Browse
Abstract
Reliably quantifying uncertainty in precipitation forecasts remains a critical challenge. This work examines the application of a deep learning (DL) architecture, Unet, for postprocessing deterministic numerical weather predictions of precipitation to improve their skills and for deriving forecast uncertainty. Daily accumulated 0–4-day precipitation forecasts are generated from a 34-yr reforecast based on the West Weather Research and Forecasting (West-WRF) mesoscale model, developed by the Center for Western Weather and Water Extremes. The Unet learns the distributional parameters associated with a censored, shifted gamma distribution. In addition, the DL framework is tested against state-of-the-art benchmark methods, including an analog ensemble, nonhomogeneous regression, and mixed-type meta-Gaussian distribution. These methods are evaluated over four years of data and the western United States. The Unet outperforms the benchmark methods at all lead times as measured by continuous ranked probability and Brier skill scores. The Unet also produces a reliable estimation of forecast uncertainty, as measured by binned spread–skill relationship diagrams. Additionally, the Unet has the best performance for extreme events (i.e., the 95th and 99th percentiles of the distribution) and for these cases, its performance improves as more training data are available.
Significance Statement
Accurate precipitation forecasts are critical for social and economic sectors. They also play an important role in our daily activity planning. The objective of this research is to investigate how to use a deep learning architecture to postprocess high-resolution (4 km) precipitation forecasts and generate accurate and reliable forecasts with quantified uncertainty. The proposed approach performs well with extreme cases and its performance improves as more data are available in training.
Abstract
Reliably quantifying uncertainty in precipitation forecasts remains a critical challenge. This work examines the application of a deep learning (DL) architecture, Unet, for postprocessing deterministic numerical weather predictions of precipitation to improve their skills and for deriving forecast uncertainty. Daily accumulated 0–4-day precipitation forecasts are generated from a 34-yr reforecast based on the West Weather Research and Forecasting (West-WRF) mesoscale model, developed by the Center for Western Weather and Water Extremes. The Unet learns the distributional parameters associated with a censored, shifted gamma distribution. In addition, the DL framework is tested against state-of-the-art benchmark methods, including an analog ensemble, nonhomogeneous regression, and mixed-type meta-Gaussian distribution. These methods are evaluated over four years of data and the western United States. The Unet outperforms the benchmark methods at all lead times as measured by continuous ranked probability and Brier skill scores. The Unet also produces a reliable estimation of forecast uncertainty, as measured by binned spread–skill relationship diagrams. Additionally, the Unet has the best performance for extreme events (i.e., the 95th and 99th percentiles of the distribution) and for these cases, its performance improves as more training data are available.
Significance Statement
Accurate precipitation forecasts are critical for social and economic sectors. They also play an important role in our daily activity planning. The objective of this research is to investigate how to use a deep learning architecture to postprocess high-resolution (4 km) precipitation forecasts and generate accurate and reliable forecasts with quantified uncertainty. The proposed approach performs well with extreme cases and its performance improves as more data are available in training.
Abstract
A multiscale alignment (MSA) ensemble filtering method was introduced by Ying to reduce nonlinear position errors effectively during data assimilation. The MSA method extends the traditional ensemble Kalman filter (EnKF) to update states from large to small scales sequentially, during which it leverages the displacement vectors derived from the large-scale analysis increments to reduce position errors at smaller scales through warping of the model grid. This study stress tests the MSA method in various scenarios using an idealized vortex model. We show that the MSA improves filter performance as number of scales (Ns ) increases in the presence of nonlinear position errors. We tuned localization parameters for the cross-scale EnKF updates to find the best performance when assimilating an observation network. To further reduce the scale mismatch between observations and states, a new option called MSA-O is introduced to decompose observations into scale components during assimilation. Cycling DA experiments show that the MSA-O consistently outperforms the traditional EnKF at equal computational cost. A more challenging scenario for the MSA is identified when the large-scale background flow and the small-scale vortex are incoherent in terms of their errors, making the displacement vectors not effective in reducing vortex position errors. Observation availability for the small scales also limits the use of large Ns for the MSA. Potential remedies for these issues are discussed.
