Abstract
The simulations of clouds and surface radiation from 10 CMIP6 models and their CMIP5 predecessors are compared to the ARM ground-based observations over different climate regions. Compared to the ARM radar-lidar derived total cloud fractions (CFT ) and cloud fraction vertical distributions over the six selected sites, both CMIP5 and CMIP6 significantly underestimated CFT and low-level CF over the North Hemispheric mid-latitude sites (SGPC1 and ENAC1), although the biases are generally smaller in CMIP6. Over the tropical oceanic site (TWPC2), five out of 10 CMIP6 models better simulated low-level CF than their CMIP5 predecessors. CMIP6 simulations generally agreed well with the ARM observations in CFT and cloud fraction vertical distributions over the tropical continental (MAOM1) and coastal (TWPC3) sites but missed the transitions between dry and wet seasons, similar to CMIP5 simulations. The improvements in downwelling shortwave fluxes (SWdn ) at the surface from the majority of CMIP6 compared to CMIP5 primarily resulted from the improved cloud fraction simulations, especially over the SGPC1, ENAC1 and TWPC3 sites. By contrast, both CMIP5 and CMIP6 models exhibited diverse performances of clouds and shortwave radiations over the Arctic site (NSAC1), where CMIP6 models produced more clouds than CMIP5 models, especially for the low-level clouds. The comparisons between observations and CMIP5 and CMIP6 simulations provide valuable quantitative assessments of the accuracy of mean states and variabilities in the model simulations and shed light on general directions to improve climate models in different regions.
Abstract
The simulations of clouds and surface radiation from 10 CMIP6 models and their CMIP5 predecessors are compared to the ARM ground-based observations over different climate regions. Compared to the ARM radar-lidar derived total cloud fractions (CFT ) and cloud fraction vertical distributions over the six selected sites, both CMIP5 and CMIP6 significantly underestimated CFT and low-level CF over the North Hemispheric mid-latitude sites (SGPC1 and ENAC1), although the biases are generally smaller in CMIP6. Over the tropical oceanic site (TWPC2), five out of 10 CMIP6 models better simulated low-level CF than their CMIP5 predecessors. CMIP6 simulations generally agreed well with the ARM observations in CFT and cloud fraction vertical distributions over the tropical continental (MAOM1) and coastal (TWPC3) sites but missed the transitions between dry and wet seasons, similar to CMIP5 simulations. The improvements in downwelling shortwave fluxes (SWdn ) at the surface from the majority of CMIP6 compared to CMIP5 primarily resulted from the improved cloud fraction simulations, especially over the SGPC1, ENAC1 and TWPC3 sites. By contrast, both CMIP5 and CMIP6 models exhibited diverse performances of clouds and shortwave radiations over the Arctic site (NSAC1), where CMIP6 models produced more clouds than CMIP5 models, especially for the low-level clouds. The comparisons between observations and CMIP5 and CMIP6 simulations provide valuable quantitative assessments of the accuracy of mean states and variabilities in the model simulations and shed light on general directions to improve climate models in different regions.
Abstract
Extreme snow ablation can greatly impact regional hydrology, affecting streamflow, soil moisture, and groundwater supplies. Relatively little is known about the climatology of extreme ablation events in the eastern U.S., and the causal atmospheric forcing mechanisms behind such events. Studying the Susquehanna River Basin over a 50-year period, here we evaluate the variability of extreme ablation and river discharge events in conjunction with a synoptic classification and global-scale teleconnection pattern analysis. Results indicate that an average of 4.2 extreme ablation events occurred within the basin per year, where some 88% of those events resulted in an increase in river discharge when evaluated at a 3-day lag. Both extreme ablation and extreme discharge events occurred most frequently during instances of southerly synoptic scale flow, accounting for 35.7% and 35.8% of events, respectively. However, extreme ablation was also regularly observed during high-pressure overhead and rain-on-snow synoptic weather types. The largest magnitude of snow ablation per extreme event occurred during occasions of rain-on-snow, where a basin-wide, areal-weighted 5.7 cm of snow depth was lost, approximately 23% larger than the average extreme event. Interannually, southerly flow synoptic weather types were more frequent during winter seasons when the Arctic and North Atlantic Oscillations were positively phased. Approximately 30% of the variance in rain-on-snow weather type frequency was explained by the Pacific/North American Pattern. Evaluating the pathway of physical forcing mechanisms from regional events up through global patterns allows for improved understanding of the processes resulting in extreme ablation and discharge across the Susquehanna Basin.
