Browse
Abstract
This paper proposes a new quantitative precipitation estimation (QPE) technique to provide accurate rainfall estimates in complex terrain, where conventional QPE has limitations. The operational radar QPE in Taiwan is mainly based on the simplified relationship between radar reflectivity Z and rain rate R [R(Z) relation] and only utilizes the single-point lowest available echo to estimate rain rates, leading to low accuracy in complex terrain. Here, we conduct QPE using deep learning that extracts features from 3D radar reflectivities to address the above issues. Convolutional neural networks (CNN) are used to analyze contoured frequency by altitude diagrams (CFADs) to generate the QPE. CNN models are trained on existing rain gauges in northern and eastern Taiwan with the 3-yr data during 2015–17 and validated and tested using 2018 data. The weights of heavy rains (≥10 mm h−1) are increased in the model loss calculation to handle the unbalanced rainfall data and improve accuracy. Results show that the CNN outperforms the R(Z) relation based on the 2018 rain gauge data. Furthermore, this research proposes methods to conduct 2D gridded QPE at every pixel by blending estimates from various trained CNN models. Verification based on independent rain gauges shows that the CNN QPE solves the underestimation of the R(Z) relation in mountainous areas. Case studies are presented to visualize the results, showing that the CNN QPE generates better small-scale rainfall features and more accurate precipitation information. This deep learning QPE technique may be helpful for the disaster prevention of small-scale flash floods in complex terrain.
Abstract
This paper proposes a new quantitative precipitation estimation (QPE) technique to provide accurate rainfall estimates in complex terrain, where conventional QPE has limitations. The operational radar QPE in Taiwan is mainly based on the simplified relationship between radar reflectivity Z and rain rate R [R(Z) relation] and only utilizes the single-point lowest available echo to estimate rain rates, leading to low accuracy in complex terrain. Here, we conduct QPE using deep learning that extracts features from 3D radar reflectivities to address the above issues. Convolutional neural networks (CNN) are used to analyze contoured frequency by altitude diagrams (CFADs) to generate the QPE. CNN models are trained on existing rain gauges in northern and eastern Taiwan with the 3-yr data during 2015–17 and validated and tested using 2018 data. The weights of heavy rains (≥10 mm h−1) are increased in the model loss calculation to handle the unbalanced rainfall data and improve accuracy. Results show that the CNN outperforms the R(Z) relation based on the 2018 rain gauge data. Furthermore, this research proposes methods to conduct 2D gridded QPE at every pixel by blending estimates from various trained CNN models. Verification based on independent rain gauges shows that the CNN QPE solves the underestimation of the R(Z) relation in mountainous areas. Case studies are presented to visualize the results, showing that the CNN QPE generates better small-scale rainfall features and more accurate precipitation information. This deep learning QPE technique may be helpful for the disaster prevention of small-scale flash floods in complex terrain.
Abstract
A machine learning method based on spatial convolution to capture complex spatial precipitation patterns is proposed to identify and reduce biases affecting predictions of a dynamical model. The method is based on a combination of a classification and dual-regression model approach using modified U-Net convolutional neural networks (CNN) to postprocess daily accumulated precipitation over the U.S. West Coast. In this study, we leverage 34 years of high-resolution deterministic Western Weather Research and Forecasting (West-WRF) precipitation reforecasts as training data for the U-Net CNN. The data are split such that the test set contains 4 water years of data that encompass characteristic West Coast precipitation regimes: El Niño, La Niña, and dry and wet El Niño–Southern Oscillation (ENSO neutral) water years. On the unseen 4-yr dataset, the trained CNN yields a 12.9%–15.9% reduction in root-mean-square error (RMSE) and 2.7%–3.4% improvement in Pearson correlation (PC) over West-WRF for lead times of 1–4 days. Compared to an adapted model output statistics correction, the CNN reduces RMSE by 7.4%–8.9% and improves PC by 3.3%–4.2% across all events. Effectively, the CNN adds more than a day of predictive skill when compared to West-WRF. The CNN outperforms the other methods also for the prediction of extreme events, which we define as the top 10% of events with the greatest average daily accumulated precipitation. The improvement over West-WRF’s RMSE (PC) for these events is 19.8%–21.0% (4.9%–5.5%) and MOS’s RMSE (PC) is 8.8%–9.7% (4.2%–4.7%). Hence, the proposed U-Net CNN shows significantly improved forecast skill over existing methods, highlighting a promising path forward for improving precipitation forecasts.
