Browse
Abstract
Accurate subseasonal-to-seasonal (S2S) weather forecasts are crucial to making important decisions in many sectors. However, significant gaps exist between the needs of society and what forecasters can produce, especially at weekly and longer lead times. We hypothesize that by clustering atmospheric states into a number of predefined categories, the noise can be reduced and, consequently, medium-range forecasts can be improved. Self-organizing map (SOM)-based clustering was used on daily mean sea level pressure (MSLP) data from the North American Regional Reanalysis to categorize the synoptic-scale circulation for eastern North America from 1979 to 2016 into 28 discrete patterns. Then, using two goodness-of-fit metrics, the relative skill of four different forecasting methods over a 90-day lead time was studied: 1) a circulation pattern (CP) forecast, 2) raw forecast output from the Climate Forecast System (CFS) operated by the National Centers for Environmental Prediction (NCEP), 3) a simple climatology forecast, and 4) a simple persistence forecast. As expected, forecast skill of both the CP forecast and the raw CFS forecast generally decreased rapidly from the first day, coming to parity with the skill of climatology after 10–12 days when using correlation, and at 7–10 days when using the root-mean-square error (RMSE). Most importantly, this study found that the CP forecast was the most skillful forecast method over the 8–11-day lead time when using RMSE. On a spatial basis, the skill of the CP forecast and the raw CFS decreases latitudinally from north to south. This study thus demonstrates the potential utility of categorical or circulation pattern–based forecasting at 1–2-week lead times.
Abstract
Accurate subseasonal-to-seasonal (S2S) weather forecasts are crucial to making important decisions in many sectors. However, significant gaps exist between the needs of society and what forecasters can produce, especially at weekly and longer lead times. We hypothesize that by clustering atmospheric states into a number of predefined categories, the noise can be reduced and, consequently, medium-range forecasts can be improved. Self-organizing map (SOM)-based clustering was used on daily mean sea level pressure (MSLP) data from the North American Regional Reanalysis to categorize the synoptic-scale circulation for eastern North America from 1979 to 2016 into 28 discrete patterns. Then, using two goodness-of-fit metrics, the relative skill of four different forecasting methods over a 90-day lead time was studied: 1) a circulation pattern (CP) forecast, 2) raw forecast output from the Climate Forecast System (CFS) operated by the National Centers for Environmental Prediction (NCEP), 3) a simple climatology forecast, and 4) a simple persistence forecast. As expected, forecast skill of both the CP forecast and the raw CFS forecast generally decreased rapidly from the first day, coming to parity with the skill of climatology after 10–12 days when using correlation, and at 7–10 days when using the root-mean-square error (RMSE). Most importantly, this study found that the CP forecast was the most skillful forecast method over the 8–11-day lead time when using RMSE. On a spatial basis, the skill of the CP forecast and the raw CFS decreases latitudinally from north to south. This study thus demonstrates the potential utility of categorical or circulation pattern–based forecasting at 1–2-week lead times.
Abstract
A 5-yr climatology and composite study of precipitation bands associated with extratropical cyclones over the British Isles from April 2017 to March 2022 is constructed. A total of 249 single bands were manually identified from radar network mosaics in association with 167 cyclones identified from surface maps. More bands formed over water near the coast than over inland areas, and most had a meridional orientation. The average lengths of bands at the times of formation and maximum length were 290 and 460 km, respectively; only 20% of bands reached a maximum length exceeding 600 km. The number of bands decreased with increasing duration, with 31% of bands lasting for 2–3 h, with bands lasting more than 10 h uncommon. The bands were classified into six categories, with occluded-frontal bands (19 yr−1), warm-frontal bands (11 yr−1), and cold-frontal bands (10 yr−1) being the most frequent. Occluded-frontal and warm-frontal bands commonly occurred west of Scotland and in the east quadrant relative to their parent cyclones. In contrast, cold-frontal bands commonly occurred southwest of Great Britain and in the south quadrant relative to their parent cyclones. Composites for northwest–southeast occluded-frontal and warm-frontal bands west of Scotland, and southwest–northeast cold-frontal bands southwest of Great Britain, show the different synoptic environments that favor bands. The low-level jet transports moisture into the band and is similar to the location and scale of the composite bands, similar to that of an atmospheric river. These results are compared to previous studies on bands from the United States.
