Browse
Abstract
This study assesses the forecast skill of the Canadian Seasonal to Interannual Prediction System (CanSIPS), version 2, in predicting Arctic sea ice extent on both the pan-Arctic and regional scales. In addition, the forecast skill is compared to that of CanSIPS, version 1. Overall, there is a net increase of forecast skill when considering detrended data due to the changes made in the development of CanSIPSv2. The most notable improvements are for forecasts of late summer and autumn target months that have been initialized in the months of April and May that, in previous studies, have been associated with the spring predictability barrier. By comparison of the skills of CanSIPSv1 and CanSIPSv2 to that of an intermediate version of CanSIPS, CanSIPSv1b, we can attribute skill differences between CanSIPSv1 and CanSIPSv2 to two main sources. First, an improved initialization procedure for sea ice initial conditions markedly improves forecast skill on the pan-Arctic scale as well as regionally in the central Arctic, Laptev Sea, Sea of Okhotsk, and Barents Sea. This conclusion is further supported by analysis of the predictive skill of the sea ice volume initialization field. Second, the change in model combination from CanSIPSv1 to CanSIPSv2 (exchanging the constituent CanCM3 model for GEM-NEMO) improves forecast skill in the Bering, Kara, Chukchi, Beaufort, East Siberian, Barents, and the Greenland–Iceland–Norwegian (GIN) Seas. In Hudson and Baffin Bay, as well as the Labrador Sea, there is limited and unsystematic improvement in forecasts of CanSIPSv2 as compared to CanSIPSv1.
Abstract
This study assesses the forecast skill of the Canadian Seasonal to Interannual Prediction System (CanSIPS), version 2, in predicting Arctic sea ice extent on both the pan-Arctic and regional scales. In addition, the forecast skill is compared to that of CanSIPS, version 1. Overall, there is a net increase of forecast skill when considering detrended data due to the changes made in the development of CanSIPSv2. The most notable improvements are for forecasts of late summer and autumn target months that have been initialized in the months of April and May that, in previous studies, have been associated with the spring predictability barrier. By comparison of the skills of CanSIPSv1 and CanSIPSv2 to that of an intermediate version of CanSIPS, CanSIPSv1b, we can attribute skill differences between CanSIPSv1 and CanSIPSv2 to two main sources. First, an improved initialization procedure for sea ice initial conditions markedly improves forecast skill on the pan-Arctic scale as well as regionally in the central Arctic, Laptev Sea, Sea of Okhotsk, and Barents Sea. This conclusion is further supported by analysis of the predictive skill of the sea ice volume initialization field. Second, the change in model combination from CanSIPSv1 to CanSIPSv2 (exchanging the constituent CanCM3 model for GEM-NEMO) improves forecast skill in the Bering, Kara, Chukchi, Beaufort, East Siberian, Barents, and the Greenland–Iceland–Norwegian (GIN) Seas. In Hudson and Baffin Bay, as well as the Labrador Sea, there is limited and unsystematic improvement in forecasts of CanSIPSv2 as compared to CanSIPSv1.
Abstract
Realistic ocean initial conditions are essential for coupled hurricane forecasts. This study focuses on the impact of assimilating high-resolution ocean observations for initialization of the Modular Ocean Model (MOM6) in a coupled configuration with the Hurricane Analysis and Forecast System (HAFS). Based on the Joint Effort for Data Assimilation Integration (JEDI) framework, numerical experiments were performed for the Hurricane Isaias (2020) case, a category-1 hurricane, with use of underwater glider datasets and satellite observations. Assimilation of ocean glider data together with satellite observations provides opportunity to further advance understanding of ocean conditions and air–sea interactions in coupled model initialization and hurricane forecast systems. This comprehensive data assimilation approach has led to a more accurate prediction of the salinity-induced barrier layer thickness that suppresses vertical mixing and sea surface temperature cooling during the storm. Increased barrier layer thickness enhances ocean enthalpy flux into the lower atmosphere and potentially increases tropical cyclone intensity. Assimilation of satellite observations demonstrates improvement in Hurricane Isaias’s intensity forecast. Assimilating glider observations with broad spatial and temporal coverage along Isaias’s track in addition to satellite observations further increase Isaias’s intensity forecast. Overall, this case study demonstrates the importance of assimilating comprehensive marine observations to a more robust ocean and hurricane forecast under a unified JEDI–HAFS hurricane forecast system.