Significance Statement
Extreme precipitation events and atmospheric rivers, which contain narrow bands of water vapor transport, can cause millions of dollars in damages. We demonstrate the utility of a computer vision-based machine learning technique for improving precipitation forecasts. We show that there is a significant increase in predictive accuracy for daily accumulated precipitation using these machine learning methods, over a 4-yr period of unseen cases, including those corresponding to the extreme precipitation associated with atmospheric rivers.
Abstract
A machine learning method based on spatial convolution to capture complex spatial precipitation patterns is proposed to identify and reduce biases affecting predictions of a dynamical model. The method is based on a combination of a classification and dual-regression model approach using modified U-Net convolutional neural networks (CNN) to postprocess daily accumulated precipitation over the U.S. West Coast. In this study, we leverage 34 years of high-resolution deterministic Western Weather Research and Forecasting (West-WRF) precipitation reforecasts as training data for the U-Net CNN. The data are split such that the test set contains 4 water years of data that encompass characteristic West Coast precipitation regimes: El Niño, La Niña, and dry and wet El Niño–Southern Oscillation (ENSO neutral) water years. On the unseen 4-yr dataset, the trained CNN yields a 12.9%–15.9% reduction in root-mean-square error (RMSE) and 2.7%–3.4% improvement in Pearson correlation (PC) over West-WRF for lead times of 1–4 days. Compared to an adapted model output statistics correction, the CNN reduces RMSE by 7.4%–8.9% and improves PC by 3.3%–4.2% across all events. Effectively, the CNN adds more than a day of predictive skill when compared to West-WRF. The CNN outperforms the other methods also for the prediction of extreme events, which we define as the top 10% of events with the greatest average daily accumulated precipitation. The improvement over West-WRF’s RMSE (PC) for these events is 19.8%–21.0% (4.9%–5.5%) and MOS’s RMSE (PC) is 8.8%–9.7% (4.2%–4.7%). Hence, the proposed U-Net CNN shows significantly improved forecast skill over existing methods, highlighting a promising path forward for improving precipitation forecasts.
Significance Statement
Extreme precipitation events and atmospheric rivers, which contain narrow bands of water vapor transport, can cause millions of dollars in damages. We demonstrate the utility of a computer vision-based machine learning technique for improving precipitation forecasts. We show that there is a significant increase in predictive accuracy for daily accumulated precipitation using these machine learning methods, over a 4-yr period of unseen cases, including those corresponding to the extreme precipitation associated with atmospheric rivers.
Abstract
Multiple linear regression models were developed for 1-3-day lead forecasts of maximum and minimum temperature for two locations in the city of Lima, in the central coast of Peru (12°S), and contrasted with the operational forecasts issued by the National Meteorological and Hydrological Service - SENAMHI and the output of a regional numerical atmospheric model. We developed empirical models, fitted to data from the 2000-2013 period, and verified their skill for the 2014-2019 period. Since El Niño produces a strong low-frequency signal, the models focus on the high-frequency weather and subseasonal variability (60-day cutoff).
The empirical models outperformed the operational forecasts and the numerical model. For instance, the high-frequency annual correlation coefficient and root mean square error (RMSE) for the 1-day lead forecasts were 0.37-0.53 and 0.74-1.76°C for the empirical model, respectively, but around −0.05-0.24 and 0.88-4.21°C in the operational case. Only three predictors were considered for the models, including persistence and large-scale atmospheric indices. Contrary to our belief, the model skill was lowest for the austral winter (June-August), when the extratropical influence is largest, suggesting an enhanced role of local effects. Including local specific humidity as a predictor for minimum temperature at the inland location substantially increased the skill and reduced its seasonality.
There were cases in which both the operational and empirical forecast had similar strong errors and we suggest mesoscale circulations, such as the Low-Level Cyclonic Vortex over the ocean, as the culprit. Incorporating such information could be valuable for improving the forecasts.
