Browse
Abstract
The sea surface temperature anomaly (SSTA) plays a key role in climate change and extreme weather processes. Usually, SSTA forecast methods consist of numerical and conventional statistical models, the former can be seriously influenced by the uncertainty of physical parameterization schemes, the nonlinearity of ocean dynamic processes, and the nonrobustness of numerical discretization algorithms. Recently, deep learning has been explored to address forecast issues in the field of oceanography. However, existing deep learning models for ocean forecasting are mainly site-specific, which were designed for forecasting on a single point or for an independent variable. Moreover, few special deep learning networks have been developed to deal with SSTA field forecasts under typhoon conditions. In this study, a multivariable convolutional neural network (MCNN) is proposed, which can be applied for synoptic-scale SSTA forecasting in the South China Sea. In addition to the SSTA itself, the surface wind speed and the surface current velocity are regarded as input variables for the prediction networks, effectively reflecting the influences of both local atmospheric dynamic forcing and nonlocal oceanic thermal advection. Experimental results demonstrate that MCNN exhibits better performance than single-variable convolutional neural network (SCNN), especially for the SSTA forecast during the typhoon passage. While forecast results deteriorate rapidly in the SCNN during the passage of a typhoon, forecast errors in the MCNN can be effectively restrained to slowly increase over the forecast time due to the introduction of the surface wind speed in this network.
Abstract
The sea surface temperature anomaly (SSTA) plays a key role in climate change and extreme weather processes. Usually, SSTA forecast methods consist of numerical and conventional statistical models, the former can be seriously influenced by the uncertainty of physical parameterization schemes, the nonlinearity of ocean dynamic processes, and the nonrobustness of numerical discretization algorithms. Recently, deep learning has been explored to address forecast issues in the field of oceanography. However, existing deep learning models for ocean forecasting are mainly site-specific, which were designed for forecasting on a single point or for an independent variable. Moreover, few special deep learning networks have been developed to deal with SSTA field forecasts under typhoon conditions. In this study, a multivariable convolutional neural network (MCNN) is proposed, which can be applied for synoptic-scale SSTA forecasting in the South China Sea. In addition to the SSTA itself, the surface wind speed and the surface current velocity are regarded as input variables for the prediction networks, effectively reflecting the influences of both local atmospheric dynamic forcing and nonlocal oceanic thermal advection. Experimental results demonstrate that MCNN exhibits better performance than single-variable convolutional neural network (SCNN), especially for the SSTA forecast during the typhoon passage. While forecast results deteriorate rapidly in the SCNN during the passage of a typhoon, forecast errors in the MCNN can be effectively restrained to slowly increase over the forecast time due to the introduction of the surface wind speed in this network.
Abstract
A 16-member convective-scale ensemble prediction system (CEPS) developed at the Central Weather Bureau (CWB) of Taiwan is evaluated for probability forecasts of convective precipitation. To address the issues of limited predictability of convective systems, the CEPS provides short-range forecasts using initial conditions from a rapid-updated ensemble data assimilation system. This study aims to identify the behavior of the CEPS forecasts, especially the impact of different ensemble configurations and forecast lead times. Warm-season afternoon thunderstorms (ATs) from 30 June to 4 July 2017 are selected. Since ATs usually occur between 1300 and 2000 LST, this study compares deterministic and probabilistic quantitative precipitation forecasts (QPFs) launched at 0500, 0800, and 1100 LST. This study demonstrates that initial and boundary perturbations (IBP) are crucial to ensure good spread–skill consistency over the 18-h forecasts. On top of IBP, additional model perturbations have insignificant impacts on upper-air and precipitation forecasts. The deterministic QPFs launched at 1100 LST outperform those launched at 0500 and 0800 LST, likely because the most-recent data assimilation analyses enhance the practical predictability. However, it cannot improve the probabilistic QPFs launched at 1100 LST due to inadequate ensemble spreads resulting from limited error growth time. This study points out the importance of sufficient initial condition uncertainty on short-range probabilistic forecasts to exploit the benefits of rapid-update data assimilation analyses.