Significance Statement
This is the first comprehensive study of marine observations’ impact on hurricane forecast using marine JEDI. This study found that assimilating satellite observations increases upper-ocean stratification during the prestorm period of Isaias. Assimilating preprocessed observations from six gliders increases salinity-induced upper ocean barrier layer thickness, which reduces sea surface temperature cooling and increases enthalpy flux during the storm. This mechanism eventually enhances hurricane intensity forecast. Overall, this study demonstrates a positive impact of assimilating comprehensive marine observations to a successful ocean and hurricane forecast under a unified JEDI–HAFS hurricane forecast system.
Abstract
Realistic ocean initial conditions are essential for coupled hurricane forecasts. This study focuses on the impact of assimilating high-resolution ocean observations for initialization of the Modular Ocean Model (MOM6) in a coupled configuration with the Hurricane Analysis and Forecast System (HAFS). Based on the Joint Effort for Data Assimilation Integration (JEDI) framework, numerical experiments were performed for the Hurricane Isaias (2020) case, a category-1 hurricane, with use of underwater glider datasets and satellite observations. Assimilation of ocean glider data together with satellite observations provides opportunity to further advance understanding of ocean conditions and air–sea interactions in coupled model initialization and hurricane forecast systems. This comprehensive data assimilation approach has led to a more accurate prediction of the salinity-induced barrier layer thickness that suppresses vertical mixing and sea surface temperature cooling during the storm. Increased barrier layer thickness enhances ocean enthalpy flux into the lower atmosphere and potentially increases tropical cyclone intensity. Assimilation of satellite observations demonstrates improvement in Hurricane Isaias’s intensity forecast. Assimilating glider observations with broad spatial and temporal coverage along Isaias’s track in addition to satellite observations further increase Isaias’s intensity forecast. Overall, this case study demonstrates the importance of assimilating comprehensive marine observations to a more robust ocean and hurricane forecast under a unified JEDI–HAFS hurricane forecast system.
Significance Statement
This is the first comprehensive study of marine observations’ impact on hurricane forecast using marine JEDI. This study found that assimilating satellite observations increases upper-ocean stratification during the prestorm period of Isaias. Assimilating preprocessed observations from six gliders increases salinity-induced upper ocean barrier layer thickness, which reduces sea surface temperature cooling and increases enthalpy flux during the storm. This mechanism eventually enhances hurricane intensity forecast. Overall, this study demonstrates a positive impact of assimilating comprehensive marine observations to a successful ocean and hurricane forecast under a unified JEDI–HAFS hurricane forecast system.
Abstract
We developed a storm surge ensemble prediction system (EPS) for lagoons and transitional environments. Lagoons are often threatened by storm surge events with consequent risks for human life and economic losses. The uncertainties connected with a classic deterministic forecast are many, thus, an ensemble forecast system is required to properly consider them and inform the end-user community accordingly. The technological resources now available allow us to investigate the possibility of operational ensemble forecasting systems that will become increasingly essential for coastal management. We show the advantages and limitations of an EPS applied to a lagoon, using a very high-resolution unstructured grid finite element model and 45 EPS members. For five recent storm surge events, the EPS generally improves the forecast skill on the third forecast day compared to just one deterministic forecast, while they are similar in the first two days. A weighting system is implemented to compute an improved ensemble mean. The uncertainties regarding sea level due to meteorological forcing, river runoff, initial boundaries, and lateral boundaries are evaluated for a special case in the northern Adriatic Sea, and the different forecasts are used to compose the EPS members. We conclude that the largest uncertainty is in the initial and lateral boundary fields at different time and space scales, including the tidal components.
