Browse
Abstract
Ensemble sensitivity analysis (ESA) is a numerical method by which the potential value of a single additional observation can be estimated using an ensemble numerical weather forecast. By performing ESA observation targeting on runs of the TTU WRF Ensemble from the Spring of 2016, a dataset of predicted variance reductions (hereafter referred to as target values) was obtained over approximately 6 weeks’ worth of convective forecasts for the central US. It was then ascertained from these cases that the geographic variation in target values is large for any one case, with local maxima often several standard deviations higher than the mean and surrounded by sharp gradients. Radiosondes launched from the surface, then, would need to take this variation into account to properly sample a specific target by launching upstream of where the target is located rather than directly underneath. In many cases, the difference between the maximum target value in the vertical and the actual target value observed along the balloon path was multiple standard deviations. This may help explain the lower-than-expected forecast improvements observed in previous ESA targeting experiments, especially the Mesoscale Predictability Experiment (MPEX). If target values are a good predictor of observation value, then it is possible that taking the balloon path into account in targeting systems for radiosonde deployment may substantially improve on the value added to the forecast by individual observations.
Abstract
Ensemble sensitivity analysis (ESA) is a numerical method by which the potential value of a single additional observation can be estimated using an ensemble numerical weather forecast. By performing ESA observation targeting on runs of the TTU WRF Ensemble from the Spring of 2016, a dataset of predicted variance reductions (hereafter referred to as target values) was obtained over approximately 6 weeks’ worth of convective forecasts for the central US. It was then ascertained from these cases that the geographic variation in target values is large for any one case, with local maxima often several standard deviations higher than the mean and surrounded by sharp gradients. Radiosondes launched from the surface, then, would need to take this variation into account to properly sample a specific target by launching upstream of where the target is located rather than directly underneath. In many cases, the difference between the maximum target value in the vertical and the actual target value observed along the balloon path was multiple standard deviations. This may help explain the lower-than-expected forecast improvements observed in previous ESA targeting experiments, especially the Mesoscale Predictability Experiment (MPEX). If target values are a good predictor of observation value, then it is possible that taking the balloon path into account in targeting systems for radiosonde deployment may substantially improve on the value added to the forecast by individual observations.
Abstract
In this study, we analyze various sources of CAPE in the environment and their contributions to its time tendency that will complement forecast models and operational analyses that are relatively temporally (∼1 h) coarse. As a case study, the relative roles of direct insolation and near-surface moisture advection in the recovery CAPE on 31 March 2016 in northern Alabama are examined using VORTEX-Southeast (VORTEX-SE) observations and numerical simulations. In between rounds of nontornadic morning storms and tornadic evening storms, CAPE over the VORTEX-SE domain increased from near zero to at least 500 J kg−1. A timeline of the day’s events is provided with a focus on the evolution of the lower levels of the atmosphere. We focus on its responses to solar insolation and moisture advection, which we hypothesize as the main mechanisms behind the recovery of CAPE. Data from the University of Massachusetts S-Band frequency-modulated, continuous-wave (FMCW) radar and NOAA National Severe Storms Laboratory (NSSL) Collaborative Lower Atmospheric Mobile Profiling System (CLAMPS), and high-resolution EnKF analyses from the Advanced Regional Prediction System (ARPS) are used to characterize the boundary layer evolution in the pre-tornadic storm environment. It is found that insolation-driven surface diabatic heating was the primary driver of rapid CAPE recovery on this day. The methodology developed in this case can be applied in other scenarios to diagnose the primary drivers of CAPE development.
Significance Statement
The mechanisms by which atmospheric instability recovers can vary widely and are often a source of uncertainty in forecasting. We want to understand how and why the environment destabilized enough to produce an evening tornado following morning storms on 31 March 2016. To do this, we used model data and observations from a collocated radar and profiler. It was found that heating from the sun at the surface was the primary cause of destabilization in the environment.
Abstract
In this study, we analyze various sources of CAPE in the environment and their contributions to its time tendency that will complement forecast models and operational analyses that are relatively temporally (∼1 h) coarse. As a case study, the relative roles of direct insolation and near-surface moisture advection in the recovery CAPE on 31 March 2016 in northern Alabama are examined using VORTEX-Southeast (VORTEX-SE) observations and numerical simulations. In between rounds of nontornadic morning storms and tornadic evening storms, CAPE over the VORTEX-SE domain increased from near zero to at least 500 J kg−1. A timeline of the day’s events is provided with a focus on the evolution of the lower levels of the atmosphere. We focus on its responses to solar insolation and moisture advection, which we hypothesize as the main mechanisms behind the recovery of CAPE. Data from the University of Massachusetts S-Band frequency-modulated, continuous-wave (FMCW) radar and NOAA National Severe Storms Laboratory (NSSL) Collaborative Lower Atmospheric Mobile Profiling System (CLAMPS), and high-resolution EnKF analyses from the Advanced Regional Prediction System (ARPS) are used to characterize the boundary layer evolution in the pre-tornadic storm environment. It is found that insolation-driven surface diabatic heating was the primary driver of rapid CAPE recovery on this day. The methodology developed in this case can be applied in other scenarios to diagnose the primary drivers of CAPE development.