Abstract
A multiscale alignment (MSA) ensemble filtering method was introduced by Ying to reduce nonlinear position errors effectively during data assimilation. The MSA method extends the traditional ensemble Kalman filter (EnKF) to update states from large to small scales sequentially, during which it leverages the displacement vectors derived from the large-scale analysis increments to reduce position errors at smaller scales through warping of the model grid. This study stress tests the MSA method in various scenarios using an idealized vortex model. We show that the MSA improves filter performance as number of scales (Ns ) increases in the presence of nonlinear position errors. We tuned localization parameters for the cross-scale EnKF updates to find the best performance when assimilating an observation network. To further reduce the scale mismatch between observations and states, a new option called MSA-O is introduced to decompose observations into scale components during assimilation. Cycling DA experiments show that the MSA-O consistently outperforms the traditional EnKF at equal computational cost. A more challenging scenario for the MSA is identified when the large-scale background flow and the small-scale vortex are incoherent in terms of their errors, making the displacement vectors not effective in reducing vortex position errors. Observation availability for the small scales also limits the use of large Ns for the MSA. Potential remedies for these issues are discussed.
Abstract
Numerical weather prediction models contain parameters that are inherently uncertain and cannot be determined exactly. It is thus desirable to have reliable objective approaches for estimation of optimal values and uncertainties of these parameters. Traditionally, the parameter tuning has been done manually, which can lead to the tuning process being a maze of subjective choices. In this paper we present how to optimize 20 key physical parameters in the atmospheric model Open Integrated Forecasting System (OpenIFS) that have a strong impact on forecast quality. The results show that simultaneous optimization of O(20) parameters is possible with O(100) algorithm steps using an ensemble of O(20) members; the results also show that the optimized parameters lead to substantial enhancement of predictive skill. The enhanced predictive skill can be attributed to reduced biases in low-level winds and upper-tropospheric humidity in the optimized model. We find that the optimization process is dependent on the starting values of the parameters that are optimized (starting from better-suited values results in a better model). The results show also that the applicability of the tuned parameter values across different model resolutions is somewhat limited because of resolution-dependent model biases, and we also found that the parameter covariances provided by the tuning algorithm seem to be uninformative.
Significance Statement
The purpose of this work is to show how to use algorithmic methods to optimize a weather model in a computationally efficient manner. Traditional manual model tuning is an extremely laborious and time-consuming process, so algorithmic methods have strong potential for saving the model developers’ time and accelerating development. This paper shows that algorithmic optimization is possible and that weather forecasts can be improved. However, potential issues related to the use of the optimized parameter values across different model resolutions are discussed as well as other shortcomings related to the tuning process.
Abstract
Numerical weather prediction models contain parameters that are inherently uncertain and cannot be determined exactly. It is thus desirable to have reliable objective approaches for estimation of optimal values and uncertainties of these parameters. Traditionally, the parameter tuning has been done manually, which can lead to the tuning process being a maze of subjective choices. In this paper we present how to optimize 20 key physical parameters in the atmospheric model Open Integrated Forecasting System (OpenIFS) that have a strong impact on forecast quality. The results show that simultaneous optimization of O(20) parameters is possible with O(100) algorithm steps using an ensemble of O(20) members; the results also show that the optimized parameters lead to substantial enhancement of predictive skill. The enhanced predictive skill can be attributed to reduced biases in low-level winds and upper-tropospheric humidity in the optimized model. We find that the optimization process is dependent on the starting values of the parameters that are optimized (starting from better-suited values results in a better model). The results show also that the applicability of the tuned parameter values across different model resolutions is somewhat limited because of resolution-dependent model biases, and we also found that the parameter covariances provided by the tuning algorithm seem to be uninformative.
