Browse
Abstract
Potential factors affecting the inland penetration and orographic modulation of lake-effect precipitation east of Lake Ontario include the environmental (lake, land, and atmospheric) conditions, mode of the lake-effect system, and orographic processes associated with flow across the downstream Tug Hill Plateau (herein Tug Hill), Black River valley, and Adirondack Mountains (herein Adirondacks). In this study we use data from the KTYX WSR-88D, ERA5 reanalysis, New York State Mesonet, and Ontario Winter Lake-effect Systems (OWLeS) field campaign to examine how these factors influence lake-effect characteristics with emphasis on the region downstream of Tug Hill. During an eight-cool-season (16 November–15 April) study period (2012/13–2019/20), total radar-estimated precipitation during lake-effect periods increased gradually from Lake Ontario to upper Tug Hill and decreased abruptly where the Tug Hill escarpment drops into the Black River valley. The axis of maximum precipitation shifted poleward across the northern Black River valley and into the northwestern Adirondacks. In the western Adirondacks, the heaviest lake-effect snowfall periods featured strong, near-zonal boundary layer flow, a deep boundary layer, and a single precipitation band aligned along the long-lake axis. Airborne profiling radar observations collected during OWLeS IOP10 revealed precipitation enhancement over Tug Hill, spillover and shadowing in the Black River valley where a resonant lee wave was present, and precipitation invigoration over the western Adirondacks. These results illustrate the orographic modulation of inland-penetrating lake-effect systems downstream of Lake Ontario and the factors favoring heavy snowfall over the western Adirondacks.
Significance Statement
Inland penetrating lake-effect storms east of Lake Ontario affect remote rural communities, enable a regional winter-sports economy, and contribute to a snowpack that contributes to runoff and flooding during thaws and rain-on-snow events. In this study we illustrate how the region’s three major geographic features—Tug Hill, the Black River valley, and the western Adirondacks—affect the characteristics of lake-effect precipitation, describe the factors contributing to heavy snowfall over the western Adirondacks, and provide an examples of terrain effects in a lake-effect storm observed with a specially instrumented research aircraft.
Abstract
Potential factors affecting the inland penetration and orographic modulation of lake-effect precipitation east of Lake Ontario include the environmental (lake, land, and atmospheric) conditions, mode of the lake-effect system, and orographic processes associated with flow across the downstream Tug Hill Plateau (herein Tug Hill), Black River valley, and Adirondack Mountains (herein Adirondacks). In this study we use data from the KTYX WSR-88D, ERA5 reanalysis, New York State Mesonet, and Ontario Winter Lake-effect Systems (OWLeS) field campaign to examine how these factors influence lake-effect characteristics with emphasis on the region downstream of Tug Hill. During an eight-cool-season (16 November–15 April) study period (2012/13–2019/20), total radar-estimated precipitation during lake-effect periods increased gradually from Lake Ontario to upper Tug Hill and decreased abruptly where the Tug Hill escarpment drops into the Black River valley. The axis of maximum precipitation shifted poleward across the northern Black River valley and into the northwestern Adirondacks. In the western Adirondacks, the heaviest lake-effect snowfall periods featured strong, near-zonal boundary layer flow, a deep boundary layer, and a single precipitation band aligned along the long-lake axis. Airborne profiling radar observations collected during OWLeS IOP10 revealed precipitation enhancement over Tug Hill, spillover and shadowing in the Black River valley where a resonant lee wave was present, and precipitation invigoration over the western Adirondacks. These results illustrate the orographic modulation of inland-penetrating lake-effect systems downstream of Lake Ontario and the factors favoring heavy snowfall over the western Adirondacks.
Significance Statement
Inland penetrating lake-effect storms east of Lake Ontario affect remote rural communities, enable a regional winter-sports economy, and contribute to a snowpack that contributes to runoff and flooding during thaws and rain-on-snow events. In this study we illustrate how the region’s three major geographic features—Tug Hill, the Black River valley, and the western Adirondacks—affect the characteristics of lake-effect precipitation, describe the factors contributing to heavy snowfall over the western Adirondacks, and provide an examples of terrain effects in a lake-effect storm observed with a specially instrumented research aircraft.