Abstract
Extreme snow ablation can greatly impact regional hydrology, affecting streamflow, soil moisture, and groundwater supplies. Relatively little is known about the climatology of extreme ablation events in the eastern U.S., and the causal atmospheric forcing mechanisms behind such events. Studying the Susquehanna River Basin over a 50-year period, here we evaluate the variability of extreme ablation and river discharge events in conjunction with a synoptic classification and global-scale teleconnection pattern analysis. Results indicate that an average of 4.2 extreme ablation events occurred within the basin per year, where some 88% of those events resulted in an increase in river discharge when evaluated at a 3-day lag. Both extreme ablation and extreme discharge events occurred most frequently during instances of southerly synoptic scale flow, accounting for 35.7% and 35.8% of events, respectively. However, extreme ablation was also regularly observed during high-pressure overhead and rain-on-snow synoptic weather types. The largest magnitude of snow ablation per extreme event occurred during occasions of rain-on-snow, where a basin-wide, areal-weighted 5.7 cm of snow depth was lost, approximately 23% larger than the average extreme event. Interannually, southerly flow synoptic weather types were more frequent during winter seasons when the Arctic and North Atlantic Oscillations were positively phased. Approximately 30% of the variance in rain-on-snow weather type frequency was explained by the Pacific/North American Pattern. Evaluating the pathway of physical forcing mechanisms from regional events up through global patterns allows for improved understanding of the processes resulting in extreme ablation and discharge across the Susquehanna Basin.
Abstract
Rapid increases in the flash rate (FR) of a thunderstorm, so-called Lightning Jumps (LJs), have potential for nowcasting applications and to increase leadtimes for severe weather warnings. To date, there are some automated LJ algorithms that were developed and tuned for ground-based lightning locating systems. This study addresses the optimization of an automated LJ algorithm for the Geostationary Lightning Mapper (GLM) lightning observations from space. The widely used σ-LJ algorithm is used in its original form, and in an adapted calculation including the footprint area of the storm cell (FRarea LJ algorithm). In addition, a new Relative Increase Level (RIL) LJ algorithm is introduced. All algorithms are tested in different configurations and detected LJs are verified against National Centers for Environmental Information (NCEI) severe weather reports. Overall, the FRarea algorithm with an activation FR threshold of 15 flashes per minute (fl/min) and a σ-level threshold of 1.0 to 1.5 as well as the RIL algorithm with FR threshold of 15 fl/min and RIL threshold of 1.1 are recommended. These algorithms scored the best Critical Success Index (CSI) of about 0.5, with a Probability of Detection of 0.6 to 0.7 and a False Alarm Ratio of about 0.4. For daytime warm season thunderstorms the CSI can exceed 0.5, reaching 0.67 for storms observed during 3 consecutive days in April 2021. The CSI is generally lower at night and in winter.
Abstract
Rapid increases in the flash rate (FR) of a thunderstorm, so-called Lightning Jumps (LJs), have potential for nowcasting applications and to increase leadtimes for severe weather warnings. To date, there are some automated LJ algorithms that were developed and tuned for ground-based lightning locating systems. This study addresses the optimization of an automated LJ algorithm for the Geostationary Lightning Mapper (GLM) lightning observations from space. The widely used σ-LJ algorithm is used in its original form, and in an adapted calculation including the footprint area of the storm cell (FRarea LJ algorithm). In addition, a new Relative Increase Level (RIL) LJ algorithm is introduced. All algorithms are tested in different configurations and detected LJs are verified against National Centers for Environmental Information (NCEI) severe weather reports. Overall, the FRarea algorithm with an activation FR threshold of 15 flashes per minute (fl/min) and a σ-level threshold of 1.0 to 1.5 as well as the RIL algorithm with FR threshold of 15 fl/min and RIL threshold of 1.1 are recommended. These algorithms scored the best Critical Success Index (CSI) of about 0.5, with a Probability of Detection of 0.6 to 0.7 and a False Alarm Ratio of about 0.4. For daytime warm season thunderstorms the CSI can exceed 0.5, reaching 0.67 for storms observed during 3 consecutive days in April 2021. The CSI is generally lower at night and in winter.