Significance Statement
The purpose of this work is to show how to use algorithmic methods to optimize a weather model in a computationally efficient manner. Traditional manual model tuning is an extremely laborious and time-consuming process, so algorithmic methods have strong potential for saving the model developers’ time and accelerating development. This paper shows that algorithmic optimization is possible and that weather forecasts can be improved. However, potential issues related to the use of the optimized parameter values across different model resolutions are discussed as well as other shortcomings related to the tuning process.
Abstract
A lack of routine environmental observations located near deepening cumulus congestus clouds limits verification of important theorized and simulated updraft–environment interaction processes occurring during deep convection initiation (CI). We analyze radiosonde profiles collected during several hundred CI events near a mountain range in central Argentina during the CACTI field campaign. Statistical analyses illustrate environmental conditions supporting radar-observed CI outcomes that span a spectrum of convective cell depths, widths, and durations, as well as events lacking precipitating convection. Tested environmental factors include a large variety of sounding-derived measurements of CAPE, CIN, moisture, terrain-relative winds, vertical shear, and lifted parcel properties, with supplemental model reanalysis of background larger-scale vertical motion. CAPE and CIN metrics do not consistently differentiate CI success from failure. Only a few environmental factors contain consistent monotonic relationships among the spectrum of cloud depths achieved during CI: (i) the depth and strength of background ascent, and (ii) the component of low-level flow oriented parallel to the ridgeline. These metrics suggest that the ability of the surrounding flow to lift parcels to their LFC and terrain-modified flow are consistently relevant processes for CI. Low- to midlevel relative humidity strongly discriminated between CI and non-CI events, likely reflecting entrainment-driven dilution processes. However, we could not confidently conclude that relative humidity similarly discriminated robust from marginal CI events. Circumstantial evidence was found linking cell width, an important cloud property governing the probability of CI, to LCL height, boundary layer depth, depth and magnitude of the CIN layer, and ambient wind shear.
Abstract
A lack of routine environmental observations located near deepening cumulus congestus clouds limits verification of important theorized and simulated updraft–environment interaction processes occurring during deep convection initiation (CI). We analyze radiosonde profiles collected during several hundred CI events near a mountain range in central Argentina during the CACTI field campaign. Statistical analyses illustrate environmental conditions supporting radar-observed CI outcomes that span a spectrum of convective cell depths, widths, and durations, as well as events lacking precipitating convection. Tested environmental factors include a large variety of sounding-derived measurements of CAPE, CIN, moisture, terrain-relative winds, vertical shear, and lifted parcel properties, with supplemental model reanalysis of background larger-scale vertical motion. CAPE and CIN metrics do not consistently differentiate CI success from failure. Only a few environmental factors contain consistent monotonic relationships among the spectrum of cloud depths achieved during CI: (i) the depth and strength of background ascent, and (ii) the component of low-level flow oriented parallel to the ridgeline. These metrics suggest that the ability of the surrounding flow to lift parcels to their LFC and terrain-modified flow are consistently relevant processes for CI. Low- to midlevel relative humidity strongly discriminated between CI and non-CI events, likely reflecting entrainment-driven dilution processes. However, we could not confidently conclude that relative humidity similarly discriminated robust from marginal CI events. Circumstantial evidence was found linking cell width, an important cloud property governing the probability of CI, to LCL height, boundary layer depth, depth and magnitude of the CIN layer, and ambient wind shear.
Abstract
This study investigates the structure and evolution of a summertime convective event that occurred on 14 July 2015 over the Arabian region. We use the WRF Model with 1-km horizontal grid spacing and test three PBL parameterizations: the Mellor–Yamada–Nakanishi–Niino (MYNN) scheme; the Asymmetrical Convective Model, version 2, (ACM2) scheme; and the quasi-normal scale-elimination (QNSE) scheme. Convection initiates near the Al Hajar Mountains of northern Oman at around 1100 local time (LT; 0700 UTC) and propagates northwestward. A nonorographic convective band along the west coast of the United Arab Emirates (UAE) develops after 1500 LT as a result of the convergence of cold pools with the sea breeze from the Arabian Gulf. The model simulation employing the QNSE scheme simulates the convection initiation and propagation well. Although the MYNN and ACM2 simulations show convective initiation near the Al Hajar Mountains, they fail to simulate the development of the convective band along the UAE west coast. The MYNN run simulates colder near-surface temperatures and a weaker sea breeze, whereas the ACM2 run simulates a stronger sea breeze but a drier lower troposphere. Sensitivity simulations using horizontal grid spacings of 9 and 3 km show that lower-resolution runs develop broader convective structures and weaker cold pools and horizontal wind divergence, affecting the development of convection along the west coast of the UAE. The 1-km run using the QNSE PBL scheme realistically captures the sequence of events that leads to the moist convection over the UAE and adjacent mountains.
