Abstract
Effective calibration of precipitation forecasts produced by numerical weather prediction (NWP) models faces challenges associated with the training sample size. Newly-operationalized NWP models may only accumulate a small number of forecasts and thus may limit robust parameter inference in forecast calibration. It is necessary to investigate how the performance of forecast calibration changes with the amount of training data, to determine an effective training sample size. In this study, we thoroughly investigate the impacts of training sample size on precipitation forecast calibration based on the Seasonally Coherent Calibration (SCC) model across Australia. Overall, the performance of the model tends to stabilize in most parts of Australia when raw forecasts of 10 months or longer are used for parameter inference. Whether the training dataset cover wet months substantially affects forecast calibration. The findings of this study are critical for understanding the impacts of training sample size on forecast calibration, and will provide implications for future forecast calibration and the generation of hindcasts.
Abstract
Effective calibration of precipitation forecasts produced by numerical weather prediction (NWP) models faces challenges associated with the training sample size. Newly-operationalized NWP models may only accumulate a small number of forecasts and thus may limit robust parameter inference in forecast calibration. It is necessary to investigate how the performance of forecast calibration changes with the amount of training data, to determine an effective training sample size. In this study, we thoroughly investigate the impacts of training sample size on precipitation forecast calibration based on the Seasonally Coherent Calibration (SCC) model across Australia. Overall, the performance of the model tends to stabilize in most parts of Australia when raw forecasts of 10 months or longer are used for parameter inference. Whether the training dataset cover wet months substantially affects forecast calibration. The findings of this study are critical for understanding the impacts of training sample size on forecast calibration, and will provide implications for future forecast calibration and the generation of hindcasts.
Abstract
The intricate cause–effect relations in air–sea interaction are investigated using a quantitative causal inference formalism. The formalism is first validated with a classical stochastic coupled model, and then applied to the observational time series of sea surface temperature (SST) and air–sea turbulent surface heat flux (SHF). We identify an overall asymmetry of causality between the two variables, namely, the causality from SHF to SST is significantly larger than that from SST to SHF over most of the global oceans. Geographically, the coupling is strongest in the tropics and gets weaker substantially in the extratropics. In the midlatitude ocean, SST makes higher contributions to the SHF variability in frontal regions. We further show that the identified causality is space- and time-scale dependent. The dominance of SHF driving SST occurs at basin scales, whereas the dominance of SST driving SHF mostly occurs at scales smaller than 10°. The causalities in both directions get larger with increasing time scale, and are less asymmetric at longer time scales. Also discussed here is the seasonality of the causality.
Abstract
The intricate cause–effect relations in air–sea interaction are investigated using a quantitative causal inference formalism. The formalism is first validated with a classical stochastic coupled model, and then applied to the observational time series of sea surface temperature (SST) and air–sea turbulent surface heat flux (SHF). We identify an overall asymmetry of causality between the two variables, namely, the causality from SHF to SST is significantly larger than that from SST to SHF over most of the global oceans. Geographically, the coupling is strongest in the tropics and gets weaker substantially in the extratropics. In the midlatitude ocean, SST makes higher contributions to the SHF variability in frontal regions. We further show that the identified causality is space- and time-scale dependent. The dominance of SHF driving SST occurs at basin scales, whereas the dominance of SST driving SHF mostly occurs at scales smaller than 10°. The causalities in both directions get larger with increasing time scale, and are less asymmetric at longer time scales. Also discussed here is the seasonality of the causality.