Abstract
Machine learning algorithms are able to capture complex, nonlinear, interacting relationships and are increasingly used to predict agricultural yield variability at regional and national scales. Using explainable artificial intelligence (XAI) methods applied to such algorithms may enable better scientific understanding of drivers of yield variability. However, XAI methods may provide misleading results when applied to spatiotemporal correlated datasets. In this study, machine learning models are trained to predict simulated crop yield from climate indices, and the impact of cross-validation strategy on the interpretation and performance of the resulting models is assessed. Using data from a process-based crop model allows us to then comment on the plausibility of the “explanations” provided by XAI methods. Our results show that the choice of evaluation strategy has an impact on (i) interpretations of the model and (ii) model skill on held-out years and regions, after the evaluation strategy is used for hyperparameter tuning and feature selection. We find that use of a cross-validation strategy based on clustering in feature space achieves the most plausible interpretations as well as the best model performance on held-out years and regions. Our results provide the first steps toward identifying domain-specific “best practices” for the use of XAI tools on spatiotemporal agricultural or climatic data.
Significance Statement
“Explainable” or “interpretable” machine learning (XAI) methods have been increasingly used in scientific research to study complex relationships between climatic and biogeoscientific variables (such as crop yield). However, these methods can return contradictory, implausible, or ambiguous results. In this study, we train machine learning models to predict maize yield anomalies and vary the model evaluation method used. We find that the evaluation (cross validation) method used has an effect on model interpretation results and on the skill of resulting models in held-out years and regions. These results have implications for the methodological design of studies that aim to use XAI tools to identify drivers of, for example, crop yield variability.
Abstract
Machine learning algorithms are able to capture complex, nonlinear, interacting relationships and are increasingly used to predict agricultural yield variability at regional and national scales. Using explainable artificial intelligence (XAI) methods applied to such algorithms may enable better scientific understanding of drivers of yield variability. However, XAI methods may provide misleading results when applied to spatiotemporal correlated datasets. In this study, machine learning models are trained to predict simulated crop yield from climate indices, and the impact of cross-validation strategy on the interpretation and performance of the resulting models is assessed. Using data from a process-based crop model allows us to then comment on the plausibility of the “explanations” provided by XAI methods. Our results show that the choice of evaluation strategy has an impact on (i) interpretations of the model and (ii) model skill on held-out years and regions, after the evaluation strategy is used for hyperparameter tuning and feature selection. We find that use of a cross-validation strategy based on clustering in feature space achieves the most plausible interpretations as well as the best model performance on held-out years and regions. Our results provide the first steps toward identifying domain-specific “best practices” for the use of XAI tools on spatiotemporal agricultural or climatic data.
Significance Statement
“Explainable” or “interpretable” machine learning (XAI) methods have been increasingly used in scientific research to study complex relationships between climatic and biogeoscientific variables (such as crop yield). However, these methods can return contradictory, implausible, or ambiguous results. In this study, we train machine learning models to predict maize yield anomalies and vary the model evaluation method used. We find that the evaluation (cross validation) method used has an effect on model interpretation results and on the skill of resulting models in held-out years and regions. These results have implications for the methodological design of studies that aim to use XAI tools to identify drivers of, for example, crop yield variability.
Abstract
Ocean surface currents introduce variations into the surface wind-stress that can change the component of the stress aligned with the thermal wind shear at fronts. This modifies the Ekman buoyancy flux, such that the current feedback on the stress tends to generate an effective flux of buoyancy and potential vorticity to the mixed-layer. Scaling arguments and idealized simulations resolving both mesoscale and submesoscale turbulence suggest that this pathway for air-sea interaction can be important both locally at individual submesoscale fronts with strong surface currents—where it can introduce equivalent advective heat fluxes exceeding several hundredWm−2—and in the spatial mean where it reduces the integrated Ekman buoyancy flux by approximately 50%. The accompanying source of surface potential vorticity injection suggests that at some fronts the current feedback modification of the Ekman buoyancy flux may be significant in terms of both submesoscale dynamics and boundary layer energetics, with an implied modification of symmetric instability growth rates and dissipation that scales similarly to the energy lost through the negative wind work generated by the current feedback. This provides an example of how the shift of dynamical regimes into the submesoscale may promote the importance of air-sea interaction mechanisms that differ from those most active at larger scale.
Abstract
Ocean surface currents introduce variations into the surface wind-stress that can change the component of the stress aligned with the thermal wind shear at fronts. This modifies the Ekman buoyancy flux, such that the current feedback on the stress tends to generate an effective flux of buoyancy and potential vorticity to the mixed-layer. Scaling arguments and idealized simulations resolving both mesoscale and submesoscale turbulence suggest that this pathway for air-sea interaction can be important both locally at individual submesoscale fronts with strong surface currents—where it can introduce equivalent advective heat fluxes exceeding several hundredWm−2—and in the spatial mean where it reduces the integrated Ekman buoyancy flux by approximately 50%. The accompanying source of surface potential vorticity injection suggests that at some fronts the current feedback modification of the Ekman buoyancy flux may be significant in terms of both submesoscale dynamics and boundary layer energetics, with an implied modification of symmetric instability growth rates and dissipation that scales similarly to the energy lost through the negative wind work generated by the current feedback. This provides an example of how the shift of dynamical regimes into the submesoscale may promote the importance of air-sea interaction mechanisms that differ from those most active at larger scale.