Abstract
This study investigates the structure and evolution of a summertime convective event that occurred on 14 July 2015 over the Arabian region. We use the WRF Model with 1-km horizontal grid spacing and test three PBL parameterizations: the Mellor–Yamada–Nakanishi–Niino (MYNN) scheme; the Asymmetrical Convective Model, version 2, (ACM2) scheme; and the quasi-normal scale-elimination (QNSE) scheme. Convection initiates near the Al Hajar Mountains of northern Oman at around 1100 local time (LT; 0700 UTC) and propagates northwestward. A nonorographic convective band along the west coast of the United Arab Emirates (UAE) develops after 1500 LT as a result of the convergence of cold pools with the sea breeze from the Arabian Gulf. The model simulation employing the QNSE scheme simulates the convection initiation and propagation well. Although the MYNN and ACM2 simulations show convective initiation near the Al Hajar Mountains, they fail to simulate the development of the convective band along the UAE west coast. The MYNN run simulates colder near-surface temperatures and a weaker sea breeze, whereas the ACM2 run simulates a stronger sea breeze but a drier lower troposphere. Sensitivity simulations using horizontal grid spacings of 9 and 3 km show that lower-resolution runs develop broader convective structures and weaker cold pools and horizontal wind divergence, affecting the development of convection along the west coast of the UAE. The 1-km run using the QNSE PBL scheme realistically captures the sequence of events that leads to the moist convection over the UAE and adjacent mountains.
Abstract
Severe rainfall has become increasingly frequent and intense in the Taipei metropolitan area. A complex thunderstorm in the Taipei Basin on 14 June 2015 produced an extreme rain rate (>130 mm h−1), leading to an urban flash flood. This paper presents storms’ microphysical and dynamic features during the organizing and heavy rain stages, mainly based on observed polarimetric variables in a Doppler radar network and ground-based raindrop size distribution. Shallower isolated cells in the early afternoon characterized by big raindrops produced a rain rate > 10 mm h−1, but the rain showers persisted for a short time. The storm’s evolution highlighted the behavior of merged convective cells before the heaviest rainfall (exceeding 60 mm within 20 min). The columnar features of differential reflectivity (Z DR) and specific differential phase (K DP) became more evident in merged cells, which correlated with the broad distribution of upward motion and mixed-phase hydrometeors. The K DP below the environmental 0°C level increased toward the ground associated with the melted graupel and resulted in subsequent intense rain rates, showing the contribution of the ice-phase process. Due to the collision–breakup process, the highest concentrations of almost all drop sizes and smaller mass-weighted mean diameter occurred during the maximum rainfall stage.
Abstract
Severe rainfall has become increasingly frequent and intense in the Taipei metropolitan area. A complex thunderstorm in the Taipei Basin on 14 June 2015 produced an extreme rain rate (>130 mm h−1), leading to an urban flash flood. This paper presents storms’ microphysical and dynamic features during the organizing and heavy rain stages, mainly based on observed polarimetric variables in a Doppler radar network and ground-based raindrop size distribution. Shallower isolated cells in the early afternoon characterized by big raindrops produced a rain rate > 10 mm h−1, but the rain showers persisted for a short time. The storm’s evolution highlighted the behavior of merged convective cells before the heaviest rainfall (exceeding 60 mm within 20 min). The columnar features of differential reflectivity (Z DR) and specific differential phase (K DP) became more evident in merged cells, which correlated with the broad distribution of upward motion and mixed-phase hydrometeors. The K DP below the environmental 0°C level increased toward the ground associated with the melted graupel and resulted in subsequent intense rain rates, showing the contribution of the ice-phase process. Due to the collision–breakup process, the highest concentrations of almost all drop sizes and smaller mass-weighted mean diameter occurred during the maximum rainfall stage.