Significance Statement
This study aims to understand the behavior of convective-scale short-range probabilistic forecasts in Taiwan and the surrounding area. Taiwan is influenced by diverse weather systems, including typhoons, mei-yu fronts, and local thunderstorms. During the past decade, there has been promising improvement in predicting mesoscale weather systems (e.g., typhoons and mei-yu fronts). However, it is still challenging to provide timely and accurate forecasts for rapid-evolving high-impact convection. This study provides a reference for the designation of convective-scale ensemble prediction systems; in particular, those with a goal to provide short-range probabilistic forecasts. While the findings cannot be extrapolated to all ensemble prediction systems, this study demonstrates that initial and boundary perturbations are the most important factors, while the model perturbation has an insignificant effect. This study suggests that in-depth studies are required to improve the convective-scale initial condition accuracy and uncertainty to provide reliable probabilistic forecasts within short lead times.
Abstract
A 16-member convective-scale ensemble prediction system (CEPS) developed at the Central Weather Bureau (CWB) of Taiwan is evaluated for probability forecasts of convective precipitation. To address the issues of limited predictability of convective systems, the CEPS provides short-range forecasts using initial conditions from a rapid-updated ensemble data assimilation system. This study aims to identify the behavior of the CEPS forecasts, especially the impact of different ensemble configurations and forecast lead times. Warm-season afternoon thunderstorms (ATs) from 30 June to 4 July 2017 are selected. Since ATs usually occur between 1300 and 2000 LST, this study compares deterministic and probabilistic quantitative precipitation forecasts (QPFs) launched at 0500, 0800, and 1100 LST. This study demonstrates that initial and boundary perturbations (IBP) are crucial to ensure good spread–skill consistency over the 18-h forecasts. On top of IBP, additional model perturbations have insignificant impacts on upper-air and precipitation forecasts. The deterministic QPFs launched at 1100 LST outperform those launched at 0500 and 0800 LST, likely because the most-recent data assimilation analyses enhance the practical predictability. However, it cannot improve the probabilistic QPFs launched at 1100 LST due to inadequate ensemble spreads resulting from limited error growth time. This study points out the importance of sufficient initial condition uncertainty on short-range probabilistic forecasts to exploit the benefits of rapid-update data assimilation analyses.
Significance Statement
This study aims to understand the behavior of convective-scale short-range probabilistic forecasts in Taiwan and the surrounding area. Taiwan is influenced by diverse weather systems, including typhoons, mei-yu fronts, and local thunderstorms. During the past decade, there has been promising improvement in predicting mesoscale weather systems (e.g., typhoons and mei-yu fronts). However, it is still challenging to provide timely and accurate forecasts for rapid-evolving high-impact convection. This study provides a reference for the designation of convective-scale ensemble prediction systems; in particular, those with a goal to provide short-range probabilistic forecasts. While the findings cannot be extrapolated to all ensemble prediction systems, this study demonstrates that initial and boundary perturbations are the most important factors, while the model perturbation has an insignificant effect. This study suggests that in-depth studies are required to improve the convective-scale initial condition accuracy and uncertainty to provide reliable probabilistic forecasts within short lead times.
Abstract
This article introduces an ensemble clustering tool developed at the Weather Prediction Center (WPC) to assist forecasters in the preparation of medium-range (3–7 day) forecasts. Effectively incorporating ensemble data into an operational forecasting process, like that used at WPC, can be challenging given time constraints and data infrastructure limitations. Often forecasters do not have time to view the large number of constituent members of an ensemble forecast, so they settle for viewing the ensemble’s mean and spread. This ignores the useful information about forecast uncertainty and the range of possible forecast outcomes that an ensemble forecast can provide. Ensemble clustering could be a solution to this problem as it can reduce a large ensemble forecast down to the most prevalent forecast scenarios. Forecasters can then quickly view these ensemble clusters to better understand and communicate forecast uncertainty and the range of possible forecast outcomes. The ensemble clustering tool developed at WPC is a variation of fuzzy clustering where operationally available ensemble members with similar 500-hPa geopotential height forecasts are grouped into four clusters. A representative case from 15 February 2021 is presented to demonstrate the clustering methodology and the overall utility of this new ensemble clustering tool. Cumulative verification statistics show that one of the four forecast scenarios identified by this ensemble clustering tool routinely outperforms all the available ensemble mean and deterministic forecasts.