Abstract
During the boreal winter, the India-Burma trough (IBT), a shortwave trough system primarily positioned over the northern Bay of Bengal, exerts a significant synoptic scale variation and impact on precipitation in South and East Asia. This study utilizes the 6-hourly ERA5 dataset to objectively identify and track 714 IBT events from 1981 to 2019. On average, IBT occurred about 54.7 days per year, with an average of 18.3 events annually and lasting around 2.5 days. IBT events are classified into two types: local and eastward-moving. Local IBTs manifest in the middle and lower troposphere with a vertical temperature structure of warm-over-cold, whereas the signals of the eastward-moving IBTs extend from the lower to the upper troposphere, exhibiting cold anomalies vertically. The impacts on winter precipitation in South and East Asia differ significantly between these two IBT types. Local IBTs are primarily linked with the eastward propagation of wave disturbances along the subtropical westerly jet, while eastward-moving IBTs are associated with robust disturbance sources from mid-high latitudes and jointly modulated by the consequent eastward propagation of Rossby wave trains along northern and southern branches of the westerly jet streams. Precipitation anomalies during local IBTs are typically positive (negative) in the Indochina Peninsula and southwestern China (Indian Peninsula), whereas eastward-moving IBTs correlate with more intense and widespread precipitation in South and East Asia, with a pronounced band of anomalies from the Indochina Peninsula to southern China.
Abstract
During the boreal winter, the India-Burma trough (IBT), a shortwave trough system primarily positioned over the northern Bay of Bengal, exerts a significant synoptic scale variation and impact on precipitation in South and East Asia. This study utilizes the 6-hourly ERA5 dataset to objectively identify and track 714 IBT events from 1981 to 2019. On average, IBT occurred about 54.7 days per year, with an average of 18.3 events annually and lasting around 2.5 days. IBT events are classified into two types: local and eastward-moving. Local IBTs manifest in the middle and lower troposphere with a vertical temperature structure of warm-over-cold, whereas the signals of the eastward-moving IBTs extend from the lower to the upper troposphere, exhibiting cold anomalies vertically. The impacts on winter precipitation in South and East Asia differ significantly between these two IBT types. Local IBTs are primarily linked with the eastward propagation of wave disturbances along the subtropical westerly jet, while eastward-moving IBTs are associated with robust disturbance sources from mid-high latitudes and jointly modulated by the consequent eastward propagation of Rossby wave trains along northern and southern branches of the westerly jet streams. Precipitation anomalies during local IBTs are typically positive (negative) in the Indochina Peninsula and southwestern China (Indian Peninsula), whereas eastward-moving IBTs correlate with more intense and widespread precipitation in South and East Asia, with a pronounced band of anomalies from the Indochina Peninsula to southern China.
Abstract
Monitoring fine particulate matter (PM2.5) is crucial for evaluating air quality and its effects on public health. However, the limited distribution of monitoring stations presents a challenge in accurately assessing air pollution, especially in areas distant from these stations. To address this challenge, our study introduces a two-step deep learning approach for estimating daily gap-free surface PM2.5 concentrations across the contiguous United States (CONUS) from 2018 to 2022, with a spatial resolution of 4 km. In the first phase, we employ a Depthwise-Partial Convolutional Neural Network (DW-PCNN) to fill gaps between surface PM2.5 stations, utilizing Aerosol Optical Depth (AOD) data from the Moderate Resolution Imaging Spectroradiometer (MODIS). In the second phase, we integrate the PM2.5 grids imputed by the DW-PCNN with meteorological and anthropogenic variables into a Deep Convolutional Neural Network (Deep-CNN) to further enhance the accuracy of our estimation. This enables us to estimate gap-free surface PM2.5 concentrations accurately, evidenced by a Pearson’s correlation coefficient (R) of 0.92 and an Index of Agreement (IOA) of 0.96 in ten-fold cross-validation. We also introduce a grid-based method for calculating PM2.5 Design Values (DV), providing a continuous spatial representation of PM2.5 DV that enhances the traditional station-based approach. Our grid-based DV representations offer a comprehensive perspective on air quality, facilitating more detailed analysis. Furthermore, our model's ability to provide spatiotemporally consistent, gap-free PM2.5 data addresses the issue of missing values, supporting health impact research, policy formulation, and the accuracy of environmental assessments.