Significance Statement
Storm surges are extreme sea level events that may threaten densely populated coastal areas. The purpose of this work is to improve the extreme sea level forecast for transitional areas with the understanding of what are the most important forcing generating uncertainties and find a technique to reach a reliable sea level forecast. This is achieved by implementing an ensemble prediction system running 45 members for each event considered. Results show that initial and lateral boundary conditions provide most of the uncertainty, including the tidal components. The weighting system applied to find the ensemble mean improves the forecast skill on the third forecast day while it is comparable with the deterministic forecast in the first two days.
Abstract
We developed a storm surge ensemble prediction system (EPS) for lagoons and transitional environments. Lagoons are often threatened by storm surge events with consequent risks for human life and economic losses. The uncertainties connected with a classic deterministic forecast are many, thus, an ensemble forecast system is required to properly consider them and inform the end-user community accordingly. The technological resources now available allow us to investigate the possibility of operational ensemble forecasting systems that will become increasingly essential for coastal management. We show the advantages and limitations of an EPS applied to a lagoon, using a very high-resolution unstructured grid finite element model and 45 EPS members. For five recent storm surge events, the EPS generally improves the forecast skill on the third forecast day compared to just one deterministic forecast, while they are similar in the first two days. A weighting system is implemented to compute an improved ensemble mean. The uncertainties regarding sea level due to meteorological forcing, river runoff, initial boundaries, and lateral boundaries are evaluated for a special case in the northern Adriatic Sea, and the different forecasts are used to compose the EPS members. We conclude that the largest uncertainty is in the initial and lateral boundary fields at different time and space scales, including the tidal components.
Significance Statement
Storm surges are extreme sea level events that may threaten densely populated coastal areas. The purpose of this work is to improve the extreme sea level forecast for transitional areas with the understanding of what are the most important forcing generating uncertainties and find a technique to reach a reliable sea level forecast. This is achieved by implementing an ensemble prediction system running 45 members for each event considered. Results show that initial and lateral boundary conditions provide most of the uncertainty, including the tidal components. The weighting system applied to find the ensemble mean improves the forecast skill on the third forecast day while it is comparable with the deterministic forecast in the first two days.
Abstract
The assimilation of two surface-sensitive channels of the AMSU-A instruments on board the NOAA-15/NOAA-18/NOAA-19 and MetOp-A/MetOp-B satellites over land was achieved in the China Meteorological Administration Global Forecast System (CMA_GFS). The land surface emissivity was calculated by 1) the window channel retrieval method and 2) the Tool to Estimate Land Surface Emissivities at Microwave frequencies (TELSEM2). Quality controls for these satellite microwave observations over land were conducted. The predictors and regression coefficients used for oceanic satellite data were retained during the bias correction over land and found to perform well. Three batch experiments were implemented in CMA_GFS with 4D-Var: 1) assimilating only the default data, and adding the above data over land with land surface emissivity obtained from 2) TELSEM2 and 3) the window channel retrieval method. The results indicated that the window channel retrieval method can better reduce the departure between the observed and simulated brightness temperature. Over most land types, the positive impacts of this method exceed those of TELSEM2. Both TELSEM2 and the window channel retrieval method improve the humidity analysis near the ground, as well as the forecast capability globally, particularly in those regions where the land coverage is greater, such as in the Northern Hemisphere. The data utilization of the two surface-sensitive channels increase by 6% and 12%, respectively, and the additional data every 6 h can cover most land, where there was no surface-sensitive data assimilated before. This study marks the beginning of near-surface channel assimilation over land in CMA_GFS and represents a breakthrough in the assimilation of other surface-sensitive channels in other satellite instruments.
