and offer potential improvements for subseasonal forecasting in EA.
Significance Statement
The relationship between El Niño–Southern Oscillation (ENSO) and MJO has been extensively explored recently. However, variations in the MJO behavior and its climate effects under mega and equatorial La Niña winters have not been thoroughly investigated. In this work, we utilized reanalysis datasets and an idealized linear baroclinic model to address these questions. We observed more robust and abundant intraseasonal convection activity and planetary boundary layer–integrated moisture anomaly averaged over the tropical WP at MJO phases 5–6 under equatorial conditions than mega conditions, which is closely linked with the vertical gradient change in low-frequency background moisture. Additionally, MJO-related teleconnections also exhibit some changes. This work may provide a new perspective on understanding the relationship between the MJO and ENSO and offer potential improvements for the subseasonal forecast.
Abstract
This study investigated variations in the Madden–Julian oscillation (MJO) behavior between two types of La Niña winters: mega and equatorial La Niñas. Results of this work show that in contrast to mega conditions, accompanied by more abundant intraseasonal column-integrated moisture anomalies over the planetary boundary layer, intensities of intraseasonal outgoing longwave radiation anomalies are stronger over the tropical western Pacific (WP) at MJO phases 5–6 under equatorial conditions. The occurrence of the moisture anomaly change averaged over the tropical WP is supported by a distinct moisture tendency difference. Moisture budget and multiscale interaction diagnoses underscore the pivotal effect of an area-averaged vertical gradient change in low-frequency background moisture in driving such tendency change. Considering notable MJO convection variations, the teleconnections associated with MJO phases 5–6 over East Asia (EA) were also explored, including warmer surface air temperature anomalies and stronger positive geopotential height anomalies at 200 hPa on the intraseasonal time scale under equatorial conditions. Results from a linear baroclinic model demonstrate that such teleconnection changes could be effectively explained by the linear response to the MJO’s diabatic heating anomaly. Roles of mean state and MJO itself variations in the linear response change are also discussed. These findings provide novel insights into MJO activity and offer potential improvements for subseasonal forecasting in EA.
Significance Statement
The relationship between El Niño–Southern Oscillation (ENSO) and MJO has been extensively explored recently. However, variations in the MJO behavior and its climate effects under mega and equatorial La Niña winters have not been thoroughly investigated. In this work, we utilized reanalysis datasets and an idealized linear baroclinic model to address these questions. We observed more robust and abundant intraseasonal convection activity and planetary boundary layer–integrated moisture anomaly averaged over the tropical WP at MJO phases 5–6 under equatorial conditions than mega conditions, which is closely linked with the vertical gradient change in low-frequency background moisture. Additionally, MJO-related teleconnections also exhibit some changes. This work may provide a new perspective on understanding the relationship between the MJO and ENSO and offer potential improvements for the subseasonal forecast.
Abstract
Land–atmospheric feedback influences the occurrence and severity of flash droughts. However, the observed and projected changes in flash droughts and associated land–atmospheric coupling have not been examined over India. Moreover, the causes of the rapid depletion of soil moisture during flash droughts are not well known. We identify major flash droughts and associated soil moisture–vapor pressure deficit (SM–VPD) coupling in India using ERA5 and simulations from global climate models (CMIP6-GCMs). The summer monsoon season (June–September) witnesses more than 60% of the flash drought events and a relatively higher rate of flash drought development. The flash drought frequency has mainly decreased during India’s observed climate (1980–2019), which is projected to decline further in the future warming climate. On the other hand, the flash drought development rate has significantly increased during the observed period, which is projected to enhance further under the warming climate. SM–VPD coupling during the flash drought onset-development phase is considerably higher (threefold to fivefold) than during the normal condition (in the absence of flash drought). The high (low) SM–VPD coupling explains the faster (slower) flash drought development rate in the observed and future warming climate. The strength of SM–VPD coupling has increased in the recent period and is projected to increase further in the future warming climate. The increased SM–VPD coupling can intensify future flash droughts in India, especially during the summer monsoon season, with considerable implications for agriculture, water resources, and ecosystems.