Abstract
Monitoring fine particulate matter (PM2.5) is crucial for evaluating air quality and its effects on public health. However, the limited distribution of monitoring stations presents a challenge in accurately assessing air pollution, especially in areas distant from these stations. To address this challenge, our study introduces a two-step deep learning approach for estimating daily gap-free surface PM2.5 concentrations across the contiguous United States (CONUS) from 2018 to 2022, with a spatial resolution of 4 km. In the first phase, we employ a Depthwise-Partial Convolutional Neural Network (DW-PCNN) to fill gaps between surface PM2.5 stations, utilizing Aerosol Optical Depth (AOD) data from the Moderate Resolution Imaging Spectroradiometer (MODIS). In the second phase, we integrate the PM2.5 grids imputed by the DW-PCNN with meteorological and anthropogenic variables into a Deep Convolutional Neural Network (Deep-CNN) to further enhance the accuracy of our estimation. This enables us to estimate gap-free surface PM2.5 concentrations accurately, evidenced by a Pearson’s correlation coefficient (R) of 0.92 and an Index of Agreement (IOA) of 0.96 in ten-fold cross-validation. We also introduce a grid-based method for calculating PM2.5 Design Values (DV), providing a continuous spatial representation of PM2.5 DV that enhances the traditional station-based approach. Our grid-based DV representations offer a comprehensive perspective on air quality, facilitating more detailed analysis. Furthermore, our model's ability to provide spatiotemporally consistent, gap-free PM2.5 data addresses the issue of missing values, supporting health impact research, policy formulation, and the accuracy of environmental assessments.
Abstract
We present the Energy Balance Model – Kalman Filter (EBM-KF), a hybrid model projecting and assimilating the global mean surface temperature (GMST) and ocean heat content anomaly (OHCA). It combines an annual energy balance model (difference equations) with 17 parameters drawn from the literature and a statistical Extended Kalman Filter assimilating GMST and OHCA, either observed timeseries or simulated by earth system models. Our motivation is to create an efficient and natural estimator of the climate state and its uncertainty, which we believe to be Gaussian at a global scale. We illustrate four applications: 1) EBM-KF generates a similar estimate to the 30-year time-averaged climate state 15 years sooner, or a model-simulated hindcasts’ annual ensemble average, depending on whether volcanic forcing is filtered or not. 2) EBM-KF conveniently assesses annually likelihoods of crossing a policy threshold. For example, based on temperature records up to the end of 2023, p=0.0017 that the climate state was 1.5°C over preindustrial, but there is a 16% likelihood that the GMST in 2023 itself could have been over that threshold. 3) A variant of the EBM-KF also approximates the spread of an entire climate model large ensemble using only one or a few ensemble members. 4) All variants of the EBM-KF are sufficiently fast to allow thorough sampling from non-Gaussian probabilistic futures, e.g., the impact of rare but significant volcanic eruptions. This sampling with the EBM-KF better determines how future volcanism may affect when policy thresholds will be crossed and what an ensemble with thousands of members exploring future intermittent volcanism reveals.
Abstract
We present the Energy Balance Model – Kalman Filter (EBM-KF), a hybrid model projecting and assimilating the global mean surface temperature (GMST) and ocean heat content anomaly (OHCA). It combines an annual energy balance model (difference equations) with 17 parameters drawn from the literature and a statistical Extended Kalman Filter assimilating GMST and OHCA, either observed timeseries or simulated by earth system models. Our motivation is to create an efficient and natural estimator of the climate state and its uncertainty, which we believe to be Gaussian at a global scale. We illustrate four applications: 1) EBM-KF generates a similar estimate to the 30-year time-averaged climate state 15 years sooner, or a model-simulated hindcasts’ annual ensemble average, depending on whether volcanic forcing is filtered or not. 2) EBM-KF conveniently assesses annually likelihoods of crossing a policy threshold. For example, based on temperature records up to the end of 2023, p=0.0017 that the climate state was 1.5°C over preindustrial, but there is a 16% likelihood that the GMST in 2023 itself could have been over that threshold. 3) A variant of the EBM-KF also approximates the spread of an entire climate model large ensemble using only one or a few ensemble members. 4) All variants of the EBM-KF are sufficiently fast to allow thorough sampling from non-Gaussian probabilistic futures, e.g., the impact of rare but significant volcanic eruptions. This sampling with the EBM-KF better determines how future volcanism may affect when policy thresholds will be crossed and what an ensemble with thousands of members exploring future intermittent volcanism reveals.