Significance Statement
Ensemble forecasts could be used more effectively in medium-range (3–7 day) forecasting. Currently, the onus is put on forecasters to view and synthesize all of the data contained in an ensemble forecast. This is a task they often do not have time to adequately execute. This work proposes a solution to this problem. An automated tool was developed that would split the available ensemble members into four groups of broadly similar members. These groups were presented to forecasters as four potential forecast outcomes. Forecasters felt this tool helped them to better incorporate ensemble forecasts into their forecast process. Verification shows that presenting ensemble forecasts in this manner is an improvement on currently used ensemble forecast visualization techniques.
Abstract
This article introduces an ensemble clustering tool developed at the Weather Prediction Center (WPC) to assist forecasters in the preparation of medium-range (3–7 day) forecasts. Effectively incorporating ensemble data into an operational forecasting process, like that used at WPC, can be challenging given time constraints and data infrastructure limitations. Often forecasters do not have time to view the large number of constituent members of an ensemble forecast, so they settle for viewing the ensemble’s mean and spread. This ignores the useful information about forecast uncertainty and the range of possible forecast outcomes that an ensemble forecast can provide. Ensemble clustering could be a solution to this problem as it can reduce a large ensemble forecast down to the most prevalent forecast scenarios. Forecasters can then quickly view these ensemble clusters to better understand and communicate forecast uncertainty and the range of possible forecast outcomes. The ensemble clustering tool developed at WPC is a variation of fuzzy clustering where operationally available ensemble members with similar 500-hPa geopotential height forecasts are grouped into four clusters. A representative case from 15 February 2021 is presented to demonstrate the clustering methodology and the overall utility of this new ensemble clustering tool. Cumulative verification statistics show that one of the four forecast scenarios identified by this ensemble clustering tool routinely outperforms all the available ensemble mean and deterministic forecasts.
Significance Statement
Ensemble forecasts could be used more effectively in medium-range (3–7 day) forecasting. Currently, the onus is put on forecasters to view and synthesize all of the data contained in an ensemble forecast. This is a task they often do not have time to adequately execute. This work proposes a solution to this problem. An automated tool was developed that would split the available ensemble members into four groups of broadly similar members. These groups were presented to forecasters as four potential forecast outcomes. Forecasters felt this tool helped them to better incorporate ensemble forecasts into their forecast process. Verification shows that presenting ensemble forecasts in this manner is an improvement on currently used ensemble forecast visualization techniques.
Abstract
A Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC) follow-on constellation, COSMIC-2, was successfully launched into equatorial orbit on June 24, 2019. With an increased signal-to-noise ratio due to improved receivers and digital beam-steering antennas, COSMIC-2 is producing about 5,000 high-quality radio-occultation (RO) profiles daily over the tropics and subtropics. The initial evaluation of the impact of assimilating COSMIC-2 into NOAA’s Global Forecast System (GFS) showed mixed results, and adjustments to quality control procedures and observation error characteristics had to be made prior to the assimilation of this dataset in the operational configuration in May 2020. Additional changes in the GFS that followed this initial operational implementation resulted in a larger percentage of rejection (~ 90 %) of all RO observations, including COSMIC-2, in the mid-lower troposphere. Since then, two software upgrades directly related to the assimilation of RO bending angle observations were developed. These improvements aimed at optimizing the utilization of COSMIC-2 and other RO observations to improve global weather analyses and forecasts. The first upgrade was implemented operationally in September 2021 and the second one in November 2022. This study describes both RO software upgrades and evaluates the impact of COSMIC-2 with this most recently improved configuration. Specifically, we show that the assimilation of COSMIC-2 observations has a significant impact in improving temperature and winds in the tropics, though benefits also extend to the extra-tropical latitudes.