Significance Statement
This study aims to understand better the role of land–atmospheric coupling in explaining flash drought characteristics (frequency and development rate) in India. Strong land–atmospheric (SM–VPD) feedback might influence the regional weather patterns during flash drought, which often negatively impacts humans and the ecosystem. We explain the causes and drivers of increasing (decreasing) flash drought development rate (frequency) in India. The long-term change in SM–VPD coupling drives the frequency of flash drought, whereas an anomalous instantaneous change in coupling controls the flash drought development rate. More intense flash droughts contributed by the increased land–atmospheric coupling are projected in the future. Predicting the SM–VPD coupling metric can facilitate more time for preparedness, resulting in minimizing the flash drought impacts.
Abstract
Land–atmospheric feedback influences the occurrence and severity of flash droughts. However, the observed and projected changes in flash droughts and associated land–atmospheric coupling have not been examined over India. Moreover, the causes of the rapid depletion of soil moisture during flash droughts are not well known. We identify major flash droughts and associated soil moisture–vapor pressure deficit (SM–VPD) coupling in India using ERA5 and simulations from global climate models (CMIP6-GCMs). The summer monsoon season (June–September) witnesses more than 60% of the flash drought events and a relatively higher rate of flash drought development. The flash drought frequency has mainly decreased during India’s observed climate (1980–2019), which is projected to decline further in the future warming climate. On the other hand, the flash drought development rate has significantly increased during the observed period, which is projected to enhance further under the warming climate. SM–VPD coupling during the flash drought onset-development phase is considerably higher (threefold to fivefold) than during the normal condition (in the absence of flash drought). The high (low) SM–VPD coupling explains the faster (slower) flash drought development rate in the observed and future warming climate. The strength of SM–VPD coupling has increased in the recent period and is projected to increase further in the future warming climate. The increased SM–VPD coupling can intensify future flash droughts in India, especially during the summer monsoon season, with considerable implications for agriculture, water resources, and ecosystems.
Significance Statement
This study aims to understand better the role of land–atmospheric coupling in explaining flash drought characteristics (frequency and development rate) in India. Strong land–atmospheric (SM–VPD) feedback might influence the regional weather patterns during flash drought, which often negatively impacts humans and the ecosystem. We explain the causes and drivers of increasing (decreasing) flash drought development rate (frequency) in India. The long-term change in SM–VPD coupling drives the frequency of flash drought, whereas an anomalous instantaneous change in coupling controls the flash drought development rate. More intense flash droughts contributed by the increased land–atmospheric coupling are projected in the future. Predicting the SM–VPD coupling metric can facilitate more time for preparedness, resulting in minimizing the flash drought impacts.
Abstract
This paper analyzes the climatology, prediction skill, and predictability of tropical cyclones (TCs) in NASA’s Goddard Earth Observing System Subseasonal to Seasonal (GEOS-S2S) forecast system version 2. GEOS reasonably simulates the number and spatial distribution of TCs compared to observations except in the Atlantic where the model simulates too few TCs due to low genesis rates in the Caribbean Sea and Gulf of Mexico. The environmental conditions, diagnosed through a genesis potential index, do not clearly explain model biases in the genesis rates, especially in the Atlantic. At the storm scale, GEOS reforecasts replicate several key aspects of the thermodynamic and dynamic structures of observed TCs, such as a warm core and the secondary circulation. The model, however, fails to simulate an off-center eyewall when evaluating vertical velocity, precipitation, and moisture. The analysis of the prediction skill of TC genesis and occurrence shows that GEOS has comparable skill to other global models in WMO S2S archive and that its skill could be further improved by increasing the ensemble size. After calibration, GEOS forecasts are skillful in the western North Pacific and southern Indian Ocean up to 20 days in advance. A model-based predictability analysis demonstrates the importance of the Madden–Julian oscillation (MJO) as a source of predictability of TC occurrence beyond the 14-day lead time. Forecasts initialized under strong MJO conditions show evidence of predictability beyond week 3. However, due to model biases in the forecast distribution, there are notable gaps between the MJO-related prediction skill and predictability, which require further study.