Abstract
This paper analyzes the formation dates of the n th storm in a sequence for all named North Atlantic tropical cyclones and assesses whether the intraseasonal length of the Atlantic hurricane season has changed temporally. The record-breaking 2020 season, with 30 named storms, set records for the earliest 3rd TC formation (Cristobal) and from the 6th TC (Fay) onward. Analysis of season length from 1851–2022 identifies only one statistically significant breakpoint detected in the early 1970s, roughly coinciding with the introduction of satellite observations. Since 1970, we also find a trend towards longer North Atlantic hurricane seasons The 1971–2022 trend is robust and statistically significant, whether assessed as the number of days between the first and last storm, or by the distance between intermediate percentiles (e.g., 10th to 90th, 15th to 85th). These increases are mainly associated with storms forming earlier in the calendar year and are best described as an upward trend rather than a stepwise shift between eras. Although a simple trend fits better than a stepwise model, improvement falls short of significance, so we do not formally reject the stepwise hypothesis. If, following this hypothesis, the 1950–2022 period is segmented into a high activity era (HAE1; 1950–1969), a low activity era (LAE; 1970–1994), and second high activity era (HAE2; 1995–2022), the median season length HAE2 is 12 days longer than in the first HAE, but this difference is not statistically significant (p = 0.58) and could be explained by the substantial difference in the observational network. The median season length in both HAE1 and AE2 are significantly longer (by 36 and 48 days, respectively) than the intervening LAE.
Abstract
This paper analyzes the formation dates of the n th storm in a sequence for all named North Atlantic tropical cyclones and assesses whether the intraseasonal length of the Atlantic hurricane season has changed temporally. The record-breaking 2020 season, with 30 named storms, set records for the earliest 3rd TC formation (Cristobal) and from the 6th TC (Fay) onward. Analysis of season length from 1851–2022 identifies only one statistically significant breakpoint detected in the early 1970s, roughly coinciding with the introduction of satellite observations. Since 1970, we also find a trend towards longer North Atlantic hurricane seasons The 1971–2022 trend is robust and statistically significant, whether assessed as the number of days between the first and last storm, or by the distance between intermediate percentiles (e.g., 10th to 90th, 15th to 85th). These increases are mainly associated with storms forming earlier in the calendar year and are best described as an upward trend rather than a stepwise shift between eras. Although a simple trend fits better than a stepwise model, improvement falls short of significance, so we do not formally reject the stepwise hypothesis. If, following this hypothesis, the 1950–2022 period is segmented into a high activity era (HAE1; 1950–1969), a low activity era (LAE; 1970–1994), and second high activity era (HAE2; 1995–2022), the median season length HAE2 is 12 days longer than in the first HAE, but this difference is not statistically significant (p = 0.58) and could be explained by the substantial difference in the observational network. The median season length in both HAE1 and AE2 are significantly longer (by 36 and 48 days, respectively) than the intervening LAE.
Abstract
Accurate sub-seasonal (2-8 weeks) prediction of monsoon precipitation is crucial for mitigating flood and heatwave disasters caused by intra-seasonal variability (ISV). However, current state-of-the-art sub-seasonal-to-seasonal (S2S) models have limited prediction skills beyond one week when predicting weekly precipitation. Our findings suggest that predictability primarily arises from strong ISV events, and the prediction skills for ISV events depend on the propagation stability of preceding signals, regardless of models. This allows us to identify opportunities and barriers (OBs) within S2S models, clarifying what the models can and cannot achieve in ISV event prediction. Focusing on the complex East Asian summer monsoon (EASM), we discover that stable propagation of Eurasian and tropical atmospheric wave trains towards East Asia serves as an opportunity. This opportunity offers a one-week leading prediction skill of up to 0.85 and skillful prediction up to 13 days ahead for 43% of all ISV events. However, the Tibetan Plateau barrier highlights the limitation of EASM predictability. Identifying these OBs will help us gain confidence in making more accurate sub-seasonal prediction.