Abstract
A Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC) follow-on constellation, COSMIC-2, was successfully launched into equatorial orbit on June 24, 2019. With an increased signal-to-noise ratio due to improved receivers and digital beam-steering antennas, COSMIC-2 is producing about 5,000 high-quality radio-occultation (RO) profiles daily over the tropics and subtropics. The initial evaluation of the impact of assimilating COSMIC-2 into NOAA’s Global Forecast System (GFS) showed mixed results, and adjustments to quality control procedures and observation error characteristics had to be made prior to the assimilation of this dataset in the operational configuration in May 2020. Additional changes in the GFS that followed this initial operational implementation resulted in a larger percentage of rejection (~ 90 %) of all RO observations, including COSMIC-2, in the mid-lower troposphere. Since then, two software upgrades directly related to the assimilation of RO bending angle observations were developed. These improvements aimed at optimizing the utilization of COSMIC-2 and other RO observations to improve global weather analyses and forecasts. The first upgrade was implemented operationally in September 2021 and the second one in November 2022. This study describes both RO software upgrades and evaluates the impact of COSMIC-2 with this most recently improved configuration. Specifically, we show that the assimilation of COSMIC-2 observations has a significant impact in improving temperature and winds in the tropics, though benefits also extend to the extra-tropical latitudes.
Abstract
This study marks the most comprehensive assessment of the overall impact of dropsondes on tropical cyclone (TC) forecasts to date. We compare two experiments to quantify dropsonde impact: one that assimilated and another that denied dropsonde observations. These experiments used a basin-scale, multi-storm configuration of the Hurricane Weather Research and Forecasting model (HWRF) and covered active North Atlantic basin periods during the 2017–2020 hurricane seasons. The importance of a sufficiently large sample size as well as thoroughly understanding the error distribution by stratifying results are highlighted by this work. Overall, dropsondes directly improved forecasts during sampled periods and indirectly impacted forecasts during unsampled periods. Benefits for forecasts of track, intensity, and outer wind radii were more pronounced during sampled periods. The forecast improvements of outer wind radii were most notable given the impact that TC size has on TC-hazards forecasts. Additionally, robustly observing the inner and near-core regionwas necessary for 64-kt-wind-radii forecasts. Yet, these benefitswere heavily dependent on the data assimilation (DA) system quality. More specifically, dropsondes only improved forecasts when the analysis used mesoscale error covariance derived from a cycled HWRF ensemble, suggesting that it is a vital DA component. Further, while forecast improvements were found regardless of initial classification and in steady-state TCs, TCs undergoing intensity change had diminished benefits. The diminished benefits during intensity change probably reflects continued DA deficiencies. Thus, improving DA-system quality and observing system limitations would likely enhance dropsonde impacts.
Abstract
This study marks the most comprehensive assessment of the overall impact of dropsondes on tropical cyclone (TC) forecasts to date. We compare two experiments to quantify dropsonde impact: one that assimilated and another that denied dropsonde observations. These experiments used a basin-scale, multi-storm configuration of the Hurricane Weather Research and Forecasting model (HWRF) and covered active North Atlantic basin periods during the 2017–2020 hurricane seasons. The importance of a sufficiently large sample size as well as thoroughly understanding the error distribution by stratifying results are highlighted by this work. Overall, dropsondes directly improved forecasts during sampled periods and indirectly impacted forecasts during unsampled periods. Benefits for forecasts of track, intensity, and outer wind radii were more pronounced during sampled periods. The forecast improvements of outer wind radii were most notable given the impact that TC size has on TC-hazards forecasts. Additionally, robustly observing the inner and near-core regionwas necessary for 64-kt-wind-radii forecasts. Yet, these benefitswere heavily dependent on the data assimilation (DA) system quality. More specifically, dropsondes only improved forecasts when the analysis used mesoscale error covariance derived from a cycled HWRF ensemble, suggesting that it is a vital DA component. Further, while forecast improvements were found regardless of initial classification and in steady-state TCs, TCs undergoing intensity change had diminished benefits. The diminished benefits during intensity change probably reflects continued DA deficiencies. Thus, improving DA-system quality and observing system limitations would likely enhance dropsonde impacts.