Significance Statement
Precipitation bands are lines of heavy precipitation as seen on weather radar. Most studies of bands in extratropical cyclones have occurred in the United States. We examine 5 years of bands in extratropical cyclones over the British Isles to better understand their characteristics. Bands form in preferred geographic regions: offshore of the west coasts of Scotland, Wales, and southwest England. The most common bands are associated with occluded fronts (37% of all bands). The average scale of the bands is associated with the average scale of wind maxima 1–2 km above ground. These results provide a better understanding of the typical characteristics and conditions under which bands form and their geographical variability compared to the United States.
Abstract
A 5-yr climatology and composite study of precipitation bands associated with extratropical cyclones over the British Isles from April 2017 to March 2022 is constructed. A total of 249 single bands were manually identified from radar network mosaics in association with 167 cyclones identified from surface maps. More bands formed over water near the coast than over inland areas, and most had a meridional orientation. The average lengths of bands at the times of formation and maximum length were 290 and 460 km, respectively; only 20% of bands reached a maximum length exceeding 600 km. The number of bands decreased with increasing duration, with 31% of bands lasting for 2–3 h, with bands lasting more than 10 h uncommon. The bands were classified into six categories, with occluded-frontal bands (19 yr−1), warm-frontal bands (11 yr−1), and cold-frontal bands (10 yr−1) being the most frequent. Occluded-frontal and warm-frontal bands commonly occurred west of Scotland and in the east quadrant relative to their parent cyclones. In contrast, cold-frontal bands commonly occurred southwest of Great Britain and in the south quadrant relative to their parent cyclones. Composites for northwest–southeast occluded-frontal and warm-frontal bands west of Scotland, and southwest–northeast cold-frontal bands southwest of Great Britain, show the different synoptic environments that favor bands. The low-level jet transports moisture into the band and is similar to the location and scale of the composite bands, similar to that of an atmospheric river. These results are compared to previous studies on bands from the United States.
Significance Statement
Precipitation bands are lines of heavy precipitation as seen on weather radar. Most studies of bands in extratropical cyclones have occurred in the United States. We examine 5 years of bands in extratropical cyclones over the British Isles to better understand their characteristics. Bands form in preferred geographic regions: offshore of the west coasts of Scotland, Wales, and southwest England. The most common bands are associated with occluded fronts (37% of all bands). The average scale of the bands is associated with the average scale of wind maxima 1–2 km above ground. These results provide a better understanding of the typical characteristics and conditions under which bands form and their geographical variability compared to the United States.
Abstract
A severe derecho impacted the Midwestern United States on 10 August 2020, causing over $12 billion (U.S. dollars) in damage, and producing peak winds estimated at 63 m s−1, with the worst impacts in Iowa. The event was not forecast well by operational forecasters, nor even by operational and quasi-operational convection-allowing models. In the present study, nine simulations are performed using the Limited Area Model version of the Finite-Volume-Cubed-Sphere model (FV3-LAM) with three horizontal grid spacings and two physics suites. In addition, when a prototype of the Rapid Refresh Forecast System (RRFS) physics is used, sensitivity tests are performed to examine the impact of using the Grell–Freitas (GF) convective scheme. Several unusual results are obtained. With both the RRFS (not using GF) and Global Forecast System (GFS) physics suites, simulations using relatively coarse 13- and 25-km horizontal grid spacing do a much better job of showing an organized convective system in Iowa during the daylight hours of 10 August than the 3-km grid spacing runs. In addition, the RRFS run with 25-km grid spacing becomes much worse when the GF convective scheme is used. The 3-km RRFS run that does not use the GF scheme develops spurious nocturnal convection the night before the derecho, removing instability and preventing the derecho from being simulated at all. When GF is used, the spurious storms are removed and an excellent forecast is obtained with an intense bowing echo, exceptionally strong cold pool, and roughly 50 m s−1 surface wind gusts.