Abstract
Based on the principle “learn from past errors to correct current forecasts,” statistical postprocessing consists of optimizing forecasts generated by numerical weather prediction (NWP) models. In this context, machine learning (ML) offers state-of-the-art tools for training statistical models and making predictions based on large datasets. In our study, ML-based solutions are developed to reduce forecast errors of 2-m temperature and 10-m wind speed of the ECMWF’s operational medium-range, high-resolution forecasts produced with the Integrated Forecasting System (IFS). IFS forecasts and other spatiotemporal indicators are used as predictors after careful selection with the help of ML interpretability tools. Different ML approaches are tested: linear regression, random forest decision trees, and neural networks. Statistical models of systematic and random errors are derived sequentially where the random error is defined as the residual error after bias correction. In terms of output, bias correction and forecast uncertainty prediction are made available at any point from locations around the world. All three ML methods show a similar ability to capture situation-dependent biases leading to noteworthy performance improvements (between 10% and 15% improvement in terms of root-mean-square error for all lead times and variables), and a similar ability to provide reliable uncertainty predictions.
Abstract
Based on the principle “learn from past errors to correct current forecasts,” statistical postprocessing consists of optimizing forecasts generated by numerical weather prediction (NWP) models. In this context, machine learning (ML) offers state-of-the-art tools for training statistical models and making predictions based on large datasets. In our study, ML-based solutions are developed to reduce forecast errors of 2-m temperature and 10-m wind speed of the ECMWF’s operational medium-range, high-resolution forecasts produced with the Integrated Forecasting System (IFS). IFS forecasts and other spatiotemporal indicators are used as predictors after careful selection with the help of ML interpretability tools. Different ML approaches are tested: linear regression, random forest decision trees, and neural networks. Statistical models of systematic and random errors are derived sequentially where the random error is defined as the residual error after bias correction. In terms of output, bias correction and forecast uncertainty prediction are made available at any point from locations around the world. All three ML methods show a similar ability to capture situation-dependent biases leading to noteworthy performance improvements (between 10% and 15% improvement in terms of root-mean-square error for all lead times and variables), and a similar ability to provide reliable uncertainty predictions.
Abstract
As a part of the Tropical Cyclone Rapid Intensification Project (TCRI), observations were made of the rapid intensification of Hurricane Sally (2020) as it passed over the Gulf of Mexico. High-altitude dropsondes and radar observations from NOAA’s Gulfstream IV, radar observations from WP-3D aircraft, the WSR-88D ground radar network, satellite images, and satellite-detected lightning strikes are used to apply recently developed theoretical knowledge about tropical cyclone intensification. As observed in many other tropical cyclones, strong, bottom-heavy vertical mass flux profiles are correlated with low (but positive) values of low- to midlevel moist convective instability along with high column relative humidity. Such mass flux profiles produce rapid spinup at low levels and the environmental conditions giving rise to them are associated with an intense midlevel vortex. This low-level spinup underneath the midlevel vortex results in the vertical alignment of the vortex column, which is a key step in the rapid intensification process. In the case of Sally, the spinup of the low-level vortex resulted from vorticity stretching, while the spinup of the midlevel vortex at 6 km resulted from vorticity tilting produced by the interaction of convective ascent with moderate vertical shear.
Significance Statement
The purpose of this study is to investigate the rapid intensification of Hurricane Sally as it was approaching the Florida Panhandle. We do that by analyzing an unprecedented dataset from the NOAA WP-3D and Gulfstream-IV aircraft, together with ground-based radar and satellite data. We find that both the dynamics (vorticity structure and evolution) and thermodynamics (instability index, saturation fraction, heating/mass flux profiles) need to be considered in diagnosing intensification processes. Further field projects with continuous high-altitude dropsondes and research are needed to see if these are applicable to other reformation events as well as genesis.