Abstract
Multiple linear regression models were developed for 1-3-day lead forecasts of maximum and minimum temperature for two locations in the city of Lima, in the central coast of Peru (12°S), and contrasted with the operational forecasts issued by the National Meteorological and Hydrological Service - SENAMHI and the output of a regional numerical atmospheric model. We developed empirical models, fitted to data from the 2000-2013 period, and verified their skill for the 2014-2019 period. Since El Niño produces a strong low-frequency signal, the models focus on the high-frequency weather and subseasonal variability (60-day cutoff).
The empirical models outperformed the operational forecasts and the numerical model. For instance, the high-frequency annual correlation coefficient and root mean square error (RMSE) for the 1-day lead forecasts were 0.37-0.53 and 0.74-1.76°C for the empirical model, respectively, but around −0.05-0.24 and 0.88-4.21°C in the operational case. Only three predictors were considered for the models, including persistence and large-scale atmospheric indices. Contrary to our belief, the model skill was lowest for the austral winter (June-August), when the extratropical influence is largest, suggesting an enhanced role of local effects. Including local specific humidity as a predictor for minimum temperature at the inland location substantially increased the skill and reduced its seasonality.
There were cases in which both the operational and empirical forecast had similar strong errors and we suggest mesoscale circulations, such as the Low-Level Cyclonic Vortex over the ocean, as the culprit. Incorporating such information could be valuable for improving the forecasts.
Abstract
The impact of assimilating dropsonde data from the 2020 Atmospheric River (AR) Reconnaissance (ARR) field campaign on operational numerical weather forecasts was assessed. Two experiments were executed for the period from 24 January to 18 March 2020 using the National Centers for Environmental Prediction (NCEP) Global Forecast System version 15 (GFSv15) with a four-dimensional hybrid ensemble-variational (4DEnVar) data assimilation system. The control run (CTRL) used all of the routinely assimilated data and included data from 628 ARR dropsondes, whereas the denial run (DENY) excluded the dropsonde data. Results from 17 Intensive Observing Periods (IOPs) indicate a mixed impact for mean sea-level pressure and geopotential height over the Pacific North American (PNA) region in CTRL compared to DENY. The overall local impact over the U.S. West Coast and Gulf of Alaska for the 17 IOPs is neutral (−0.45%) for Integrated Vapor Transport (IVT), but positive for wind and moisture profiles (0.5-1.0%), with a spectrum of statistically significant positive and negative impacts for various IOPs. The positive dropsonde data impact on precipitation forecasts over U.S. West Coast domains appears driven, in part, by improved low-level moisture and wind fields at short-forecast lead times. Indeed, data gaps, especially for accurate and unbiased moisture profiles and wind fields, can be at least partially mitigated to improve U.S. West Coast precipitation forecasts.
Abstract
The impact of assimilating dropsonde data from the 2020 Atmospheric River (AR) Reconnaissance (ARR) field campaign on operational numerical weather forecasts was assessed. Two experiments were executed for the period from 24 January to 18 March 2020 using the National Centers for Environmental Prediction (NCEP) Global Forecast System version 15 (GFSv15) with a four-dimensional hybrid ensemble-variational (4DEnVar) data assimilation system. The control run (CTRL) used all of the routinely assimilated data and included data from 628 ARR dropsondes, whereas the denial run (DENY) excluded the dropsonde data. Results from 17 Intensive Observing Periods (IOPs) indicate a mixed impact for mean sea-level pressure and geopotential height over the Pacific North American (PNA) region in CTRL compared to DENY. The overall local impact over the U.S. West Coast and Gulf of Alaska for the 17 IOPs is neutral (−0.45%) for Integrated Vapor Transport (IVT), but positive for wind and moisture profiles (0.5-1.0%), with a spectrum of statistically significant positive and negative impacts for various IOPs. The positive dropsonde data impact on precipitation forecasts over U.S. West Coast domains appears driven, in part, by improved low-level moisture and wind fields at short-forecast lead times. Indeed, data gaps, especially for accurate and unbiased moisture profiles and wind fields, can be at least partially mitigated to improve U.S. West Coast precipitation forecasts.