Abstract
A severe derecho impacted the Midwestern United States on 10 August 2020, causing over $12 billion (U.S. dollars) in damage, and producing peak winds estimated at 63 m s−1, with the worst impacts in Iowa. The event was not forecast well by operational forecasters, nor even by operational and quasi-operational convection-allowing models. In the present study, nine simulations are performed using the Limited Area Model version of the Finite-Volume-Cubed-Sphere model (FV3-LAM) with three horizontal grid spacings and two physics suites. In addition, when a prototype of the Rapid Refresh Forecast System (RRFS) physics is used, sensitivity tests are performed to examine the impact of using the Grell–Freitas (GF) convective scheme. Several unusual results are obtained. With both the RRFS (not using GF) and Global Forecast System (GFS) physics suites, simulations using relatively coarse 13- and 25-km horizontal grid spacing do a much better job of showing an organized convective system in Iowa during the daylight hours of 10 August than the 3-km grid spacing runs. In addition, the RRFS run with 25-km grid spacing becomes much worse when the GF convective scheme is used. The 3-km RRFS run that does not use the GF scheme develops spurious nocturnal convection the night before the derecho, removing instability and preventing the derecho from being simulated at all. When GF is used, the spurious storms are removed and an excellent forecast is obtained with an intense bowing echo, exceptionally strong cold pool, and roughly 50 m s−1 surface wind gusts.
Abstract
Eyewall replacement cycles (ERCs) in tropical cyclones (TCs) are generally associated with rapid changes in TC wind intensity and broadening of the TC wind field, both of which can create unique forecasting challenges. As part of the NOAA Joint Hurricane Testbed Project, a new model was developed to provide operational probabilistic guidance on ERC onset. The model is based on the time evolution of TC wind intensity and passive satellite microwave imagery and is named “M-PERC” for Microwave-Based Probability of Eyewall Replacement Cycle. The model was initially developed in the Atlantic basin but is found to be globally applicable and skillful. The development of M-PERC and its performance characteristics are described here, as well as a new intensity prediction model that extends previous work. Application of these models is expected to contribute to a reduction of TC intensity forecast error.
Abstract
Eyewall replacement cycles (ERCs) in tropical cyclones (TCs) are generally associated with rapid changes in TC wind intensity and broadening of the TC wind field, both of which can create unique forecasting challenges. As part of the NOAA Joint Hurricane Testbed Project, a new model was developed to provide operational probabilistic guidance on ERC onset. The model is based on the time evolution of TC wind intensity and passive satellite microwave imagery and is named “M-PERC” for Microwave-Based Probability of Eyewall Replacement Cycle. The model was initially developed in the Atlantic basin but is found to be globally applicable and skillful. The development of M-PERC and its performance characteristics are described here, as well as a new intensity prediction model that extends previous work. Application of these models is expected to contribute to a reduction of TC intensity forecast error.
Abstract
Accurate and reliable seasonal forecasts are important for water and energy supply management. Recognizing the important role of snow water equivalent (SWE) for water management, here we include the seasonal forecast of SWE in addition to precipitation (P) and 2-m temperature (T2m) over hydrologically defined regions of the western United States. A two-stage process is applied to seasonal predictions from two models (NCEP CFSv2 and ECMWF SEAS5) through 1) postprocessing to remove biases in the mean, variance, and ensemble spread and 2) further reducing the residual errors by linear regression using climate indices. The adjusted forecasts from the two models are combined to form a superensemble using weights based on their prior skill. The adjusted forecasts are consistently improved over raw model forecasts probabilistically for all variables and deterministically for SWE forecasts. Overall skill of the superensemble usually improves upon the skill of forecasts from individual models; however, the percentage of seasons and regions with increased skill was approximately the same as those with decreased skill relative to the top performing postprocessed individual model. Seasonal SWE has the highest prediction skill, followed by T2m, with P showing lower prediction skill. Persistence contributes strongly to the skill of SWE and moderately to the skill of T2m. Furthermore, a distinct seasonality in the skill is seen in SWE, with a higher skill from late spring through early summer.