Abstract
As a part of the Tropical Cyclone Rapid Intensification Project (TCRI), observations were made of the rapid intensification of Hurricane Sally (2020) as it passed over the Gulf of Mexico. High-altitude dropsondes and radar observations from NOAA’s Gulfstream IV, radar observations from WP-3D aircraft, the WSR-88D ground radar network, satellite images, and satellite-detected lightning strikes are used to apply recently developed theoretical knowledge about tropical cyclone intensification. As observed in many other tropical cyclones, strong, bottom-heavy vertical mass flux profiles are correlated with low (but positive) values of low- to midlevel moist convective instability along with high column relative humidity. Such mass flux profiles produce rapid spinup at low levels and the environmental conditions giving rise to them are associated with an intense midlevel vortex. This low-level spinup underneath the midlevel vortex results in the vertical alignment of the vortex column, which is a key step in the rapid intensification process. In the case of Sally, the spinup of the low-level vortex resulted from vorticity stretching, while the spinup of the midlevel vortex at 6 km resulted from vorticity tilting produced by the interaction of convective ascent with moderate vertical shear.
Significance Statement
The purpose of this study is to investigate the rapid intensification of Hurricane Sally as it was approaching the Florida Panhandle. We do that by analyzing an unprecedented dataset from the NOAA WP-3D and Gulfstream-IV aircraft, together with ground-based radar and satellite data. We find that both the dynamics (vorticity structure and evolution) and thermodynamics (instability index, saturation fraction, heating/mass flux profiles) need to be considered in diagnosing intensification processes. Further field projects with continuous high-altitude dropsondes and research are needed to see if these are applicable to other reformation events as well as genesis.
Abstract
The local spectral width (LSW) of a radio occultation (RO) observation in impact parameter representation is a useful parameter for providing information on the uncertainty associated with the RO bending angle measurement. The LSW can potentially be used to specify the bending angle observation error (BaOE) in the lower troposphere for each individual sounding. This study assesses the usefulness and limitations of LSW in representing BaOE for a global data assimilation system. A two-step scheme is proposed to derive profile-dependent BaOE from LSW. Since the LSW-based BaOE varies with each individual RO observation, it is here designated as a dynamic BaOE (DBOE) in contrast to the traditional statistics-based BaOE specification. A benchmark control run and two sensitivity experiments are conducted with continuous cycling data assimilation using the National Centers for Environmental Prediction (NCEP) Gridpoint Statistical Interpolation (GSI) and Global Forecast System (GFS). The usefulness and impact of the LSW-based DBOE are evaluated using radiosonde observations and global analyses. Results show that DBOE is able to improve the assimilation of RO data, leading to better forecast skill scores. Another experiment, in which the GSI statistical observation error of the benchmark run is replaced by the average of LSW-based DBOE, shows that the ability to assign larger weighting for high-quality observation and lower weighting for low-quality observation is the key factor for the success of the LSW-based DBOE.
Abstract
The local spectral width (LSW) of a radio occultation (RO) observation in impact parameter representation is a useful parameter for providing information on the uncertainty associated with the RO bending angle measurement. The LSW can potentially be used to specify the bending angle observation error (BaOE) in the lower troposphere for each individual sounding. This study assesses the usefulness and limitations of LSW in representing BaOE for a global data assimilation system. A two-step scheme is proposed to derive profile-dependent BaOE from LSW. Since the LSW-based BaOE varies with each individual RO observation, it is here designated as a dynamic BaOE (DBOE) in contrast to the traditional statistics-based BaOE specification. A benchmark control run and two sensitivity experiments are conducted with continuous cycling data assimilation using the National Centers for Environmental Prediction (NCEP) Gridpoint Statistical Interpolation (GSI) and Global Forecast System (GFS). The usefulness and impact of the LSW-based DBOE are evaluated using radiosonde observations and global analyses. Results show that DBOE is able to improve the assimilation of RO data, leading to better forecast skill scores. Another experiment, in which the GSI statistical observation error of the benchmark run is replaced by the average of LSW-based DBOE, shows that the ability to assign larger weighting for high-quality observation and lower weighting for low-quality observation is the key factor for the success of the LSW-based DBOE.