Significance Statement
Surface-sensitive microwave channels are difficult to assimilate in NWP due to the lack of both direct measurement and appropriate modeling for instantaneous land surface emissivity. This paper discusses a method that improves the surface emissivity estimates, which has allowed the utilization of surface-sensitive microwave channels in CMA_GFS. Those capabilities have resulted in better data utilization, improved forecasts of temperature, geopotential height, and winds in the Northern Hemisphere at 3–7 days, and represent an incremental and important improvement to CMA_GFS.
Abstract
The assimilation of two surface-sensitive channels of the AMSU-A instruments on board the NOAA-15/NOAA-18/NOAA-19 and MetOp-A/MetOp-B satellites over land was achieved in the China Meteorological Administration Global Forecast System (CMA_GFS). The land surface emissivity was calculated by 1) the window channel retrieval method and 2) the Tool to Estimate Land Surface Emissivities at Microwave frequencies (TELSEM2). Quality controls for these satellite microwave observations over land were conducted. The predictors and regression coefficients used for oceanic satellite data were retained during the bias correction over land and found to perform well. Three batch experiments were implemented in CMA_GFS with 4D-Var: 1) assimilating only the default data, and adding the above data over land with land surface emissivity obtained from 2) TELSEM2 and 3) the window channel retrieval method. The results indicated that the window channel retrieval method can better reduce the departure between the observed and simulated brightness temperature. Over most land types, the positive impacts of this method exceed those of TELSEM2. Both TELSEM2 and the window channel retrieval method improve the humidity analysis near the ground, as well as the forecast capability globally, particularly in those regions where the land coverage is greater, such as in the Northern Hemisphere. The data utilization of the two surface-sensitive channels increase by 6% and 12%, respectively, and the additional data every 6 h can cover most land, where there was no surface-sensitive data assimilated before. This study marks the beginning of near-surface channel assimilation over land in CMA_GFS and represents a breakthrough in the assimilation of other surface-sensitive channels in other satellite instruments.
Significance Statement
Surface-sensitive microwave channels are difficult to assimilate in NWP due to the lack of both direct measurement and appropriate modeling for instantaneous land surface emissivity. This paper discusses a method that improves the surface emissivity estimates, which has allowed the utilization of surface-sensitive microwave channels in CMA_GFS. Those capabilities have resulted in better data utilization, improved forecasts of temperature, geopotential height, and winds in the Northern Hemisphere at 3–7 days, and represent an incremental and important improvement to CMA_GFS.
Abstract
This study evaluates the representation of tropical cyclone precipitation (TCP) in reforecasts from the Subseasonal to Seasonal (S2S) Prediction Project. The global distribution of precipitation in S2S models shows relevant biases in the multimodel mean ensemble that are characterized by wet biases in total precipitation and TCP, except for the Atlantic. The TCP biases can contribute more than 50% of the total precipitation biases in basins such as the southern Indian Ocean and South Pacific. The magnitude and spatial pattern of these biases exhibit little variation with lead time. The origins of TCP biases can be attributed to biases in the frequency of tropical cyclone occurrence. The S2S models simulate too few TCs in the Atlantic and western North Pacific and too many TCs in the Southern Hemisphere and eastern North Pacific. At the storm scale, the average peak precipitation near the storm center is lower in the models than observations due to a too high proportion of weak TCs. However, this bias is offset in some models by higher than observed precipitation rates at larger radii (300–500 km). An analysis of the mean TCP for each TC at each grid point reveals an overestimation of TCP rates, particularly in the near-equatorial Indian and western Pacific Oceans. These findings suggest that the simulation of TC occurrence and the storm-scale precipitation require better representation in order to reduce TCP biases and enhance the subseasonal prediction skill of mean and extreme total precipitation.