Significance Statement
Here we test the postprocessing of seasonal forecasts from two state-of-the-art seasonal prediction models for traditionally forecasted elements of precipitation and temperature as well as snowpack, which is important for water management. A two-stage procedure is utilized, including ocean and atmospheric teleconnection indices that have been shown to impact seasonal weather across the western United States. First, we adjust model output based on the average error in historic runs and then relate the remaining error to these teleconnection indices. A final step combines each adjusted model based on its historic performance. Forecasts are shown to improve upon the original models when assessed probabilistically. The snowpack forecasts perform better than temperature and precipitation forecasts with the best performance from late winter through early summer. Persistence is found to contribute strongly to the skill of snowpack and moderately to the skill of temperature.
Abstract
Accurate and reliable seasonal forecasts are important for water and energy supply management. Recognizing the important role of snow water equivalent (SWE) for water management, here we include the seasonal forecast of SWE in addition to precipitation (P) and 2-m temperature (T2m) over hydrologically defined regions of the western United States. A two-stage process is applied to seasonal predictions from two models (NCEP CFSv2 and ECMWF SEAS5) through 1) postprocessing to remove biases in the mean, variance, and ensemble spread and 2) further reducing the residual errors by linear regression using climate indices. The adjusted forecasts from the two models are combined to form a superensemble using weights based on their prior skill. The adjusted forecasts are consistently improved over raw model forecasts probabilistically for all variables and deterministically for SWE forecasts. Overall skill of the superensemble usually improves upon the skill of forecasts from individual models; however, the percentage of seasons and regions with increased skill was approximately the same as those with decreased skill relative to the top performing postprocessed individual model. Seasonal SWE has the highest prediction skill, followed by T2m, with P showing lower prediction skill. Persistence contributes strongly to the skill of SWE and moderately to the skill of T2m. Furthermore, a distinct seasonality in the skill is seen in SWE, with a higher skill from late spring through early summer.
Significance Statement
Here we test the postprocessing of seasonal forecasts from two state-of-the-art seasonal prediction models for traditionally forecasted elements of precipitation and temperature as well as snowpack, which is important for water management. A two-stage procedure is utilized, including ocean and atmospheric teleconnection indices that have been shown to impact seasonal weather across the western United States. First, we adjust model output based on the average error in historic runs and then relate the remaining error to these teleconnection indices. A final step combines each adjusted model based on its historic performance. Forecasts are shown to improve upon the original models when assessed probabilistically. The snowpack forecasts perform better than temperature and precipitation forecasts with the best performance from late winter through early summer. Persistence is found to contribute strongly to the skill of snowpack and moderately to the skill of temperature.
Abstract
Improving estimates of tropical cyclone forecast uncertainty remains an important goal of the Hurricane Forecast Improvement Project (HFIP). Intensity forecast uncertainty near landfall is especially complicated because intensity forecasts depend on track forecasts. Ensembles can be difficult to interpret near land due to differences in both spatial and temporal resolution and differences in landfall timing (if at all) and location. The Monte Carlo Wind Speed Probability (WSP) model is a statistical ensemble based on the error characteristics of forecasts by the National Hurricane Center (NHC) and the spread of several track forecast models. The landfall distribution product (LDP) introduced in this paper was developed to use the statistical ensemble of forecasts from the WSP model to estimate both the track and intensity forecast uncertainty associated with potential landfalls. The LDP includes probabilistic intensity estimates as well as estimates of the most likely and reasonable strongest intensity at landfall. These products could communicate concise intensity uncertainty information to users at risk for tropical cyclone impacts. Demonstration on a retrospective dataset from 2010 to 2018 and evaluation of the LDP on the 2020–21 Atlantic hurricane seasons shows that the probability of landfall and the landfall intensity probabilities generated by the WSP model are reliable and potentially useful for preparedness decision-making. A case study of Hurricane Ida (2021) highlights how the LDP can be implemented to communicate landfall uncertainty to a broad range of users.
Significance Statement
With the new landfall distribution product (LDP), the National Hurricane Center can provide both track and intensity forecast uncertainty surrounding the landfall of hurricanes. The issuance of a reasonable worst case scenario for the strongest winds that could impact a region could amplify messaging to encourage people to take appropriate action prior to a landfall.