Abstract
This study investigates the use of numerical weather prediction (NWP) ensembles to aid refractivity inversion problems during surface ducting conditions. Thirteen sets of measured thermodynamic atmospheric data from an instrumented helicopter during the Wallops Island field experiment are fit to a two-layer parametric surface duct model to characterize the duct. This modeled refractivity is considered “ground truth” for the environment and is used to generate the synthetic radar propagation loss field that then drives the inversion process. The inverse solution (refractivity derived from the synthetic radar data) is compared with this ground truth refractivity. For the inversion process, parameters of the two-layer model are iteratively estimated using genetic algorithms to determine which parameters likely produced the synthetic radar propagation field. Three numerical inversion experiments are conducted. The first experiment utilizes a randomized set of two-layer model parameters to initialize the inversion process, while the second experiment initializes the inversion using NWP ensembles, and the third experiment uses NWP ensembles to both initialize and restrict the parameter search intervals used in the inversion process. The results show that incorporation of NWP data benefits the accuracy and speed of the inversion result. However, in a few cases, an extended NWP ensemble forecast period was needed to encompass the ground truth parameters in the restricted search space. Furthermore, it is found that NWP ensemble populations with smaller spreads are more likely to hinder the inverse process than to aid it.
Abstract
This study investigates the use of numerical weather prediction (NWP) ensembles to aid refractivity inversion problems during surface ducting conditions. Thirteen sets of measured thermodynamic atmospheric data from an instrumented helicopter during the Wallops Island field experiment are fit to a two-layer parametric surface duct model to characterize the duct. This modeled refractivity is considered “ground truth” for the environment and is used to generate the synthetic radar propagation loss field that then drives the inversion process. The inverse solution (refractivity derived from the synthetic radar data) is compared with this ground truth refractivity. For the inversion process, parameters of the two-layer model are iteratively estimated using genetic algorithms to determine which parameters likely produced the synthetic radar propagation field. Three numerical inversion experiments are conducted. The first experiment utilizes a randomized set of two-layer model parameters to initialize the inversion process, while the second experiment initializes the inversion using NWP ensembles, and the third experiment uses NWP ensembles to both initialize and restrict the parameter search intervals used in the inversion process. The results show that incorporation of NWP data benefits the accuracy and speed of the inversion result. However, in a few cases, an extended NWP ensemble forecast period was needed to encompass the ground truth parameters in the restricted search space. Furthermore, it is found that NWP ensemble populations with smaller spreads are more likely to hinder the inverse process than to aid it.
Abstract
We assess the sensitivity of the Weather Research and Forecasting (WRF) Model to the use of different planetary boundary layer (PBL) parameterizations focusing on air temperature and extreme heat conditions. This work aims to evaluate the performance of the WRF Model in simulating temperatures across the Middle East–North Africa (MENA) domain, explain the model biases resulting from the choice of different PBL schemes, and identify the best-performing configuration for the MENA region. Three different PBL schemes are used to downscale the ECMWF ERA-Interim climate over the MENA region at a horizontal resolution of 24 km, for the period 2000–10. These are the Mellor–Yamada–Janjić (MYJ), Yonsei University (YSU), and the asymmetric convective model, version 2 (ACM2). For the evaluation of the WRF runs, we used related meteorological variables from the ERA5 reanalysis, including summer maximum and minimum 2-m air temperature and heat extreme indices. Our results indicate that simulations tend to overestimate maximum temperatures and underestimate minimum temperatures, and we find that model errors are very dependent on the geographic location. The possible physical causes of model biases are investigated through the analysis of additional variables (such as boundary layer height, moisture, and heat fluxes). It is shown that differences among the PBL schemes are associated with differences in vertical mixing strength, which alters the magnitude of the entrainment of free-tropospheric air into the PBL. The YSU is found to be the best-performing scheme, and it is recommended in WRF climate simulations for the MENA region.
Abstract
We assess the sensitivity of the Weather Research and Forecasting (WRF) Model to the use of different planetary boundary layer (PBL) parameterizations focusing on air temperature and extreme heat conditions. This work aims to evaluate the performance of the WRF Model in simulating temperatures across the Middle East–North Africa (MENA) domain, explain the model biases resulting from the choice of different PBL schemes, and identify the best-performing configuration for the MENA region. Three different PBL schemes are used to downscale the ECMWF ERA-Interim climate over the MENA region at a horizontal resolution of 24 km, for the period 2000–10. These are the Mellor–Yamada–Janjić (MYJ), Yonsei University (YSU), and the asymmetric convective model, version 2 (ACM2). For the evaluation of the WRF runs, we used related meteorological variables from the ERA5 reanalysis, including summer maximum and minimum 2-m air temperature and heat extreme indices. Our results indicate that simulations tend to overestimate maximum temperatures and underestimate minimum temperatures, and we find that model errors are very dependent on the geographic location. The possible physical causes of model biases are investigated through the analysis of additional variables (such as boundary layer height, moisture, and heat fluxes). It is shown that differences among the PBL schemes are associated with differences in vertical mixing strength, which alters the magnitude of the entrainment of free-tropospheric air into the PBL. The YSU is found to be the best-performing scheme, and it is recommended in WRF climate simulations for the MENA region.