Abstract
This study evaluates the representation of tropical cyclone precipitation (TCP) in reforecasts from the Subseasonal to Seasonal (S2S) Prediction Project. The global distribution of precipitation in S2S models shows relevant biases in the multimodel mean ensemble that are characterized by wet biases in total precipitation and TCP, except for the Atlantic. The TCP biases can contribute more than 50% of the total precipitation biases in basins such as the southern Indian Ocean and South Pacific. The magnitude and spatial pattern of these biases exhibit little variation with lead time. The origins of TCP biases can be attributed to biases in the frequency of tropical cyclone occurrence. The S2S models simulate too few TCs in the Atlantic and western North Pacific and too many TCs in the Southern Hemisphere and eastern North Pacific. At the storm scale, the average peak precipitation near the storm center is lower in the models than observations due to a too high proportion of weak TCs. However, this bias is offset in some models by higher than observed precipitation rates at larger radii (300–500 km). An analysis of the mean TCP for each TC at each grid point reveals an overestimation of TCP rates, particularly in the near-equatorial Indian and western Pacific Oceans. These findings suggest that the simulation of TC occurrence and the storm-scale precipitation require better representation in order to reduce TCP biases and enhance the subseasonal prediction skill of mean and extreme total precipitation.
Abstract
Parametric models of tropical cyclone winds are widely used for risk assessment. Although tropical cyclones often present their worst wind risk to humanity during landfall, parametric models that represent land–sea differences are rare. This paper presents a parametric model with explicit representation of land–sea differences. Statistical models were developed over each surface of the frictional wind speed reduction from gradient level to 10 m, and of the surface inflow angle, based on about 1200 simulations with a three-dimensional dynamical boundary layer model. The wind profile of Willoughby et al. is used to represent the gradient flow, and a maximum likelihood scheme used to fit this profile to best track data. The mean RMS difference between the statistical and dynamical surface winds within 100 km of the storm center is 0.78 m s−1 and 4.26° over sea, and 1.04 m s−1 and 4.59° over land. During landfall, the use of a common gradient-level structure, but different surface roughnesses, provides dynamical consistency between the estimated winds over sea and land. A simple representation of internal boundary layers is applied near the coast. Analysis of the dynamical simulations revealed substantial consistency with observational studies of the tropical cyclone boundary layer, including that the azimuth of the surface wind maximum is on average 65° from the front of the storm, in the left-forward quadrant in the Southern Hemisphere. There was, however, substantial variability around this figure, with the maximum occurring in the opposite forward quadrant in storms that were intense, and/or had a relatively rapid decrease in wind speed outside of the radius of maximum winds.
Abstract
Parametric models of tropical cyclone winds are widely used for risk assessment. Although tropical cyclones often present their worst wind risk to humanity during landfall, parametric models that represent land–sea differences are rare. This paper presents a parametric model with explicit representation of land–sea differences. Statistical models were developed over each surface of the frictional wind speed reduction from gradient level to 10 m, and of the surface inflow angle, based on about 1200 simulations with a three-dimensional dynamical boundary layer model. The wind profile of Willoughby et al. is used to represent the gradient flow, and a maximum likelihood scheme used to fit this profile to best track data. The mean RMS difference between the statistical and dynamical surface winds within 100 km of the storm center is 0.78 m s−1 and 4.26° over sea, and 1.04 m s−1 and 4.59° over land. During landfall, the use of a common gradient-level structure, but different surface roughnesses, provides dynamical consistency between the estimated winds over sea and land. A simple representation of internal boundary layers is applied near the coast. Analysis of the dynamical simulations revealed substantial consistency with observational studies of the tropical cyclone boundary layer, including that the azimuth of the surface wind maximum is on average 65° from the front of the storm, in the left-forward quadrant in the Southern Hemisphere. There was, however, substantial variability around this figure, with the maximum occurring in the opposite forward quadrant in storms that were intense, and/or had a relatively rapid decrease in wind speed outside of the radius of maximum winds.