Abstract
Improving estimates of tropical cyclone forecast uncertainty remains an important goal of the Hurricane Forecast Improvement Project (HFIP). Intensity forecast uncertainty near landfall is especially complicated because intensity forecasts depend on track forecasts. Ensembles can be difficult to interpret near land due to differences in both spatial and temporal resolution and differences in landfall timing (if at all) and location. The Monte Carlo Wind Speed Probability (WSP) model is a statistical ensemble based on the error characteristics of forecasts by the National Hurricane Center (NHC) and the spread of several track forecast models. The landfall distribution product (LDP) introduced in this paper was developed to use the statistical ensemble of forecasts from the WSP model to estimate both the track and intensity forecast uncertainty associated with potential landfalls. The LDP includes probabilistic intensity estimates as well as estimates of the most likely and reasonable strongest intensity at landfall. These products could communicate concise intensity uncertainty information to users at risk for tropical cyclone impacts. Demonstration on a retrospective dataset from 2010 to 2018 and evaluation of the LDP on the 2020–21 Atlantic hurricane seasons shows that the probability of landfall and the landfall intensity probabilities generated by the WSP model are reliable and potentially useful for preparedness decision-making. A case study of Hurricane Ida (2021) highlights how the LDP can be implemented to communicate landfall uncertainty to a broad range of users.
Significance Statement
With the new landfall distribution product (LDP), the National Hurricane Center can provide both track and intensity forecast uncertainty surrounding the landfall of hurricanes. The issuance of a reasonable worst case scenario for the strongest winds that could impact a region could amplify messaging to encourage people to take appropriate action prior to a landfall.
Abstract
Strong downslope windstorms can cause extensive property damage and extreme wildfire spread, so their accurate prediction is important. Although some early studies suggested high predictability for downslope windstorms, more recent analyses have found limited predictability for such winds. Nevertheless, there is a theoretical basis for expecting higher downslope wind predictability in cases with a mean-state critical level, and this is supported by one previous effort to forecast actual events. To more thoroughly investigate downslope windstorm predictability, we compare archived simulations from the NCAR ensemble, a 10-member mesoscale ensemble run at 3-km horizontal grid spacing over the entire contiguous United States, to observed events at 15 stations in the western United States susceptible to strong downslope winds. We assess predictability in three contexts: the average ensemble spread, which provides an estimate of potential predictability; a forecast evaluation based upon binary-decision criteria, which is representative of operational hazard warnings; and a probabilistic forecast evaluation using the continuous ranked probability score (CRPS), which is a measure of an ensemble’s ability to generate the proper probability distribution for the events under consideration. We do find better predictive skill for the mean-state critical-level regime in comparison to other downslope windstorm–generating mechanisms. Our downslope windstorm warning performance, calculated using binary-decision criteria from the bias-corrected ensemble forecasts, performed slightly worse for no-critical-level events, and slightly better for critical-level events, than National Weather Service high-wind warnings aggregated over all types of high-wind events throughout the United States and annually averaged for each year between 2008 and 2019.
Abstract
Strong downslope windstorms can cause extensive property damage and extreme wildfire spread, so their accurate prediction is important. Although some early studies suggested high predictability for downslope windstorms, more recent analyses have found limited predictability for such winds. Nevertheless, there is a theoretical basis for expecting higher downslope wind predictability in cases with a mean-state critical level, and this is supported by one previous effort to forecast actual events. To more thoroughly investigate downslope windstorm predictability, we compare archived simulations from the NCAR ensemble, a 10-member mesoscale ensemble run at 3-km horizontal grid spacing over the entire contiguous United States, to observed events at 15 stations in the western United States susceptible to strong downslope winds. We assess predictability in three contexts: the average ensemble spread, which provides an estimate of potential predictability; a forecast evaluation based upon binary-decision criteria, which is representative of operational hazard warnings; and a probabilistic forecast evaluation using the continuous ranked probability score (CRPS), which is a measure of an ensemble’s ability to generate the proper probability distribution for the events under consideration. We do find better predictive skill for the mean-state critical-level regime in comparison to other downslope windstorm–generating mechanisms. Our downslope windstorm warning performance, calculated using binary-decision criteria from the bias-corrected ensemble forecasts, performed slightly worse for no-critical-level events, and slightly better for critical-level events, than National Weather Service high-wind warnings aggregated over all types of high-wind events throughout the United States and annually averaged for each year between 2008 and 2019.