Significance Statement
The mechanisms by which atmospheric instability recovers can vary widely and are often a source of uncertainty in forecasting. We want to understand how and why the environment destabilized enough to produce an evening tornado following morning storms on 31 March 2016. To do this, we used model data and observations from a collocated radar and profiler. It was found that heating from the sun at the surface was the primary cause of destabilization in the environment.
Abstract
Atmospheric reanalyses are widely used to estimate the past atmospheric near-surface state over sea ice. They provide boundary conditions for sea ice and ocean numerical simulations and relevant information for studying polar variability and anthropogenic climate change. Previous research revealed the existence of large near-surface temperature biases (mostly warm) over the Arctic sea ice in the current generation of atmospheric reanalyses, which is linked to a poor representation of the snow over the sea ice and the stably stratified boundary layer in the forecast models used to produce the reanalyses. These errors can compromise the employment of reanalysis products in support of polar research. Here, we train a fully connected neural network that learns from remote sensing infrared temperature observations to correct the existing generation of uncoupled atmospheric reanalyses (ERA5, JRA-55) based on a set of sea ice and atmospheric predictors, which are themselves reanalysis products. The advantages of the proposed correction scheme over previous calibration attempts are the consideration of the synoptic weather and cloud state, compatibility of the predictors with the mechanism responsible for the bias, and a self-emerging seasonality and multidecadal trend consistent with the declining sea ice state in the Arctic. The correction leads on average to a 27% temperature bias reduction for ERA5 and 7% for JRA-55 if compared to independent in situ observations from the MOSAiC campaign (respectively, 32% and 10% under clear-sky conditions). These improvements can be beneficial for forced sea ice and ocean simulations, which rely on reanalyses surface fields as boundary conditions.
Significance Statement
This study illustrates a novel method based on machine learning for reducing the systematic surface temperature errors that characterize multiple atmospheric reanalyses in sea ice–covered regions of the Arctic under clear-sky conditions. The correction applied to the temperature field is consistent with the local weather and the sea ice and snow conditions, meaning that it responds to seasonal changes in sea ice cover as well as to its long-term decline due to global warming. The corrected reanalysis temperature can be employed to support polar research activities, and in particular to better simulate the evolution of the interacting sea ice and ocean system within numerical models.
Abstract
Atmospheric reanalyses are widely used to estimate the past atmospheric near-surface state over sea ice. They provide boundary conditions for sea ice and ocean numerical simulations and relevant information for studying polar variability and anthropogenic climate change. Previous research revealed the existence of large near-surface temperature biases (mostly warm) over the Arctic sea ice in the current generation of atmospheric reanalyses, which is linked to a poor representation of the snow over the sea ice and the stably stratified boundary layer in the forecast models used to produce the reanalyses. These errors can compromise the employment of reanalysis products in support of polar research. Here, we train a fully connected neural network that learns from remote sensing infrared temperature observations to correct the existing generation of uncoupled atmospheric reanalyses (ERA5, JRA-55) based on a set of sea ice and atmospheric predictors, which are themselves reanalysis products. The advantages of the proposed correction scheme over previous calibration attempts are the consideration of the synoptic weather and cloud state, compatibility of the predictors with the mechanism responsible for the bias, and a self-emerging seasonality and multidecadal trend consistent with the declining sea ice state in the Arctic. The correction leads on average to a 27% temperature bias reduction for ERA5 and 7% for JRA-55 if compared to independent in situ observations from the MOSAiC campaign (respectively, 32% and 10% under clear-sky conditions). These improvements can be beneficial for forced sea ice and ocean simulations, which rely on reanalyses surface fields as boundary conditions.
Significance Statement
This study illustrates a novel method based on machine learning for reducing the systematic surface temperature errors that characterize multiple atmospheric reanalyses in sea ice–covered regions of the Arctic under clear-sky conditions. The correction applied to the temperature field is consistent with the local weather and the sea ice and snow conditions, meaning that it responds to seasonal changes in sea ice cover as well as to its long-term decline due to global warming. The corrected reanalysis temperature can be employed to support polar research activities, and in particular to better simulate the evolution of the interacting sea ice and ocean system within numerical models.