Abstract
The Hurricane Analysis and Forecasting System (HAFS) is the next-generation, FV3-based tropical cyclone (TC) forecasting system. Unlike operational implementations of NOAA’s Hurricane Weather Research and Forecast (HWRF) modeling system, current data assimilation (DA) capabilities in HAFS permit the uninterrupted basin-wide assimilation of measurements. This feature of HAFS opens a variety of new research directions for TC prediction, including new strategies for DA algorithm development and self-contained probabilistic forecasting. The present research focuses more narrowly on new opportunities HAFS brings for optimizing the use of satellite measurements for TC prediction. While satellite radiometers provide a wealth of information for characterizing temperature, moisture, and wind in TC environments, the provided measurements are often biased and contain unknown cross-channel error correlations. For mature global modeling systems, these statistics are estimated from information gathered during DA, namely, innovations collected over large spatial and temporal training periods. The estimated statistics, however, are imperfect owing to unknown error sources such as model process error, which are difficult to separate from observation error. As such, bias and uncertainty specifications that rely on information from external models are suboptimal—as is the current strategy for HWRF. In this paper, it will be demonstrated that bias estimation for satellite radiance observations is particularly sensitive to common design choices, such as using a bias model trained from the Global Data Assimilation System instead of within the native modeling system. Implications of this finding for TC prediction are examined over a 6-week period from 2020, which included the development and intensification of nine tropical cyclones.
Significance Statement
Tropical cyclone–focused numerical weather prediction is difficult due to complex nonlinear physical processes and a lack of in situ observations over open ocean. Prediction systems rely heavily on satellite radiance measurements, which have high spatial–temporal resolution over the entire domain but require bias correction. Estimation of observation bias requires long training periods and large spatial domain coverage, which is typically not permitted outside of global models. However, bias specification is strongly model dependent, as bias correction methods cannot easily separate model and observation bias. In this study, we perform satellite radiance bias specification internally for an experimental version of the NOAA Hurricane Analysis and Forecast System and demonstrate major implications for tropical cyclone prediction.
Abstract
The Hurricane Analysis and Forecasting System (HAFS) is the next-generation, FV3-based tropical cyclone (TC) forecasting system. Unlike operational implementations of NOAA’s Hurricane Weather Research and Forecast (HWRF) modeling system, current data assimilation (DA) capabilities in HAFS permit the uninterrupted basin-wide assimilation of measurements. This feature of HAFS opens a variety of new research directions for TC prediction, including new strategies for DA algorithm development and self-contained probabilistic forecasting. The present research focuses more narrowly on new opportunities HAFS brings for optimizing the use of satellite measurements for TC prediction. While satellite radiometers provide a wealth of information for characterizing temperature, moisture, and wind in TC environments, the provided measurements are often biased and contain unknown cross-channel error correlations. For mature global modeling systems, these statistics are estimated from information gathered during DA, namely, innovations collected over large spatial and temporal training periods. The estimated statistics, however, are imperfect owing to unknown error sources such as model process error, which are difficult to separate from observation error. As such, bias and uncertainty specifications that rely on information from external models are suboptimal—as is the current strategy for HWRF. In this paper, it will be demonstrated that bias estimation for satellite radiance observations is particularly sensitive to common design choices, such as using a bias model trained from the Global Data Assimilation System instead of within the native modeling system. Implications of this finding for TC prediction are examined over a 6-week period from 2020, which included the development and intensification of nine tropical cyclones.
Significance Statement
Tropical cyclone–focused numerical weather prediction is difficult due to complex nonlinear physical processes and a lack of in situ observations over open ocean. Prediction systems rely heavily on satellite radiance measurements, which have high spatial–temporal resolution over the entire domain but require bias correction. Estimation of observation bias requires long training periods and large spatial domain coverage, which is typically not permitted outside of global models. However, bias specification is strongly model dependent, as bias correction methods cannot easily separate model and observation bias. In this study, we perform satellite radiance bias specification internally for an experimental version of the NOAA Hurricane Analysis and Forecast System and demonstrate major implications for tropical cyclone prediction.