Abstract
This work set out to assess the performance of four forecast systems [the Short-Range Ensemble Forecast (SREF), High-Resolution Rapid Refresh Ensemble (HRRRE), the National Blend of Models (NBM), and the Probabilistic Snow Accumulation product (PSA) from the National Weather Service (NWS) Boulder, Colorado, Weather Forecast Office] when predicting snowfall events around the Intermountain West to advise winter weather decision-making processes at Denver International Airport. The goal was to provide airport personnel and the Boulder NWS Forecast Office with operationally relevant verification results on the timing and severity of these events so they are able to make better-informed decisions to minimize negative impacts of storms. Forecasts of snow events using various probability thresholds and a climatological snow-to-liquid ratio of 15:1 were evaluated against Meteorological Aerodrome Reports (METARs) for 24-h periods following four decision-making times spaced equally throughout the day. For the ensembles, a frequentist approach was used: the forecast probability equaled the percentage of ensemble members that predicted a snow event. The results show that the NBM had the best timing of snow events out of the products, while all the products tended to overforecast snow amount. Additionally, NBM had fewer snow events and rarely had high probabilities of snow, unlike the other forecast products.
Abstract
This work set out to assess the performance of four forecast systems [the Short-Range Ensemble Forecast (SREF), High-Resolution Rapid Refresh Ensemble (HRRRE), the National Blend of Models (NBM), and the Probabilistic Snow Accumulation product (PSA) from the National Weather Service (NWS) Boulder, Colorado, Weather Forecast Office] when predicting snowfall events around the Intermountain West to advise winter weather decision-making processes at Denver International Airport. The goal was to provide airport personnel and the Boulder NWS Forecast Office with operationally relevant verification results on the timing and severity of these events so they are able to make better-informed decisions to minimize negative impacts of storms. Forecasts of snow events using various probability thresholds and a climatological snow-to-liquid ratio of 15:1 were evaluated against Meteorological Aerodrome Reports (METARs) for 24-h periods following four decision-making times spaced equally throughout the day. For the ensembles, a frequentist approach was used: the forecast probability equaled the percentage of ensemble members that predicted a snow event. The results show that the NBM had the best timing of snow events out of the products, while all the products tended to overforecast snow amount. Additionally, NBM had fewer snow events and rarely had high probabilities of snow, unlike the other forecast products.
Abstract
Analog ensembles (AnEns) traditionally use a single numerical weather prediction (NWP) model to make a forecast, then search an archive to find a number of past similar forecasts (analogs) from that same model, and finally retrieve the actual observations corresponding to those past forecasts to serve as members of an ensemble forecast. This study investigates new statistical methods to combine analogs into ensemble forecasts and validates them for 3-hourly precipitation over the complex terrain of British Columbia, Canada. Applying the past analog error to the target forecast (instead of using the observations directly) reduces the AnEn dry bias and makes prediction of heavy-precipitation events probabilistically more reliable—typically the most impactful forecasts for society. Two variants of this new technique enable AnEn members to obtain values outside the distribution of the finite archived observational dataset—that is, they are theoretically capable of forecasting record events, whereas traditional analog methods cannot. While both variants similarly improve heavier precipitation events, one variant predicts measurable precipitation more often, which enhances accuracy during winter. A multimodel AnEn further improves predictive skill, albeit at higher computational cost. AnEn performance shows larger sensitivity to the grid spacing of the NWP than to the physics configuration. The final AnEn prediction system improves the skill and reliability of point forecasts across all precipitation intensities.
Significance Statement
The analog ensemble (AnEn) technique is a data-driven method that can improve local weather forecasts. It improves raw model forecasts using past similar model predictions and observations, reducing future forecast errors and providing probabilities for a range of possible outcomes. One limitation of AnEns is that they commonly tend to make rare-event (e.g., heavy precipitation) forecasts appear less extreme. Usually, heavier precipitation events have a higher impact on society and the economy. This study introduces two new AnEn techniques that make operational forecasts of both probabilities and most likely amounts more accurate for heavy precipitation.