Abstract
This paper presents the Thunderstorm Nowcasting Tool (ThunderCast), a 24-hour, year round model for predicting the location of convection that is likely to initiate or remain a thunderstorm in the next 0-60 minutes in the continental United States, adapted from existing deep learning convection applications. ThunderCast utilizes a U-Net convolutional neural network for semantic segmentation trained on 320 km by 320 km data patches with four inputs and one target dataset. The inputs are satellite bands from the Geostationary Operational Environmental Satellite (GOES-16) Advanced Baseline Imager (ABI) in the visible, shortwave infrared, and longwave infrared spectrum, and the target is Multi-Radar Multi-Sensor (MRMS) radar reflectivity at the - 10°C isothermin the atmosphere. On a pixel-by-pixel basis, ThunderCast has high accuracy, recall, and specificity but is subject to false positive predictions resulting in low precision. However, the number of false positives decreases when buffering the target values with a 15×15 km centered window indicating ThunderCast’s predictions are useful within a buffered area. To demonstrate the initial prediction capabilities of ThunderCast, three case studies are presented: a mesoscale convective vortex, sea breeze convection, and monsoonal convection in the southwestern United States. The case studies illustrate that the ThunderCast model effectively nowcasts the location of newly initiated and ongoing active convection, within the next 60 minutes, under a variety of geographic and meteorological conditions.
Abstract
This paper presents the Thunderstorm Nowcasting Tool (ThunderCast), a 24-hour, year round model for predicting the location of convection that is likely to initiate or remain a thunderstorm in the next 0-60 minutes in the continental United States, adapted from existing deep learning convection applications. ThunderCast utilizes a U-Net convolutional neural network for semantic segmentation trained on 320 km by 320 km data patches with four inputs and one target dataset. The inputs are satellite bands from the Geostationary Operational Environmental Satellite (GOES-16) Advanced Baseline Imager (ABI) in the visible, shortwave infrared, and longwave infrared spectrum, and the target is Multi-Radar Multi-Sensor (MRMS) radar reflectivity at the - 10°C isothermin the atmosphere. On a pixel-by-pixel basis, ThunderCast has high accuracy, recall, and specificity but is subject to false positive predictions resulting in low precision. However, the number of false positives decreases when buffering the target values with a 15×15 km centered window indicating ThunderCast’s predictions are useful within a buffered area. To demonstrate the initial prediction capabilities of ThunderCast, three case studies are presented: a mesoscale convective vortex, sea breeze convection, and monsoonal convection in the southwestern United States. The case studies illustrate that the ThunderCast model effectively nowcasts the location of newly initiated and ongoing active convection, within the next 60 minutes, under a variety of geographic and meteorological conditions.
Abstract
In this study, an effective method of estimating the volume transport of the Kuroshio Extension (KE) is proposed using surface geostrophic flow inferred from satellite altimetry and vertical stratification derived from climatological temperature/salinity (T/S) profiles. Based on velocity measurements by a subsurface mooring array across the KE, we found that the vertical structure of horizontal flow in this region is dominated by the barotropic and first baroclinic normal modes, which is commendably described by the leading mode of empirical orthogonal functions (EOFs) of the observed velocity profiles as well. Further analysis demonstrates that the projection coefficient of moored velocity onto the superimposed vertical normal mode can be represented by the surface geostrophic velocity as derived from satellite altimetry. Given this relationship, we proposed a dynamical method to estimate the volume transport across the KE jet, which is well verified with both ocean reanalysis and repeated hydrographic data. This finding implicates that, in the regions where the currents render quasi-barotropic structure, it takes only satellite altimetry observation and climatological T/S to estimate the volume transport across any section.
Significance Statement
The Kuroshio Extension (KE) plays an important role in the midlatitude North Pacific climate system. To better understand the KE dynamic and its influences, it is very important to estimate the KE transport. However, direct observation is very difficult in this area. Combining a subsurface mooring array and climatological temperature/salinity data, the vertical structure of the KE is explored in this study using mode decomposition methods. The relationship between the vertical structure of the zonal velocity and surface geostrophic flow observed by satellite altimetry in the KE region is further investigated. Based on this relationship, the KE transport can be well estimated by using satellite altimetry observation and historical hydrographic observation.