Abstract
In a mid-latitude coastal region such as the New York Bight (NYB), the general thermodynamic structure and dynamics of the sea breeze circulation is poorly understood. The NYB sea breeze circulation is often amplified by and coterminous with other regional characteristics and phenomena such as complex coastal topology, a low-level jet (LLJ), and coastal upwelling. While typically considered a summertime phenomenon, the NYB sea breeze circulation occurs year-round.
This study creates a methodology to objectively identify sea breeze days and their associated LLJs from 2010 to 2020. Filtering parameters include surface-based observations of sea level pressure (SLP) gradient and diurnal tendencies, afternoon wind speed and direction tendencies, air temperature gradient, and the dewpoint depression. LLJs associated with the sea breeze circulation typically occur within 150 – 300 m MSL and are identified using a coastal New York State Mesonet (NYSM) profiler site. Along coastal Long Island, there are on average 32 sea breeze days annually, featuring winds consistently backing to the south and strengthening at or around 1800 UTC (1400 EDT). The NYB LLJ is most frequent in the summer months.
Sea breeze days are classified into two categories: Classic and Hybrid. A Classic sea breeze is driven primarily by both cross-shore pressure and temperature gradients, with light background winds; while a Hybrid sea breeze occurs in combination with other larger-scale features, such as frontal systems. Both types of sea breeze are similarly distributed with a maximum frequency during July.
Abstract
In a mid-latitude coastal region such as the New York Bight (NYB), the general thermodynamic structure and dynamics of the sea breeze circulation is poorly understood. The NYB sea breeze circulation is often amplified by and coterminous with other regional characteristics and phenomena such as complex coastal topology, a low-level jet (LLJ), and coastal upwelling. While typically considered a summertime phenomenon, the NYB sea breeze circulation occurs year-round.
This study creates a methodology to objectively identify sea breeze days and their associated LLJs from 2010 to 2020. Filtering parameters include surface-based observations of sea level pressure (SLP) gradient and diurnal tendencies, afternoon wind speed and direction tendencies, air temperature gradient, and the dewpoint depression. LLJs associated with the sea breeze circulation typically occur within 150 – 300 m MSL and are identified using a coastal New York State Mesonet (NYSM) profiler site. Along coastal Long Island, there are on average 32 sea breeze days annually, featuring winds consistently backing to the south and strengthening at or around 1800 UTC (1400 EDT). The NYB LLJ is most frequent in the summer months.
Sea breeze days are classified into two categories: Classic and Hybrid. A Classic sea breeze is driven primarily by both cross-shore pressure and temperature gradients, with light background winds; while a Hybrid sea breeze occurs in combination with other larger-scale features, such as frontal systems. Both types of sea breeze are similarly distributed with a maximum frequency during July.
Abstract
The NWS/NCEP recently implemented a new global deterministic aerosol forecast model named the Global Ensemble Forecast Systems Aerosols (GEFS-Aerosols), which is based on the Finite Volume version 3 GFS (FV3GFS). It replaced the operational NOAA Environmental Modeling System (NEMS) GFS Aerosol Component version 2 (NGACv2), which was based on a global spectral model (GSM). GEFS-Aerosols uses aerosol modules from the GOCART previously integrated in the WRF Model with Chemistry (WRF-Chem), FENGSHA dust scheme, and several other updates. In this study, we have extensively evaluated aerosol optical depth (AOD) forecasts from GEFS-Aerosols against various observations over a timespan longer than one year (2019–20). The total AOD improvement (in terms of seasonal mean) in GEFS-Aerosols is about 40% compared to NGACv2 in the fall and winter season of 2019. In terms of aerosol species, the biggest improvement came from the enhanced representation of biomass burning aerosol species as GEFS-Aerosols is able to capture more fire events in southern Africa, South America, and Asia than its predecessor. Dust AODs reproduce the seasonal variation over Africa and the Middle East. We have found that correlation of total AOD over large regions of the globe remains consistent for forecast days 3–5. However, we have found that GEFS-Aerosols generates some systematic positive biases for organic carbon AOD near biomass burning regions and sulfate AOD over prediction over East Asia. The addition of a data assimilation capability to GEFS-Aerosols in the near future is expected to address these biases and provide a positive impact to aerosol forecasts by the model.