Abstract
Midlatitude cyclones approaching coastal mountain ranges experience flow modifications on a variety of scales including orographic lift, blocking, mountain waves, and valley flows. During the 2015/16 Olympic Mountain Experiment (OLYMPEX), a pair of scanning ground radars observed precipitating clouds as they were modified by these orographically induced flows. The DOW radar, positioned to scan up the windward Quinault Valley, conducted RHI scans during 285 h of precipitation, 80% of which contained reversed, down-valley flow at lower levels. The existence of down-valley flow in the Quinault Valley was found to be well correlated with upstream flow blocking and the large-scale sea level pressure gradient orientated down the valley. Deep down-valley flow occurred in environments with high moist static stability and southerly winds, conditions that are common in prefrontal sectors of midlatitude cyclones in the coastal Pacific Northwest. Finally, a case study of prolonged down-valley flow in a prefrontal storm sector was simulated to investigate whether latent heat absorption (cooling) contributed to the event. Three experiments were conducted: a Control simulation and two simulations where the temperature tendencies from melting and evaporation were separately turned off. Results indicated that evaporative cooling had a stronger impact on the event’s down-valley flow than melting, likely because evaporation occurred within the low-level down-valley flow layer. Through these experiments, we show that evaporation helped prolong down-valley flow for several hours past the time of the event’s warm frontal passage.
Significance Statement
This paper analyzes the characteristics of down-valley flow over the windward Quinault Valley on the Olympic Peninsula of Washington State using data from OLYMPEX, with an emphasis on regional pressure differences and blocking metrics. Results demonstrate that the location of precipitation over the Olympic Peninsula is shifted upstream during events with deep down-valley flow, consistent with blocked upstream airflow. A case study of down-valley flow highlights the role of evaporative cooling to prolong the flow reversal.
Abstract
Midlatitude cyclones approaching coastal mountain ranges experience flow modifications on a variety of scales including orographic lift, blocking, mountain waves, and valley flows. During the 2015/16 Olympic Mountain Experiment (OLYMPEX), a pair of scanning ground radars observed precipitating clouds as they were modified by these orographically induced flows. The DOW radar, positioned to scan up the windward Quinault Valley, conducted RHI scans during 285 h of precipitation, 80% of which contained reversed, down-valley flow at lower levels. The existence of down-valley flow in the Quinault Valley was found to be well correlated with upstream flow blocking and the large-scale sea level pressure gradient orientated down the valley. Deep down-valley flow occurred in environments with high moist static stability and southerly winds, conditions that are common in prefrontal sectors of midlatitude cyclones in the coastal Pacific Northwest. Finally, a case study of prolonged down-valley flow in a prefrontal storm sector was simulated to investigate whether latent heat absorption (cooling) contributed to the event. Three experiments were conducted: a Control simulation and two simulations where the temperature tendencies from melting and evaporation were separately turned off. Results indicated that evaporative cooling had a stronger impact on the event’s down-valley flow than melting, likely because evaporation occurred within the low-level down-valley flow layer. Through these experiments, we show that evaporation helped prolong down-valley flow for several hours past the time of the event’s warm frontal passage.
Significance Statement
This paper analyzes the characteristics of down-valley flow over the windward Quinault Valley on the Olympic Peninsula of Washington State using data from OLYMPEX, with an emphasis on regional pressure differences and blocking metrics. Results demonstrate that the location of precipitation over the Olympic Peninsula is shifted upstream during events with deep down-valley flow, consistent with blocked upstream airflow. A case study of down-valley flow highlights the role of evaporative cooling to prolong the flow reversal.
Abstract
It has been long recognized that in the retrieved thermodynamic fields using multiple-Doppler radar synthesized winds, an unknown constant exists on each horizontal level, leading to an ambiguity in the retrieved vertical structure. In this study, the traditional thermodynamic retrieval scheme is significantly improved by the implementation of the Equation of State (EoS) as an additional constraint. With this new formulation, the ambiguity of the vertical structure can be explicitly identified and removed from the retrieved three-dimensional thermodynamic fields. The only in situ independent observations needed to perform the correction are the pressure and temperature measurements taken at a single surface station. If data from multiple surface stations are available, a strategy is proposed to obtain a better estimate of the unknown constant. Experiments in this research were conducted under the observation system simulation experiment (OSSE) framework to demonstrate the validity of the new approach. Problems and possible solutions associated with using real datasets and potential future extended applications of this new method are discussed.