Abstract
Accurate sub-seasonal (2-8 weeks) prediction of monsoon precipitation is crucial for mitigating flood and heatwave disasters caused by intra-seasonal variability (ISV). However, current state-of-the-art sub-seasonal-to-seasonal (S2S) models have limited prediction skills beyond one week when predicting weekly precipitation. Our findings suggest that predictability primarily arises from strong ISV events, and the prediction skills for ISV events depend on the propagation stability of preceding signals, regardless of models. This allows us to identify opportunities and barriers (OBs) within S2S models, clarifying what the models can and cannot achieve in ISV event prediction. Focusing on the complex East Asian summer monsoon (EASM), we discover that stable propagation of Eurasian and tropical atmospheric wave trains towards East Asia serves as an opportunity. This opportunity offers a one-week leading prediction skill of up to 0.85 and skillful prediction up to 13 days ahead for 43% of all ISV events. However, the Tibetan Plateau barrier highlights the limitation of EASM predictability. Identifying these OBs will help us gain confidence in making more accurate sub-seasonal prediction.
Abstract
We analyze century-end projections for the tropical Pacific upper-ocean currents simulated within the Climate Model Intercomparison Project Phase 6 (CMIP6) under global warming. We find that while the intensity of precipitation within the Intertropical Convergence Zone (ITCZ) increases, the ITCZ also shifts towards the equator and broadens, which reduces wind stress curl north of the equator. Consequently, the North Equatorial Countercurrent (NECC) shifts equatorward, following the ITCZ, and weakens, despite the more intense ITCZ. The strength of the North Equatorial Current (NEC) and the South Equatorial Current (SEC) also decreases due to the weakening of the Walker circulation and the corresponding wind stress. However, despite the weaker winds, the Equatorial Undercurrent (EUC) intensifies as it shoals due to stronger vertical stratification induced by surface warming. Furthermore, we find a slightly stronger zonal pressure gradient along the core of the EUC, instead of a weaker one expected from weaker wind stress and sea surface height gradient along the equator. Ultimately, we argue that it is reduced vertical friction theory due to increased ocean stratification and hence larger Richardson number that explains the faster EUC. These intricate balances control future changes in equatorial currents, and the uncertainties of projected changes need to be further examined.
Abstract
We analyze century-end projections for the tropical Pacific upper-ocean currents simulated within the Climate Model Intercomparison Project Phase 6 (CMIP6) under global warming. We find that while the intensity of precipitation within the Intertropical Convergence Zone (ITCZ) increases, the ITCZ also shifts towards the equator and broadens, which reduces wind stress curl north of the equator. Consequently, the North Equatorial Countercurrent (NECC) shifts equatorward, following the ITCZ, and weakens, despite the more intense ITCZ. The strength of the North Equatorial Current (NEC) and the South Equatorial Current (SEC) also decreases due to the weakening of the Walker circulation and the corresponding wind stress. However, despite the weaker winds, the Equatorial Undercurrent (EUC) intensifies as it shoals due to stronger vertical stratification induced by surface warming. Furthermore, we find a slightly stronger zonal pressure gradient along the core of the EUC, instead of a weaker one expected from weaker wind stress and sea surface height gradient along the equator. Ultimately, we argue that it is reduced vertical friction theory due to increased ocean stratification and hence larger Richardson number that explains the faster EUC. These intricate balances control future changes in equatorial currents, and the uncertainties of projected changes need to be further examined.