Significance Statement
The purpose of this study is to quantify improvements associated with the newly implemented global aerosol forecast model at NWS/NCEP. The monthly and seasonal variations of AOD forecasts of various aerosol regimes are overall consistent with the observations. Our results provide a guide to downstream regional air quality models like CMAQ that will use GEFS-Aerosols to provide lateral boundary conditions.
Abstract
The NWS/NCEP recently implemented a new global deterministic aerosol forecast model named the Global Ensemble Forecast Systems Aerosols (GEFS-Aerosols), which is based on the Finite Volume version 3 GFS (FV3GFS). It replaced the operational NOAA Environmental Modeling System (NEMS) GFS Aerosol Component version 2 (NGACv2), which was based on a global spectral model (GSM). GEFS-Aerosols uses aerosol modules from the GOCART previously integrated in the WRF Model with Chemistry (WRF-Chem), FENGSHA dust scheme, and several other updates. In this study, we have extensively evaluated aerosol optical depth (AOD) forecasts from GEFS-Aerosols against various observations over a timespan longer than one year (2019–20). The total AOD improvement (in terms of seasonal mean) in GEFS-Aerosols is about 40% compared to NGACv2 in the fall and winter season of 2019. In terms of aerosol species, the biggest improvement came from the enhanced representation of biomass burning aerosol species as GEFS-Aerosols is able to capture more fire events in southern Africa, South America, and Asia than its predecessor. Dust AODs reproduce the seasonal variation over Africa and the Middle East. We have found that correlation of total AOD over large regions of the globe remains consistent for forecast days 3–5. However, we have found that GEFS-Aerosols generates some systematic positive biases for organic carbon AOD near biomass burning regions and sulfate AOD over prediction over East Asia. The addition of a data assimilation capability to GEFS-Aerosols in the near future is expected to address these biases and provide a positive impact to aerosol forecasts by the model.
Significance Statement
The purpose of this study is to quantify improvements associated with the newly implemented global aerosol forecast model at NWS/NCEP. The monthly and seasonal variations of AOD forecasts of various aerosol regimes are overall consistent with the observations. Our results provide a guide to downstream regional air quality models like CMAQ that will use GEFS-Aerosols to provide lateral boundary conditions.
Abstract
Historical observations of severe weather and simulated severe weather environments (i.e., features) from the Global Ensemble Forecast System v12 (GEFSv12) Reforecast Dataset (GEFS/R) are used in conjunction to train and test random forest (RF) machine learning (ML) models to probabilistically forecast severe weather out to days 4–8. RFs are trained with ∼9 years of the GEFS/R and severe weather reports to establish statistical relationships. Feature engineering is briefly explored to examine alternative methods for gathering features around observed events, including simplifying features using spatial averaging and increasing the GEFS/R ensemble size with time lagging. Validated RF models are tested with ∼1.5 years of real-time forecast output from the operational GEFSv12 ensemble and are evaluated alongside expert human-generated outlooks from the Storm Prediction Center (SPC). Both RF-based forecasts and SPC outlooks are skillful with respect to climatology at days 4 and 5 with diminishing skill thereafter. The RF-based forecasts exhibit tendencies to slightly underforecast severe weather events, but they tend to be well-calibrated at lower probability thresholds. Spatially averaging predictors during RF training allows for prior-day thermodynamic and kinematic environments to generate skillful forecasts, while time lagging acts to expand the forecast areas, increasing resolution but decreasing overall skill. The results highlight the utility of ML-generated products to aid SPC forecast operations into the medium range.
Significance Statement
Medium-range severe weather forecasts generated from statistical models are explored here alongside operational forecasts from the Storm Prediction Center (SPC). Human forecasters at the SPC rely on traditional numerical weather prediction model output to make medium-range outlooks and statistical products that mimic operational forecasts can be used as guidance tools for forecasters. The statistical models relate simulated severe weather environments from a global weather model to historical records of severe weather and perform noticeably better than human-generated outlooks at shorter lead times (e.g., day 4 and 5) and are capable of capturing the general location of severe weather events 8 days in advance. The results highlight the value in these data-driven methods in supporting operational forecasting.