Abstract
We used object-oriented verification and self-organizing maps (SOMs) to identify patterns in environmental parameters correlating with mesoscale snowband predictive skill by the High-Resolution Ensemble Forecast (HREF) system between 2017 and 2022. First, HREF snowband forecasts for 305 banding events were verified based on similarities between forecast and observed feature properties. HREF members performed comparably, demonstrating large positional errors, but the non-time-lagged High-Resolution Rapid Refresh member demonstrated the greatest overall skill. Observed banding events were clustered by 500-hPa geopotential height anomalies, mean sea level pressure, vertical velocity, frontogenesis, and saturation equivalent potential vorticity from the European Centre for Medium-Range Weather Forecasts Reanalysis version 5 using SOMs. Clusters reaffirmed the presence of midlevel frontogenesis, ascent, and reduced stability in most banding cases, and the predominant synoptic environments conducive to band development. Clusters were compared to determine whether patterns in the variables were correlated with predictive skill. Strength of upward motion was correlated with skill, with the strongest upward motion cases verifying 10% better than the weakest upward motion cases due to smaller positional error. Additionally, events with a single region of strong upward motion verified better than events with disorganized, but comparably intense, upward motion. The magnitude of frontogenesis was uncorrelated with skill, but events with more upright frontogenesis collocated with the band centroid were better predicted than events with shallower slopes and low-level frontogenesis displaced toward warmer air. The skill variance associated with different vertical motion magnitudes could assist forecasters in modulating forecast confidence, while the most common types of errors documented here may be beneficial to model developers in refining HREF member snowfall forecasts.
Significance Statement
High-resolution numerical weather prediction (NWP) models generally have limited predictive skill for mesoscale snowband forecasts. Even so, some snowbands are forecast by NWP models with much greater skill than others. In this work, we apply artificial intelligence to group snowband events based on atmospheric conditions and then determine whether different groups are easier or harder for models to predict. Identification of these groups could help forecasters know when to trust or be skeptical of NWP output and help developers improve snowband formation processes in NWP models.
Abstract
We used object-oriented verification and self-organizing maps (SOMs) to identify patterns in environmental parameters correlating with mesoscale snowband predictive skill by the High-Resolution Ensemble Forecast (HREF) system between 2017 and 2022. First, HREF snowband forecasts for 305 banding events were verified based on similarities between forecast and observed feature properties. HREF members performed comparably, demonstrating large positional errors, but the non-time-lagged High-Resolution Rapid Refresh member demonstrated the greatest overall skill. Observed banding events were clustered by 500-hPa geopotential height anomalies, mean sea level pressure, vertical velocity, frontogenesis, and saturation equivalent potential vorticity from the European Centre for Medium-Range Weather Forecasts Reanalysis version 5 using SOMs. Clusters reaffirmed the presence of midlevel frontogenesis, ascent, and reduced stability in most banding cases, and the predominant synoptic environments conducive to band development. Clusters were compared to determine whether patterns in the variables were correlated with predictive skill. Strength of upward motion was correlated with skill, with the strongest upward motion cases verifying 10% better than the weakest upward motion cases due to smaller positional error. Additionally, events with a single region of strong upward motion verified better than events with disorganized, but comparably intense, upward motion. The magnitude of frontogenesis was uncorrelated with skill, but events with more upright frontogenesis collocated with the band centroid were better predicted than events with shallower slopes and low-level frontogenesis displaced toward warmer air. The skill variance associated with different vertical motion magnitudes could assist forecasters in modulating forecast confidence, while the most common types of errors documented here may be beneficial to model developers in refining HREF member snowfall forecasts.
Significance Statement
High-resolution numerical weather prediction (NWP) models generally have limited predictive skill for mesoscale snowband forecasts. Even so, some snowbands are forecast by NWP models with much greater skill than others. In this work, we apply artificial intelligence to group snowband events based on atmospheric conditions and then determine whether different groups are easier or harder for models to predict. Identification of these groups could help forecasters know when to trust or be skeptical of NWP output and help developers improve snowband formation processes in NWP models.