Abstract
This paper analyzes the climatology, prediction skill, and predictability of tropical cyclones (TCs) in NASA’s Goddard Earth Observing System Subseasonal to Seasonal (GEOS-S2S) forecast system version 2. GEOS reasonably simulates the number and spatial distribution of TCs compared to observations except in the Atlantic where the model simulates too few TCs due to low genesis rates in the Caribbean Sea and Gulf of Mexico. The environmental conditions, diagnosed through a genesis potential index, do not clearly explain model biases in the genesis rates, especially in the Atlantic. At the storm scale, GEOS reforecasts replicate several key aspects of the thermodynamic and dynamic structures of observed TCs, such as a warm core and the secondary circulation. The model, however, fails to simulate an off-center eyewall when evaluating vertical velocity, precipitation, and moisture. The analysis of the prediction skill of TC genesis and occurrence shows that GEOS has comparable skill to other global models in WMO S2S archive and that its skill could be further improved by increasing the ensemble size. After calibration, GEOS forecasts are skillful in the western North Pacific and southern Indian Ocean up to 20 days in advance. A model-based predictability analysis demonstrates the importance of the Madden–Julian oscillation (MJO) as a source of predictability of TC occurrence beyond the 14-day lead time. Forecasts initialized under strong MJO conditions show evidence of predictability beyond week 3. However, due to model biases in the forecast distribution, there are notable gaps between the MJO-related prediction skill and predictability, which require further study.
Abstract
The prediction skill for individual Madden–Julian oscillation (MJO) events is highly variable, but the key factors behind this remain unclear. Using the latest hindcast results from the subseasonal-to-seasonal (S2S) phase II models, this study attempts to understand the diverse prediction skill for the MJO events with an enhanced convective anomaly over the eastern Indian Ocean (IO) at the forecast start date, by investigating the preference of the prediction skill to the MJO-associated convective anomalies and low-frequency background states (LFBS). Compared to the low-skill MJO events, the high-skill events are characterized by a stronger intraseasonal convection–circulation couplet over the IO before the forecast start date, which could result in a longer zonal propagation range during the forecast period, thereby leading to a higher score for assessing the prediction skill. The difference in intraseasonal fields can further be attributed to the LFBS of IO sea surface temperature (SST) and quasi-biannual oscillation (QBO), with the high-skill (low-skill) events corresponding to a warmer (colder) IO and easterly (westerly) QBO phase. The physical link is that a warm IO could increase the low-level convective instability and thus amplify MJO convection over the IO, whereas an easterly QBO phase could weaken the Maritime Continent barrier effect by weakening the static stability near the tropopause, thus favoring eastward propagation of the MJO. It is also found that the combined effects of IO SST and QBO phases are more effective in influencing MJO prediction skill than individual LFBS.
Significance Statement
Given the importance of the Madden–Julian oscillation (MJO) in the subseasonal predictability of global weather and climate, how well the MJO itself can be predicted is a matter of concern. This study reveals the critical observational factors that determine the variation in MJO prediction skill across events. Without making any presumptions about what the factors would be, the observed MJO events are separated according to their individual prediction skill, and the difference between the MJO events with higher and lower skill is then investigated. The results show that the low-frequency background states of the Indian Ocean sea surface temperature and the quasi-biannual oscillation are good indicators for MJO prediction skill, for their modulatory role in the MJO propagation range.
Abstract
The prediction skill for individual Madden–Julian oscillation (MJO) events is highly variable, but the key factors behind this remain unclear. Using the latest hindcast results from the subseasonal-to-seasonal (S2S) phase II models, this study attempts to understand the diverse prediction skill for the MJO events with an enhanced convective anomaly over the eastern Indian Ocean (IO) at the forecast start date, by investigating the preference of the prediction skill to the MJO-associated convective anomalies and low-frequency background states (LFBS). Compared to the low-skill MJO events, the high-skill events are characterized by a stronger intraseasonal convection–circulation couplet over the IO before the forecast start date, which could result in a longer zonal propagation range during the forecast period, thereby leading to a higher score for assessing the prediction skill. The difference in intraseasonal fields can further be attributed to the LFBS of IO sea surface temperature (SST) and quasi-biannual oscillation (QBO), with the high-skill (low-skill) events corresponding to a warmer (colder) IO and easterly (westerly) QBO phase. The physical link is that a warm IO could increase the low-level convective instability and thus amplify MJO convection over the IO, whereas an easterly QBO phase could weaken the Maritime Continent barrier effect by weakening the static stability near the tropopause, thus favoring eastward propagation of the MJO. It is also found that the combined effects of IO SST and QBO phases are more effective in influencing MJO prediction skill than individual LFBS.