Abstract
Historical observations of severe weather and simulated severe weather environments (i.e., features) from the Global Ensemble Forecast System v12 (GEFSv12) Reforecast Dataset (GEFS/R) are used in conjunction to train and test random forest (RF) machine learning (ML) models to probabilistically forecast severe weather out to days 4–8. RFs are trained with ∼9 years of the GEFS/R and severe weather reports to establish statistical relationships. Feature engineering is briefly explored to examine alternative methods for gathering features around observed events, including simplifying features using spatial averaging and increasing the GEFS/R ensemble size with time lagging. Validated RF models are tested with ∼1.5 years of real-time forecast output from the operational GEFSv12 ensemble and are evaluated alongside expert human-generated outlooks from the Storm Prediction Center (SPC). Both RF-based forecasts and SPC outlooks are skillful with respect to climatology at days 4 and 5 with diminishing skill thereafter. The RF-based forecasts exhibit tendencies to slightly underforecast severe weather events, but they tend to be well-calibrated at lower probability thresholds. Spatially averaging predictors during RF training allows for prior-day thermodynamic and kinematic environments to generate skillful forecasts, while time lagging acts to expand the forecast areas, increasing resolution but decreasing overall skill. The results highlight the utility of ML-generated products to aid SPC forecast operations into the medium range.
Significance Statement
Medium-range severe weather forecasts generated from statistical models are explored here alongside operational forecasts from the Storm Prediction Center (SPC). Human forecasters at the SPC rely on traditional numerical weather prediction model output to make medium-range outlooks and statistical products that mimic operational forecasts can be used as guidance tools for forecasters. The statistical models relate simulated severe weather environments from a global weather model to historical records of severe weather and perform noticeably better than human-generated outlooks at shorter lead times (e.g., day 4 and 5) and are capable of capturing the general location of severe weather events 8 days in advance. The results highlight the value in these data-driven methods in supporting operational forecasting.
Abstract
A sensitivity analysis for horizontal localization scale is performed for a numerical weather prediction (NWP) system that uses a 30-second update to refresh a 500-m mesh with observations from a new-generation multi-parameter phased array weather radar (MP-PAWR). Testing is performed using three case studies of convective weather events that occurred during August/September 2019, with the aim to determine the most suitable scale for short-range forecasting of precipitating convective systems and better understand model behavior to a rapid update cycle. Results showed that while the model could provide useful skill at lead times up to 30-minutes, forecasts would consistently over-estimate rainfall and were unable to outperform nowcasts performed with a simple advection model. Using a larger localization scale e.g., 4-km, generated stronger convective and dynamical instability in the analyzes that made conditions more favorable for spurious and intense convection to develop in forecasts. It was demonstrated that lowering the localization scale reduced the size of analysis increments during early cycling, limiting the buildup of these conditions. Improved representation of the localized convection in the initial conditions was suggested as an important step to mitigating this issue in the model.
Abstract
A sensitivity analysis for horizontal localization scale is performed for a numerical weather prediction (NWP) system that uses a 30-second update to refresh a 500-m mesh with observations from a new-generation multi-parameter phased array weather radar (MP-PAWR). Testing is performed using three case studies of convective weather events that occurred during August/September 2019, with the aim to determine the most suitable scale for short-range forecasting of precipitating convective systems and better understand model behavior to a rapid update cycle. Results showed that while the model could provide useful skill at lead times up to 30-minutes, forecasts would consistently over-estimate rainfall and were unable to outperform nowcasts performed with a simple advection model. Using a larger localization scale e.g., 4-km, generated stronger convective and dynamical instability in the analyzes that made conditions more favorable for spurious and intense convection to develop in forecasts. It was demonstrated that lowering the localization scale reduced the size of analysis increments during early cycling, limiting the buildup of these conditions. Improved representation of the localized convection in the initial conditions was suggested as an important step to mitigating this issue in the model.