Abstract
In this study, an effective method of estimating the volume transport of the Kuroshio Extension (KE) is proposed using surface geostrophic flow inferred from satellite altimetry and vertical stratification derived from climatological temperature/salinity (T/S) profiles. Based on velocity measurements by a subsurface mooring array across the KE, we found that the vertical structure of horizontal flow in this region is dominated by the barotropic and first baroclinic normal modes, which is commendably described by the leading mode of empirical orthogonal functions (EOFs) of the observed velocity profiles as well. Further analysis demonstrates that the projection coefficient of moored velocity onto the superimposed vertical normal mode can be represented by the surface geostrophic velocity as derived from satellite altimetry. Given this relationship, we proposed a dynamical method to estimate the volume transport across the KE jet, which is well verified with both ocean reanalysis and repeated hydrographic data. This finding implicates that, in the regions where the currents render quasi-barotropic structure, it takes only satellite altimetry observation and climatological T/S to estimate the volume transport across any section.
Significance Statement
The Kuroshio Extension (KE) plays an important role in the midlatitude North Pacific climate system. To better understand the KE dynamic and its influences, it is very important to estimate the KE transport. However, direct observation is very difficult in this area. Combining a subsurface mooring array and climatological temperature/salinity data, the vertical structure of the KE is explored in this study using mode decomposition methods. The relationship between the vertical structure of the zonal velocity and surface geostrophic flow observed by satellite altimetry in the KE region is further investigated. Based on this relationship, the KE transport can be well estimated by using satellite altimetry observation and historical hydrographic observation.
Abstract
Across the globe, there has been an increasing interest in improving the predictability of subseasonal hydrometeorological forecasts, as they play a valuable role in medium- to long-term planning in many sectors, such as agriculture, navigation, hydropower, and emergency management. However, these forecasts still have very limited skill at the monthly time scale; hence, this study explores the possibilities for improving forecasts through different pre- and postprocessing techniques at the interface with a Precipitationn–Runoff–Evapotranspiration Hydrological Response Unit Model (PREVAH). Specifically, this research aims to assess the benefit of European weather regime (WR) data within a hybrid forecasting setup, a combination of a traditional hydrological model and a machine learning (ML) algorithm, to improve the performance of subseasonal hydrometeorological forecasts in Switzerland. The WR data contain information about the large-scale atmospheric circulation in the North Atlantic–European region, and thus allow the hydrological model to exploit potential flow-dependent predictability. Four hydrological variables are investigated: total runoff, baseflow, soil moisture, and snowmelt. The improvements in the forecasts achieved with the pre- and postprocessing techniques vary with catchments, lead times, and variables. Adding WR data has clear benefits, but these benefits are not consistent across the study area or among the variables. The usefulness of WR data is generally observed for longer lead times, e.g., beyond the third week. Furthermore, a multimodel approach is applied to determine the “best practice” for each catchment and improve forecast skill over the entire study area. This study highlights the potential and limitations of using WR information to improve subseasonal hydrometeorological forecasts in a hybrid forecasting system in an operational mode.
Abstract
Across the globe, there has been an increasing interest in improving the predictability of subseasonal hydrometeorological forecasts, as they play a valuable role in medium- to long-term planning in many sectors, such as agriculture, navigation, hydropower, and emergency management. However, these forecasts still have very limited skill at the monthly time scale; hence, this study explores the possibilities for improving forecasts through different pre- and postprocessing techniques at the interface with a Precipitationn–Runoff–Evapotranspiration Hydrological Response Unit Model (PREVAH). Specifically, this research aims to assess the benefit of European weather regime (WR) data within a hybrid forecasting setup, a combination of a traditional hydrological model and a machine learning (ML) algorithm, to improve the performance of subseasonal hydrometeorological forecasts in Switzerland. The WR data contain information about the large-scale atmospheric circulation in the North Atlantic–European region, and thus allow the hydrological model to exploit potential flow-dependent predictability. Four hydrological variables are investigated: total runoff, baseflow, soil moisture, and snowmelt. The improvements in the forecasts achieved with the pre- and postprocessing techniques vary with catchments, lead times, and variables. Adding WR data has clear benefits, but these benefits are not consistent across the study area or among the variables. The usefulness of WR data is generally observed for longer lead times, e.g., beyond the third week. Furthermore, a multimodel approach is applied to determine the “best practice” for each catchment and improve forecast skill over the entire study area. This study highlights the potential and limitations of using WR information to improve subseasonal hydrometeorological forecasts in a hybrid forecasting system in an operational mode.