Abstract
The National Weather Service plays a critical role in alerting the public when dangerous weather occurs. Tornado warnings are one of the most publicly visible products the NWS issues given the large societal impacts tornadoes can have . Understanding the performance of these warnings is crucial for providing adequate warning during tornadic events and improving overall warning performance. This study aims to understand warning performance during the lifetimes of individual storms (specifically in terms of probability of detection and lead time). For example, does probability of detection vary based on if the tornado was the first produced by the storm, or the last? We use tornado outbreak data from 2008–2014, archived NEXRAD radar data, and the NWS verification database to associate each tornado report with a storm object. This approach allows for an analysis of warning performance based on chronological order of tornado occurrence within each storm. Results show that the probability of detection and lead time increase with later tornadoes in the storm; the first tornadoes of each storm are less likely to be warned and on average have less lead time. Probability of detection also decreases overnight, especially for first tornadoes and storms that only produce one tornado. These results are important for understanding how tornado warning performance varies during individual storm lifecycles and how upstream forecast products (e.g., Storm Prediction Center tornado watches, mesoscale discussions, etc.) may increase warning confidence for the first tornado produced by each storm.
Abstract
The National Weather Service plays a critical role in alerting the public when dangerous weather occurs. Tornado warnings are one of the most publicly visible products the NWS issues given the large societal impacts tornadoes can have . Understanding the performance of these warnings is crucial for providing adequate warning during tornadic events and improving overall warning performance. This study aims to understand warning performance during the lifetimes of individual storms (specifically in terms of probability of detection and lead time). For example, does probability of detection vary based on if the tornado was the first produced by the storm, or the last? We use tornado outbreak data from 2008–2014, archived NEXRAD radar data, and the NWS verification database to associate each tornado report with a storm object. This approach allows for an analysis of warning performance based on chronological order of tornado occurrence within each storm. Results show that the probability of detection and lead time increase with later tornadoes in the storm; the first tornadoes of each storm are less likely to be warned and on average have less lead time. Probability of detection also decreases overnight, especially for first tornadoes and storms that only produce one tornado. These results are important for understanding how tornado warning performance varies during individual storm lifecycles and how upstream forecast products (e.g., Storm Prediction Center tornado watches, mesoscale discussions, etc.) may increase warning confidence for the first tornado produced by each storm.
Abstract
This study presents an initial demonstration of assimilating small Uncrewed Aircraft System (sUAS) data into an operational model with a goal to ultimately improve tropical cyclone (TC) analyses and forecasts. The observations, obtained using the Coyote sUAS in Hurricane Maria (2017), were assimilated into the operational Hurricane Weather Research and Forecast system (HWRF) as they could be in operations. Results suggest that the Coyote data can benefit HWRF forecasts. A single-cycle case study produced the best results when the Coyote observations were assimilated at greater horizontal resolution with more relaxed quality control (QC) than comparable flight-level high-density observations currently used in operations. The case study results guided experiments that cycled HWRF for a roughly four-day period that covered all Coyote flights into Maria. The cycled experiment that assimilated the most data improved initial inner-core structure in the analyses and better agreed with other aircraft observations. The average errors in track and intensity decreased in the subsequent forecasts. Intensity forecasts were too weak when no Coyote data was assimilated, and assimilating the Coyote data made the forecasts stronger. Results also suggest that a symmetric distribution of Coyote data around the TC center is necessary to maximize its benefits in the current configuration of operational HWRF. Although the sample size was limited, these experiments provide insight for potential operational use of data from newer sUAS platforms in future TC applications.
Abstract
This study presents an initial demonstration of assimilating small Uncrewed Aircraft System (sUAS) data into an operational model with a goal to ultimately improve tropical cyclone (TC) analyses and forecasts. The observations, obtained using the Coyote sUAS in Hurricane Maria (2017), were assimilated into the operational Hurricane Weather Research and Forecast system (HWRF) as they could be in operations. Results suggest that the Coyote data can benefit HWRF forecasts. A single-cycle case study produced the best results when the Coyote observations were assimilated at greater horizontal resolution with more relaxed quality control (QC) than comparable flight-level high-density observations currently used in operations. The case study results guided experiments that cycled HWRF for a roughly four-day period that covered all Coyote flights into Maria. The cycled experiment that assimilated the most data improved initial inner-core structure in the analyses and better agreed with other aircraft observations. The average errors in track and intensity decreased in the subsequent forecasts. Intensity forecasts were too weak when no Coyote data was assimilated, and assimilating the Coyote data made the forecasts stronger. Results also suggest that a symmetric distribution of Coyote data around the TC center is necessary to maximize its benefits in the current configuration of operational HWRF. Although the sample size was limited, these experiments provide insight for potential operational use of data from newer sUAS platforms in future TC applications.