Abstract
The western United States region, an economic and agricultural powerhouse, is highly dependent on winter snowpack from the mountain west. Coupled with increasing water and renewable electricity demands, the predictability and viability of snowpack resources in a changing climate are becoming increasingly important. In Idaho, specifically, up to 75% of the state’s electricity production comes from hydropower, which is dependent on the timing and volume of spring snowmelt. While we know that 1 April snowpack is declining from SNOTEL observations and is expected to continue to decline as indicated by GCM predictions, our ability to understand the variability of snowfall accumulation and distribution at the regional level is less robust. In this paper, we analyze snowfall events using 0.9-km-resolution WRF simulations to understand the variability of snowfall accumulation and distribution in the mountains of Idaho between 1 October 2016 and 31 April 2017. Various characteristics of snowfall events throughout the season are evaluated, including the spatial coverage, event durations, and snowfall rates, along with the relationship between cloud microphysical variables—particularly liquid and ice water content—on snowfall amounts. Our findings suggest that efficient snowfall conditions—for example, higher levels of elevated supercooled liquid water—can exist throughout the winter season but are more impactful when surface temperatures are near or below freezing. Inefficient snowfall events are common, exceeding 50% of the total snowfall events for the year, with some of those occurring in peak winter. For such events, glaciogenic cloud seeding could make a significant impact on snowpack development and viability in the region.
Significance Statement
The purpose and significance of this study is to better understand the variability of snowfall event accumulation and distribution in the Payette Mountains region of Idaho as it relates to the local topography, the drivers of snowfall events, the cloud microphysical properties, and what constitutes an efficient or inefficient snowfall event (i.e., its ability to convert atmospheric liquid water into snowfall). As part of this process, we identify how many snowfall events in a season are inefficient to determine the number of snowfall events in a season that are candidates for enhancement by glaciogenic cloud seeding.
Abstract
The western United States region, an economic and agricultural powerhouse, is highly dependent on winter snowpack from the mountain west. Coupled with increasing water and renewable electricity demands, the predictability and viability of snowpack resources in a changing climate are becoming increasingly important. In Idaho, specifically, up to 75% of the state’s electricity production comes from hydropower, which is dependent on the timing and volume of spring snowmelt. While we know that 1 April snowpack is declining from SNOTEL observations and is expected to continue to decline as indicated by GCM predictions, our ability to understand the variability of snowfall accumulation and distribution at the regional level is less robust. In this paper, we analyze snowfall events using 0.9-km-resolution WRF simulations to understand the variability of snowfall accumulation and distribution in the mountains of Idaho between 1 October 2016 and 31 April 2017. Various characteristics of snowfall events throughout the season are evaluated, including the spatial coverage, event durations, and snowfall rates, along with the relationship between cloud microphysical variables—particularly liquid and ice water content—on snowfall amounts. Our findings suggest that efficient snowfall conditions—for example, higher levels of elevated supercooled liquid water—can exist throughout the winter season but are more impactful when surface temperatures are near or below freezing. Inefficient snowfall events are common, exceeding 50% of the total snowfall events for the year, with some of those occurring in peak winter. For such events, glaciogenic cloud seeding could make a significant impact on snowpack development and viability in the region.
Significance Statement
The purpose and significance of this study is to better understand the variability of snowfall event accumulation and distribution in the Payette Mountains region of Idaho as it relates to the local topography, the drivers of snowfall events, the cloud microphysical properties, and what constitutes an efficient or inefficient snowfall event (i.e., its ability to convert atmospheric liquid water into snowfall). As part of this process, we identify how many snowfall events in a season are inefficient to determine the number of snowfall events in a season that are candidates for enhancement by glaciogenic cloud seeding.