Significance Statement
Given the importance of the Madden–Julian oscillation (MJO) in the subseasonal predictability of global weather and climate, how well the MJO itself can be predicted is a matter of concern. This study reveals the critical observational factors that determine the variation in MJO prediction skill across events. Without making any presumptions about what the factors would be, the observed MJO events are separated according to their individual prediction skill, and the difference between the MJO events with higher and lower skill is then investigated. The results show that the low-frequency background states of the Indian Ocean sea surface temperature and the quasi-biannual oscillation are good indicators for MJO prediction skill, for their modulatory role in the MJO propagation range.
Abstract
Realistic computational simulations in different oceanic basins reveal prevalent prograde mean flows (in the direction of topographic Rossby wave propagation along isobaths; a.k.a. topostrophy) on topographic slopes in the deep ocean, consistent with the barotropic theory of eddy-driven mean flows. Attention is focused on the Western Mediterranean Sea with strong currents and steep topography. These prograde mean currents induce an opposing bottom drag stress and thus a turbulent boundary-layer mean flow in the downhill direction, evidenced by a near-bottom negative mean vertical velocity. The slope-normal profile of diapycnal buoyancy mixing results in down-slope mean advection near the bottom (a tendency to locally increase the mean buoyancy) and up-slope buoyancy mixing (a tendency to decrease buoyancy) with associated buoyancy fluxes across the mean isopycnal surfaces (diapycnal downwelling). In the upper part of the boundary layer and nearby interior, the diapycnal turbulent buoyancy flux divergence reverses sign (diapycnal upwelling), with upward Eulerian mean buoyancy advection across isopycnal surfaces. These near-slope tendencies abate with further distance from the boundary. An along-isobath mean momentum balance shows an advective acceleration and a bottom-drag retardation of the prograde flow. The eddy buoyancy advection is significant near the slope, and the associated eddy potential energy conversion is negative, consistent with mean vertical shear flow generation for the eddies. This cross-isobath flow structure differs from previous proposals, and a new one-dimensional model is constructed for a topostrophic, stratified, slope bottom boundary layer. The broader issue of the return pathways of the global thermohaline circulation remains open, but the abyssal slope region is likely to play a dominant role.
Abstract
Realistic computational simulations in different oceanic basins reveal prevalent prograde mean flows (in the direction of topographic Rossby wave propagation along isobaths; a.k.a. topostrophy) on topographic slopes in the deep ocean, consistent with the barotropic theory of eddy-driven mean flows. Attention is focused on the Western Mediterranean Sea with strong currents and steep topography. These prograde mean currents induce an opposing bottom drag stress and thus a turbulent boundary-layer mean flow in the downhill direction, evidenced by a near-bottom negative mean vertical velocity. The slope-normal profile of diapycnal buoyancy mixing results in down-slope mean advection near the bottom (a tendency to locally increase the mean buoyancy) and up-slope buoyancy mixing (a tendency to decrease buoyancy) with associated buoyancy fluxes across the mean isopycnal surfaces (diapycnal downwelling). In the upper part of the boundary layer and nearby interior, the diapycnal turbulent buoyancy flux divergence reverses sign (diapycnal upwelling), with upward Eulerian mean buoyancy advection across isopycnal surfaces. These near-slope tendencies abate with further distance from the boundary. An along-isobath mean momentum balance shows an advective acceleration and a bottom-drag retardation of the prograde flow. The eddy buoyancy advection is significant near the slope, and the associated eddy potential energy conversion is negative, consistent with mean vertical shear flow generation for the eddies. This cross-isobath flow structure differs from previous proposals, and a new one-dimensional model is constructed for a topostrophic, stratified, slope bottom boundary layer. The broader issue of the return pathways of the global thermohaline circulation remains open, but the abyssal slope region is likely to play a dominant role.