Abstract
Analog ensembles (AnEns) traditionally use a single numerical weather prediction (NWP) model to make a forecast, then search an archive to find a number of past similar forecasts (analogs) from that same model, and finally retrieve the actual observations corresponding to those past forecasts to serve as members of an ensemble forecast. This study investigates new statistical methods to combine analogs into ensemble forecasts and validates them for 3-hourly precipitation over the complex terrain of British Columbia, Canada. Applying the past analog error to the target forecast (instead of using the observations directly) reduces the AnEn dry bias and makes prediction of heavy-precipitation events probabilistically more reliable—typically the most impactful forecasts for society. Two variants of this new technique enable AnEn members to obtain values outside the distribution of the finite archived observational dataset—that is, they are theoretically capable of forecasting record events, whereas traditional analog methods cannot. While both variants similarly improve heavier precipitation events, one variant predicts measurable precipitation more often, which enhances accuracy during winter. A multimodel AnEn further improves predictive skill, albeit at higher computational cost. AnEn performance shows larger sensitivity to the grid spacing of the NWP than to the physics configuration. The final AnEn prediction system improves the skill and reliability of point forecasts across all precipitation intensities.
Significance Statement
The analog ensemble (AnEn) technique is a data-driven method that can improve local weather forecasts. It improves raw model forecasts using past similar model predictions and observations, reducing future forecast errors and providing probabilities for a range of possible outcomes. One limitation of AnEns is that they commonly tend to make rare-event (e.g., heavy precipitation) forecasts appear less extreme. Usually, heavier precipitation events have a higher impact on society and the economy. This study introduces two new AnEn techniques that make operational forecasts of both probabilities and most likely amounts more accurate for heavy precipitation.
Abstract
Tropical Cyclone (TC) Sally formed on 11 September 2020, traveled through the Gulf of Mexico (GMX), and intensified rapidly before making landfall on the Alabama coast as a devastating category-2 TC with extensive coastal and inland flooding. In this study, using a combination of observations and idealized numerical model experiments, we demonstrate that the Mississippi River plume played a key role in the intensification of Sally near the northern Gulf Coast. As the storm intensified and its translation slowed before landfall, sea surface cooling was reduced along its track, coincident with a pronounced increase in SSS. Further analysis reveals that TC Sally encountered a warm Loop Current eddy in the northern GMX close to the Mississippi River plume. Besides deepening the thermocline, the eddy advected low-salinity Mississippi River plume water into the storm’s path. This resulted in the development of strong upper-ocean salinity stratification, with a shallow layer of freshwater lying above a deep, warm “barrier layer.” Consequently, TC-induced mixing and the associated sea surface cooling were reduced, aiding Sally’s intensification. These results suggest that the Mississippi River plume and freshwater advection by the Loop Current eddies can play an important role in TC intensification near the U.S. Gulf Coast.
Abstract
Tropical Cyclone (TC) Sally formed on 11 September 2020, traveled through the Gulf of Mexico (GMX), and intensified rapidly before making landfall on the Alabama coast as a devastating category-2 TC with extensive coastal and inland flooding. In this study, using a combination of observations and idealized numerical model experiments, we demonstrate that the Mississippi River plume played a key role in the intensification of Sally near the northern Gulf Coast. As the storm intensified and its translation slowed before landfall, sea surface cooling was reduced along its track, coincident with a pronounced increase in SSS. Further analysis reveals that TC Sally encountered a warm Loop Current eddy in the northern GMX close to the Mississippi River plume. Besides deepening the thermocline, the eddy advected low-salinity Mississippi River plume water into the storm’s path. This resulted in the development of strong upper-ocean salinity stratification, with a shallow layer of freshwater lying above a deep, warm “barrier layer.” Consequently, TC-induced mixing and the associated sea surface cooling were reduced, aiding Sally’s intensification. These results suggest that the Mississippi River plume and freshwater advection by the Loop Current eddies can play an important role in TC intensification near the U.S. Gulf Coast.