Abstract
To mitigate the impacts associated with adverse weather conditions, meteorological services issue weather warnings to the general public. These warnings rely heavily on forecasts issued by underlying prediction systems. When deciding which prediction system(s) to utilize when constructing warnings, it is important to compare systems in their ability to forecast the occurrence and severity of high-impact weather events. However, evaluating forecasts for particular outcomes is known to be a challenging task. This is exacerbated further by the fact that high-impact weather often manifests as a result of several confounding features, a realization that has led to considerable research on so-called compound weather events. Both univariate and multivariate methods are therefore required to evaluate forecasts for high-impact weather. In this paper, we discuss weighted verification tools, which allow particular outcomes to be emphasized during forecast evaluation. We review and compare different approaches to construct weighted scoring rules, both in a univariate and multivariate setting, and we leverage existing results on weighted scores to introduce conditional probability integral transform (PIT) histograms, allowing forecast calibration to be assessed conditionally on particular outcomes having occurred. To illustrate the practical benefit afforded by these weighted verification tools, they are employed in a case study to evaluate probabilistic forecasts for extreme heat events issued by the Swiss Federal Office of Meteorology and Climatology (MeteoSwiss).
Abstract
To mitigate the impacts associated with adverse weather conditions, meteorological services issue weather warnings to the general public. These warnings rely heavily on forecasts issued by underlying prediction systems. When deciding which prediction system(s) to utilize when constructing warnings, it is important to compare systems in their ability to forecast the occurrence and severity of high-impact weather events. However, evaluating forecasts for particular outcomes is known to be a challenging task. This is exacerbated further by the fact that high-impact weather often manifests as a result of several confounding features, a realization that has led to considerable research on so-called compound weather events. Both univariate and multivariate methods are therefore required to evaluate forecasts for high-impact weather. In this paper, we discuss weighted verification tools, which allow particular outcomes to be emphasized during forecast evaluation. We review and compare different approaches to construct weighted scoring rules, both in a univariate and multivariate setting, and we leverage existing results on weighted scores to introduce conditional probability integral transform (PIT) histograms, allowing forecast calibration to be assessed conditionally on particular outcomes having occurred. To illustrate the practical benefit afforded by these weighted verification tools, they are employed in a case study to evaluate probabilistic forecasts for extreme heat events issued by the Swiss Federal Office of Meteorology and Climatology (MeteoSwiss).
Abstract
This study analyzes the potential of deep learning using probabilistic artificial neural networks (ANNs) for postprocessing ensemble precipitation forecasts at four observation locations. We split the precipitation forecast problem into two tasks: estimating the probability of precipitation and predicting the hourly precipitation. We then compare the performance with classical statistical postprocessing (logistical regression and GLM). ANNs show a higher performance at three of the four stations for estimating the probability of precipitation and at all stations for predicting the hourly precipitation. Further, two more general ANN models are trained using the merged data from all four stations. These general ANNs exhibit an increase in performance compared to the station-specific ANNs at most stations. However, they show a significant decay in performance at one of the stations at estimating the hourly precipitation. The general models seem capable of learning meaningful interactions in the data and generalizing these to improve the performance at other sites, which also causes the loss of local information at one station. Thus, this study indicates the potential of deep learning in weather forecasting workflows.
Abstract
This study analyzes the potential of deep learning using probabilistic artificial neural networks (ANNs) for postprocessing ensemble precipitation forecasts at four observation locations. We split the precipitation forecast problem into two tasks: estimating the probability of precipitation and predicting the hourly precipitation. We then compare the performance with classical statistical postprocessing (logistical regression and GLM). ANNs show a higher performance at three of the four stations for estimating the probability of precipitation and at all stations for predicting the hourly precipitation. Further, two more general ANN models are trained using the merged data from all four stations. These general ANNs exhibit an increase in performance compared to the station-specific ANNs at most stations. However, they show a significant decay in performance at one of the stations at estimating the hourly precipitation. The general models seem capable of learning meaningful interactions in the data and generalizing these to improve the performance at other sites, which also causes the loss of local information at one station. Thus, this study indicates the potential of deep learning in weather forecasting workflows.