Abstract
El Niño–Southern Oscillation (ENSO) is the leading mode of interannual ocean–atmosphere coupling in the tropical Pacific, greatly influencing the global climate system. Seasonal phase locking, which means that ENSO events usually peak in boreal winter, is a distinctive feature of ENSO. In model future projections, the ENSO sea surface temperature (SST) amplitude in winter shows no significant change with a large intermodel spread. However, whether and how ENSO phase locking will respond to global warming are not fully understood. In this study, using Community Earth System Model Large Ensemble (CESM-LE) projections, we found that the seasonality of ENSO events, especially its peak phase, has advanced under global warming. This phenomenon corresponds to the seasonal difference in the changes in the ENSO SST amplitude with an enhanced (weakened) amplitude from boreal summer to autumn (winter). Mixed layer ocean heat budget analysis revealed that the advanced ENSO seasonality is due to intensified positive meridional advective and thermocline feedback during the ENSO developing phase and intensified negative thermal damping during the ENSO peak phase. Furthermore, the seasonal variation in the mean El Niño–like SST warming in the tropical Pacific favors a weakened zonal advective feedback in boreal autumn–winter and earlier decay of ENSO. The advance of the ENSO peak phase is also found in most CMIP5/6 models that simulate the seasonal phase locking of ENSO well in the present climate. Thus, even though the amplitude response in the winter shows no model consensus, ENSO also significantly changes during different stages under global warming.
Abstract
El Niño–Southern Oscillation (ENSO) is the leading mode of interannual ocean–atmosphere coupling in the tropical Pacific, greatly influencing the global climate system. Seasonal phase locking, which means that ENSO events usually peak in boreal winter, is a distinctive feature of ENSO. In model future projections, the ENSO sea surface temperature (SST) amplitude in winter shows no significant change with a large intermodel spread. However, whether and how ENSO phase locking will respond to global warming are not fully understood. In this study, using Community Earth System Model Large Ensemble (CESM-LE) projections, we found that the seasonality of ENSO events, especially its peak phase, has advanced under global warming. This phenomenon corresponds to the seasonal difference in the changes in the ENSO SST amplitude with an enhanced (weakened) amplitude from boreal summer to autumn (winter). Mixed layer ocean heat budget analysis revealed that the advanced ENSO seasonality is due to intensified positive meridional advective and thermocline feedback during the ENSO developing phase and intensified negative thermal damping during the ENSO peak phase. Furthermore, the seasonal variation in the mean El Niño–like SST warming in the tropical Pacific favors a weakened zonal advective feedback in boreal autumn–winter and earlier decay of ENSO. The advance of the ENSO peak phase is also found in most CMIP5/6 models that simulate the seasonal phase locking of ENSO well in the present climate. Thus, even though the amplitude response in the winter shows no model consensus, ENSO also significantly changes during different stages under global warming.
Abstract
Producing high-quality forecasts of key climate variables, such as temperature and precipitation, on subseasonal time scales has long been a gap in operational forecasting. This study explores an application of machine learning (ML) models as postprocessing tools for subseasonal forecasting. Lagged numerical ensemble forecasts (i.e., an ensemble where the members have different initialization dates) and observational data, including relative humidity, pressure at sea level, and geopotential height, are incorporated into various ML methods to predict monthly average precipitation and 2-m temperature 2 weeks in advance for the continental United States. For regression, quantile regression, and tercile classification tasks, we consider using linear models, random forests, convolutional neural networks, and stacked models (a multimodel approach based on the prediction of the individual ML models). Unlike previous ML approaches that often use ensemble mean alone, we leverage information embedded in the ensemble forecasts to enhance prediction accuracy. Additionally, we investigate extreme event predictions that are crucial for planning and mitigation efforts. Considering ensemble members as a collection of spatial forecasts, we explore different approaches to using spatial information. Trade-offs between different approaches may be mitigated with model stacking. Our proposed models outperform standard baselines such as climatological forecasts and ensemble means. In addition, we investigate feature importance, trade-offs between using the full ensemble or only the ensemble mean, and different modes of accounting for spatial variability.
Significance Statement
Accurately forecasting temperature and precipitation on subseasonal time scales—2 weeks–2 months in advance—is extremely challenging. These forecasts would have immense value in agriculture, insurance, and economics. Our paper describes an application of machine learning techniques to improve forecasts of monthly average precipitation and 2-m temperature using lagged physics-based predictions and observational data 2 weeks in advance for the entire continental United States. For lagged ensembles, the proposed models outperform standard benchmarks such as historical averages and averages of physics-based predictions. Our findings suggest that utilizing the full set of physics-based predictions instead of the average enhances the accuracy of the final forecast.