Abstract
Estimation of the radius of maximum wind (RMW) of tropical cyclone (TC) is helpful for the disaster prevention and mitigation. If RMWs are estimated from infrared (IR) imagery taken by geostationary meteorological satellites, their estimation is available densely in time, regardless of the ocean basin. Kossin et al. showed that when TCs have clear eyes, the eye radii estimated from IR images have a high correlation with the RMW estimated from aircraft reconnaissance. The regression of the former onto that latter was shown to have a mean absolute error (MAE) of 4.7 km. We revisit the IR-based RMW estimation by using C-band synthetic aperture radar (SAR) sea surface wind estimates. The criteria for selecting clear-eye cases are simplified. The MAE of the Kossin et al. method is found to be smaller than previously suggested: 3.1 km when the proposed relation is used and 2.7 km when the regression is revised with the SAR-measured RMWs. We further propose an improvement of the IR-based method to estimate the eye radii. The resultant MAE is shown to be 1.7 km, which indicates that the IR-based RMW estimation is more accurate than has been suggested. A strong correlation between eyewall slope and eye size is confirmed. We also investigated cloud features in the eye that may be closely related to RMW and wind structure around RMW. Potential applications of highly accurate RMW estimation are discussed.
Significance Statement
The radius of maximum wind (RMW) of tropical cyclone (TC) is an important factor for TC intensity estimation and disaster prevention. A previous study suggested that the RMWs of TCs with clear eyes can be estimated from geostationary satellite images at a mean absolute error (MAE) of 4.7 km. Here we improved the method, reducing the MAE by more than one-half. Since the method does not require aircraft or satellite in low Earth orbit, it helps TC monitoring at high frequency. The method can also improve initialization of models used to predict TC hazards and further our physical understanding and the climatology of the wind structures near the centers of TCs.
Abstract
Estimation of the radius of maximum wind (RMW) of tropical cyclone (TC) is helpful for the disaster prevention and mitigation. If RMWs are estimated from infrared (IR) imagery taken by geostationary meteorological satellites, their estimation is available densely in time, regardless of the ocean basin. Kossin et al. showed that when TCs have clear eyes, the eye radii estimated from IR images have a high correlation with the RMW estimated from aircraft reconnaissance. The regression of the former onto that latter was shown to have a mean absolute error (MAE) of 4.7 km. We revisit the IR-based RMW estimation by using C-band synthetic aperture radar (SAR) sea surface wind estimates. The criteria for selecting clear-eye cases are simplified. The MAE of the Kossin et al. method is found to be smaller than previously suggested: 3.1 km when the proposed relation is used and 2.7 km when the regression is revised with the SAR-measured RMWs. We further propose an improvement of the IR-based method to estimate the eye radii. The resultant MAE is shown to be 1.7 km, which indicates that the IR-based RMW estimation is more accurate than has been suggested. A strong correlation between eyewall slope and eye size is confirmed. We also investigated cloud features in the eye that may be closely related to RMW and wind structure around RMW. Potential applications of highly accurate RMW estimation are discussed.
Significance Statement
The radius of maximum wind (RMW) of tropical cyclone (TC) is an important factor for TC intensity estimation and disaster prevention. A previous study suggested that the RMWs of TCs with clear eyes can be estimated from geostationary satellite images at a mean absolute error (MAE) of 4.7 km. Here we improved the method, reducing the MAE by more than one-half. Since the method does not require aircraft or satellite in low Earth orbit, it helps TC monitoring at high frequency. The method can also improve initialization of models used to predict TC hazards and further our physical understanding and the climatology of the wind structures near the centers of TCs.