Abstract
It has been long recognized that in the retrieved thermodynamic fields using multiple-Doppler radar synthesized winds, an unknown constant exists on each horizontal level, leading to an ambiguity in the retrieved vertical structure. In this study, the traditional thermodynamic retrieval scheme is significantly improved by the implementation of the Equation of State (EoS) as an additional constraint. With this new formulation, the ambiguity of the vertical structure can be explicitly identified and removed from the retrieved three-dimensional thermodynamic fields. The only in situ independent observations needed to perform the correction are the pressure and temperature measurements taken at a single surface station. If data from multiple surface stations are available, a strategy is proposed to obtain a better estimate of the unknown constant. Experiments in this research were conducted under the observation system simulation experiment (OSSE) framework to demonstrate the validity of the new approach. Problems and possible solutions associated with using real datasets and potential future extended applications of this new method are discussed.
Abstract
Based on the conditional nonlinear optimal perturbation (CNOP) approach, the predictability of mei-yu heavy precipitation and its underlying physical processes is investigated. As an extension of our previous work, the practical predictability of heavy precipitation events is studied using more realistic initial perturbations than previously considered. The initial perturbation reflects certain physical connections among multiple variables including zonal and meridional winds, potential temperature (T), and water vapor mixing ratio (Q). Two types of initial perturbations for the CNOP are identified, with similar spatial distributions but opposite signs and resulting effects. The accumulated precipitation is strengthened with mostly positive perturbations in the T and Q components for the CNOP, and weakened by negative perturbations. Comparing downscaling (DOWN) perturbations and random perturbations (RPs) with the CNOP, it is found that the CNOP and DOWN perturbations exhibit particularly large- and mesoscale spatial structures, respectively, while the RPs yield a spatial distribution with mostly convective-scale features. Also, the CNOP results in the largest error growth and forecast uncertainty, especially for Q, followed by the DOWN perturbations, and then those in the RPs are the smallest. These results provide important implications for optimizing the initial perturbations of convection-permitting ensemble prediction systems, especially precipitation forecasts. Moreover, it is suggested that small-scale related variables, i.e., those associated with vertical motion and microphysical processes, are much less predictable than thermodynamic variables, and the errors grow through distinct physical processes for the three types of initial perturbations, i.e., with flow-dependent features.
Abstract
Based on the conditional nonlinear optimal perturbation (CNOP) approach, the predictability of mei-yu heavy precipitation and its underlying physical processes is investigated. As an extension of our previous work, the practical predictability of heavy precipitation events is studied using more realistic initial perturbations than previously considered. The initial perturbation reflects certain physical connections among multiple variables including zonal and meridional winds, potential temperature (T), and water vapor mixing ratio (Q). Two types of initial perturbations for the CNOP are identified, with similar spatial distributions but opposite signs and resulting effects. The accumulated precipitation is strengthened with mostly positive perturbations in the T and Q components for the CNOP, and weakened by negative perturbations. Comparing downscaling (DOWN) perturbations and random perturbations (RPs) with the CNOP, it is found that the CNOP and DOWN perturbations exhibit particularly large- and mesoscale spatial structures, respectively, while the RPs yield a spatial distribution with mostly convective-scale features. Also, the CNOP results in the largest error growth and forecast uncertainty, especially for Q, followed by the DOWN perturbations, and then those in the RPs are the smallest. These results provide important implications for optimizing the initial perturbations of convection-permitting ensemble prediction systems, especially precipitation forecasts. Moreover, it is suggested that small-scale related variables, i.e., those associated with vertical motion and microphysical processes, are much less predictable than thermodynamic variables, and the errors grow through distinct physical processes for the three types of initial perturbations, i.e., with flow-dependent features.
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
Estimating and predicting the state of the atmosphere is a probabilistic problem, and often employs an ensemble modeling approach to represent uncertainty in the system. Common methods for examining uncertainty and assessing performance for ensembles emphasize pointwise statistics or marginal distributions. However, these methods lose specific information about individual ensemble members. This paper explores contour band depth (cBD), a method of analyzing uncertainty in terms of contours of scalar fields. cBD is fully nonparametric and induces an ordering on ensemble members that leads to box-and-whisker-plot-type visualizations of uncertainty for two-dimensional data. By applying cBD to synthetic ensembles, we demonstrate that it provides enhanced information about the spatial structure of ensemble uncertainty. We also find that the usefulness of the cBD analysis depends on the presence of multiple modes and multiple scales in the ensemble of contours. Finally, we apply cBD to compare various convection-permitting forecasts from different ensemble prediction systems, and find that the value it provides in real-world applications compared to standard analysis methods exhibits clear limitations. In some cases, contour boxplots can provide deeper insight into differences in spatial characteristics between the different ensemble forecasts. Nevertheless, identification of outliers using cBD is not always intuitive, and the method can be especially challenging to implement for flow that exhibits multiple spatial scales; e.g., discrete convective cells embedded within a mesoscale weather system.