Abstract
Producing high-quality forecasts of key climate variables, such as temperature and precipitation, on subseasonal time scales has long been a gap in operational forecasting. This study explores an application of machine learning (ML) models as postprocessing tools for subseasonal forecasting. Lagged numerical ensemble forecasts (i.e., an ensemble where the members have different initialization dates) and observational data, including relative humidity, pressure at sea level, and geopotential height, are incorporated into various ML methods to predict monthly average precipitation and 2-m temperature 2 weeks in advance for the continental United States. For regression, quantile regression, and tercile classification tasks, we consider using linear models, random forests, convolutional neural networks, and stacked models (a multimodel approach based on the prediction of the individual ML models). Unlike previous ML approaches that often use ensemble mean alone, we leverage information embedded in the ensemble forecasts to enhance prediction accuracy. Additionally, we investigate extreme event predictions that are crucial for planning and mitigation efforts. Considering ensemble members as a collection of spatial forecasts, we explore different approaches to using spatial information. Trade-offs between different approaches may be mitigated with model stacking. Our proposed models outperform standard baselines such as climatological forecasts and ensemble means. In addition, we investigate feature importance, trade-offs between using the full ensemble or only the ensemble mean, and different modes of accounting for spatial variability.
Significance Statement
Accurately forecasting temperature and precipitation on subseasonal time scales—2 weeks–2 months in advance—is extremely challenging. These forecasts would have immense value in agriculture, insurance, and economics. Our paper describes an application of machine learning techniques to improve forecasts of monthly average precipitation and 2-m temperature using lagged physics-based predictions and observational data 2 weeks in advance for the entire continental United States. For lagged ensembles, the proposed models outperform standard benchmarks such as historical averages and averages of physics-based predictions. Our findings suggest that utilizing the full set of physics-based predictions instead of the average enhances the accuracy of the final forecast.
Abstract
The variability of Arctic sea ice extent (SIE) on interannual and multi-decadal timescales is examined in 29 models with historical forcing participating in phase 6 of the Coupled Model Intercomparison Project (CMIP6) and in 20th-century sea ice reconstructions. Results show that during the historical period with low external forcing (1850-1919), CMIP6 models display relatively good agreement in their representation of interannual sea ice variability (IVSIE), but exhibit pronounced inter-model spread in multi-decadal sea ice variability (MVSIE), which is overestimated with respect to sea ice reconstructions and is dominated by model uncertainty in sea ice simulation in the sub-polar North Atlantic. We find that this is associated with differences in models’ sensitivity to northern hemispheric sea surface temperatures (SST). Additionally, we show that while CMIP6 models are generally capable of simulating multi-decadal changes in Arctic sea ice from the mid-20th century to present day, they tend to underestimate the observed sea ice decline during the Early Twentieth-Century Warming (ETCW; 1915-1945). These results suggest the need for an improved characterization of the sea ice response to multi-decadal climate variability, in order to address the sources of model bias and reduce the uncertainty in future projections arising from inter-model spread.
Abstract
The variability of Arctic sea ice extent (SIE) on interannual and multi-decadal timescales is examined in 29 models with historical forcing participating in phase 6 of the Coupled Model Intercomparison Project (CMIP6) and in 20th-century sea ice reconstructions. Results show that during the historical period with low external forcing (1850-1919), CMIP6 models display relatively good agreement in their representation of interannual sea ice variability (IVSIE), but exhibit pronounced inter-model spread in multi-decadal sea ice variability (MVSIE), which is overestimated with respect to sea ice reconstructions and is dominated by model uncertainty in sea ice simulation in the sub-polar North Atlantic. We find that this is associated with differences in models’ sensitivity to northern hemispheric sea surface temperatures (SST). Additionally, we show that while CMIP6 models are generally capable of simulating multi-decadal changes in Arctic sea ice from the mid-20th century to present day, they tend to underestimate the observed sea ice decline during the Early Twentieth-Century Warming (ETCW; 1915-1945). These results suggest the need for an improved characterization of the sea ice response to multi-decadal climate variability, in order to address the sources of model bias and reduce the uncertainty in future projections arising from inter-model spread.