Abstract
Using a 3-km regional ensemble prediction system (EPS), this study tested a three-dimensional (3D) rescaling mask for initial condition (IC) perturbation. Whether the 3D mask-based EPS improves ensemble forecasts over current two-dimensional (2D) mask-based EPS has been evaluated in three aspects: ensemble mean, spread, and probability. The forecasts of wind, temperature, geopotential height, sea level pressure, and precipitation were examined for a summer month (1–28 July 2018) and a winter month (1–27 February 2019) over a region in North China. The EPS was run twice per day (initiated at 0000 and 1200 UTC) to 36 h in forecast length, providing 56 warm-season forecast cases and 54 cold-season cases for verification. The warm and cold seasons are verified separately for comparison. The study found the following: 1) The vertical profile of IC perturbation becomes closer to that of analysis uncertainty with the 3D rescaling mask. 2) Ensemble performance is significantly improved in all three aspects. The biggest improvement is in the ensemble spread, followed by the probabilistic forecast, and the least improvement is in the ensemble mean forecast. Larger improvements are seen in the warm season than in the cold season. 3) More improvement is in the shorter time range (<24 h) than in the longer range. 4) Surface and lower-level variables are improved more than upper-level ones. 5) The underlying mechanism for the improvement has been investigated. Convective instability is found to be responsible for the spread increment and, thus, overall ensemble forecast improvement. Therefore, using a 3D rescaling mask is recommended for an EPS to increase its utility especially for shorter time range and surface weather elements.
Significant Statement
A weather prediction model is a complex system that consists of nonlinear differential equations. Small errors in either its inputs or model itself will grow with time during model integration, which will contaminate a forecast. To quantify such contamination (“uncertainty”) of a forecast, the ensemble forecasting technique is used. An ensemble of forecasts is a multiple of model runs at the same time but with slightly “perturbed” inputs or model versions. These small perturbations are supposed to represent true “uncertainty” in inputs or model representation. This study proposed a technique that makes a perturbation’s vertical structure more resemble real uncertainty (intrinsic error) in input data and confirmed that it can significantly improve ensemble forecast quality especially for a shorter time range and lower-level weather elements. It is found that convective instability is responsible for the improvement.
Abstract
Using a 3-km regional ensemble prediction system (EPS), this study tested a three-dimensional (3D) rescaling mask for initial condition (IC) perturbation. Whether the 3D mask-based EPS improves ensemble forecasts over current two-dimensional (2D) mask-based EPS has been evaluated in three aspects: ensemble mean, spread, and probability. The forecasts of wind, temperature, geopotential height, sea level pressure, and precipitation were examined for a summer month (1–28 July 2018) and a winter month (1–27 February 2019) over a region in North China. The EPS was run twice per day (initiated at 0000 and 1200 UTC) to 36 h in forecast length, providing 56 warm-season forecast cases and 54 cold-season cases for verification. The warm and cold seasons are verified separately for comparison. The study found the following: 1) The vertical profile of IC perturbation becomes closer to that of analysis uncertainty with the 3D rescaling mask. 2) Ensemble performance is significantly improved in all three aspects. The biggest improvement is in the ensemble spread, followed by the probabilistic forecast, and the least improvement is in the ensemble mean forecast. Larger improvements are seen in the warm season than in the cold season. 3) More improvement is in the shorter time range (<24 h) than in the longer range. 4) Surface and lower-level variables are improved more than upper-level ones. 5) The underlying mechanism for the improvement has been investigated. Convective instability is found to be responsible for the spread increment and, thus, overall ensemble forecast improvement. Therefore, using a 3D rescaling mask is recommended for an EPS to increase its utility especially for shorter time range and surface weather elements.
Significant Statement
A weather prediction model is a complex system that consists of nonlinear differential equations. Small errors in either its inputs or model itself will grow with time during model integration, which will contaminate a forecast. To quantify such contamination (“uncertainty”) of a forecast, the ensemble forecasting technique is used. An ensemble of forecasts is a multiple of model runs at the same time but with slightly “perturbed” inputs or model versions. These small perturbations are supposed to represent true “uncertainty” in inputs or model representation. This study proposed a technique that makes a perturbation’s vertical structure more resemble real uncertainty (intrinsic error) in input data and confirmed that it can significantly improve ensemble forecast quality especially for a shorter time range and lower-level weather elements. It is found that convective instability is responsible for the improvement.