Abstract
Estimating and predicting the state of the atmosphere is a probabilistic problem, and often employs an ensemble modeling approach to represent uncertainty in the system. Common methods for examining uncertainty and assessing performance for ensembles emphasize pointwise statistics or marginal distributions. However, these methods lose specific information about individual ensemble members. This paper explores contour band depth (cBD), a method of analyzing uncertainty in terms of contours of scalar fields. cBD is fully nonparametric and induces an ordering on ensemble members that leads to box-and-whisker-plot-type visualizations of uncertainty for two-dimensional data. By applying cBD to synthetic ensembles, we demonstrate that it provides enhanced information about the spatial structure of ensemble uncertainty. We also find that the usefulness of the cBD analysis depends on the presence of multiple modes and multiple scales in the ensemble of contours. Finally, we apply cBD to compare various convection-permitting forecasts from different ensemble prediction systems, and find that the value it provides in real-world applications compared to standard analysis methods exhibits clear limitations. In some cases, contour boxplots can provide deeper insight into differences in spatial characteristics between the different ensemble forecasts. Nevertheless, identification of outliers using cBD is not always intuitive, and the method can be especially challenging to implement for flow that exhibits multiple spatial scales; e.g., discrete convective cells embedded within a mesoscale weather system.
Abstract
Methods to improve the representation of hail in the Thompson-Eidhammer microphysics scheme are explored. A new two-moment and predicted density graupel category is implemented into the Thompson-Eidhammer scheme. Additionally, the one-moment graupel category’s intercept parameter is modified, based on hail observations, to shift the properties of the graupel category to become more hail-like since the category is designed to represent both graupel and hail. Finally, methods to diagnose maximum expected hail size at the surface and aloft are implemented. The original Thompson-Eidhammer version, the newly implemented two-moment and predicted density graupel version, and the modified (to be more hail-like) one-moment version are evaluated using a case that occurred during the Plains Elevated Convection at Night (PECAN) field campaign, during which hail-producing storms merged into a strong mesoscale convective system. The three versions of the scheme are evaluated for their ability to predict hail sizes compared to observed hail sizes from storm reports and estimated from radar, their ability to predict radar reflectivity signatures at various altitudes, and their ability to predict cold-pool features like temperature and wind speed. One key benefit of using the two-moment and predicted density graupel category is that the simulated reflectivity values in the upper-levels of discrete storms are clearly improved. This improvement coincides with a significant reduction in the areal extent of graupel aloft, also seen when using the updated one-moment scheme. The two-moment and predicted density graupel scheme is also better able to predict a wide variety of hail sizes at the surface, including large (> 2-inch diameter) hail that was observed during this case.
Abstract
Methods to improve the representation of hail in the Thompson-Eidhammer microphysics scheme are explored. A new two-moment and predicted density graupel category is implemented into the Thompson-Eidhammer scheme. Additionally, the one-moment graupel category’s intercept parameter is modified, based on hail observations, to shift the properties of the graupel category to become more hail-like since the category is designed to represent both graupel and hail. Finally, methods to diagnose maximum expected hail size at the surface and aloft are implemented. The original Thompson-Eidhammer version, the newly implemented two-moment and predicted density graupel version, and the modified (to be more hail-like) one-moment version are evaluated using a case that occurred during the Plains Elevated Convection at Night (PECAN) field campaign, during which hail-producing storms merged into a strong mesoscale convective system. The three versions of the scheme are evaluated for their ability to predict hail sizes compared to observed hail sizes from storm reports and estimated from radar, their ability to predict radar reflectivity signatures at various altitudes, and their ability to predict cold-pool features like temperature and wind speed. One key benefit of using the two-moment and predicted density graupel category is that the simulated reflectivity values in the upper-levels of discrete storms are clearly improved. This improvement coincides with a significant reduction in the areal extent of graupel aloft, also seen when using the updated one-moment scheme. The two-moment and predicted density graupel scheme is also better able to predict a wide variety of hail sizes at the surface, including large (> 2-inch diameter) hail that was observed during this case.
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.