Abstract
Here we introduce the Spiralling Inverse Method (SIM) that provides estimates of the small-scale and mesoscale mixing strength. The SIM uses a vertical integral over a balance between the watermass transformation equation and the thermal wind equation. The result is an equation where all terms, except for the mixing strengths, can be obtained from hydrographic data of temperature and salinity. As an advantage, the SIM estimates the mixing strengths without the need of further knowledge of a reference velocity or streamfunction. Here we apply the SIM to a small region in the North Atlantic. We find that the estimates obtained by the SIM compare well to observations and other (inverse) estimates of the mixing strength. The SIM therefore has potential to improve and constrain parameterizations used for climate and ecosystem modelling using readily available hydrographic data.
Abstract
Here we introduce the Spiralling Inverse Method (SIM) that provides estimates of the small-scale and mesoscale mixing strength. The SIM uses a vertical integral over a balance between the watermass transformation equation and the thermal wind equation. The result is an equation where all terms, except for the mixing strengths, can be obtained from hydrographic data of temperature and salinity. As an advantage, the SIM estimates the mixing strengths without the need of further knowledge of a reference velocity or streamfunction. Here we apply the SIM to a small region in the North Atlantic. We find that the estimates obtained by the SIM compare well to observations and other (inverse) estimates of the mixing strength. The SIM therefore has potential to improve and constrain parameterizations used for climate and ecosystem modelling using readily available hydrographic data.
Abstract
The Atlantic meridional overturning circulation (AMOC) plays an important role in climate, transporting heat and salt to the subpolar North Atlantic. The AMOC’s variability is sensitive to atmospheric forcing, especially the North Atlantic Oscillation (NAO). Because AMOC observations are short, climate models are a valuable tool to study the AMOC’s variability. Yet, there are known issues with climate models, like uncertainties and systematic biases. To investigate this, preindustrial control experiments from models participating in the phase 6 of Coupled Model Intercomparison Project (CMIP6) are evaluated. There is a large, but correlated, spread in the models’ subpolar gyre mean surface temperature and salinity. By splitting models into groups of either a warm–salty or cold–fresh subpolar gyre, it is shown that warm–salty models have a lower sea ice cover in the Labrador Sea and, hence, enable a larger heat loss during a positive NAO. Stratification in the Labrador Sea is also weaker in warm–salty models, such that the larger NAO-related heat loss can also affect greater depths. As a result, subsurface density anomalies are much stronger in the warm–salty models than in those that tend to be cold and fresh. As these anomalies propagate southward along the western boundary, they establish a zonal density gradient anomaly that promotes a stronger delayed AMOC response to the NAO in the warm–salty models. These findings demonstrate how model mean state errors are linked across variables and affect variability, emphasizing the need for improvement of the subpolar North Atlantic mean states in models.
Abstract
The Atlantic meridional overturning circulation (AMOC) plays an important role in climate, transporting heat and salt to the subpolar North Atlantic. The AMOC’s variability is sensitive to atmospheric forcing, especially the North Atlantic Oscillation (NAO). Because AMOC observations are short, climate models are a valuable tool to study the AMOC’s variability. Yet, there are known issues with climate models, like uncertainties and systematic biases. To investigate this, preindustrial control experiments from models participating in the phase 6 of Coupled Model Intercomparison Project (CMIP6) are evaluated. There is a large, but correlated, spread in the models’ subpolar gyre mean surface temperature and salinity. By splitting models into groups of either a warm–salty or cold–fresh subpolar gyre, it is shown that warm–salty models have a lower sea ice cover in the Labrador Sea and, hence, enable a larger heat loss during a positive NAO. Stratification in the Labrador Sea is also weaker in warm–salty models, such that the larger NAO-related heat loss can also affect greater depths. As a result, subsurface density anomalies are much stronger in the warm–salty models than in those that tend to be cold and fresh. As these anomalies propagate southward along the western boundary, they establish a zonal density gradient anomaly that promotes a stronger delayed AMOC response to the NAO in the warm–salty models. These findings demonstrate how model mean state errors are linked across variables and affect variability, emphasizing the need for improvement of the subpolar North Atlantic mean states in models.