Abstract
Ice fog typically occurs at temperatures below approximately −30°C. Ice fog formation and persistence are affected by atmospheric processes at different spatial and temporal scales and can be influenced by anthropogenic activities that add vapor to the near-surface atmosphere. Based on meteorological observations from Fairbanks International Airport and Eielson Air Force Base (Alaska) from 1948/49 to 2021/22, we provide an overview of general ice fog climatology at the sites, changes over time, and synoptic-scale upper-level weather patterns common during ice fog occurrence. On average, ice fog occurrence has decreased by 60%–70% over the study period (median number of ice fog days at Fairbanks airport in the period 1950/51–1979/80: 16.5; median in the period 1990/91–2019/20: 6). The average lengths of ice fog events and of the ice fog season have also decreased. Trends are not linear, and rates of change vary over time. The greatest reduction in ice fog occurred during the 1970s and 1980s. Trends in ice fog hours roughly track decreasing trends in hours with cold temperatures. However, the percentage of cold hours in which ice fog occurs has decreased since approximately the 1980s. This result suggests that local changes in air pollution or near-surface moisture may also play an important role in trends in ice fog occurrence. We use self-organizing maps to assess recurring synoptic-scale weather patterns in the upper atmosphere during ice fog conditions in Fairbanks. Ice fog is typically associated with a northerly flow or low pressure gradients over the study area.
Significance Statement
We aim to show when and how often ice fog occurs in the Fairbanks region, how this has changed over time, and what kind of larger-scale weather patterns are common during ice fog. Ice fog strongly reduces visibility and represents a hazard to aviation and other traffic. The number of ice fog hours and days per winter has decreased substantially over the 70-yr period of record. Ice fog is, on average, less persistent now than in the past. The reduction is related to fewer days with cold temperatures, but changes in air pollution and other local factors may also play an important role. Further study is needed to fully attribute the causes of the observed changes.
Abstract
Ice fog typically occurs at temperatures below approximately −30°C. Ice fog formation and persistence are affected by atmospheric processes at different spatial and temporal scales and can be influenced by anthropogenic activities that add vapor to the near-surface atmosphere. Based on meteorological observations from Fairbanks International Airport and Eielson Air Force Base (Alaska) from 1948/49 to 2021/22, we provide an overview of general ice fog climatology at the sites, changes over time, and synoptic-scale upper-level weather patterns common during ice fog occurrence. On average, ice fog occurrence has decreased by 60%–70% over the study period (median number of ice fog days at Fairbanks airport in the period 1950/51–1979/80: 16.5; median in the period 1990/91–2019/20: 6). The average lengths of ice fog events and of the ice fog season have also decreased. Trends are not linear, and rates of change vary over time. The greatest reduction in ice fog occurred during the 1970s and 1980s. Trends in ice fog hours roughly track decreasing trends in hours with cold temperatures. However, the percentage of cold hours in which ice fog occurs has decreased since approximately the 1980s. This result suggests that local changes in air pollution or near-surface moisture may also play an important role in trends in ice fog occurrence. We use self-organizing maps to assess recurring synoptic-scale weather patterns in the upper atmosphere during ice fog conditions in Fairbanks. Ice fog is typically associated with a northerly flow or low pressure gradients over the study area.
Significance Statement
We aim to show when and how often ice fog occurs in the Fairbanks region, how this has changed over time, and what kind of larger-scale weather patterns are common during ice fog. Ice fog strongly reduces visibility and represents a hazard to aviation and other traffic. The number of ice fog hours and days per winter has decreased substantially over the 70-yr period of record. Ice fog is, on average, less persistent now than in the past. The reduction is related to fewer days with cold temperatures, but changes in air pollution and other local factors may also play an important role. Further study is needed to fully attribute the causes of the observed changes.
Abstract
This study analyzed the macro- and microphysical response characteristics of a typical multicell hailstorm after seeding on 27 April 2019 in Weining, China, using X-band dual-polarization radar (YLD1-D) data. An improved X-pol hydrometeor identification algorithm was employed for hydrometeor identification. According to the diffusion of the seeding agents, the seeded hailstorm was graded into three study regions, and the evolution of the seeded hailstorm was divided into four periods. The response characteristics of the seeded hailstorm in each region and period were compared and analyzed. The results show that 1) macroscopically, the decrease in the reflectivity and the height of strong echo mainly occurred in the seeded period, whereas the decrease in the echo top and the height of the storm was mainly in the postseeded period; the echo height variation of the unseeded hailstorm is obviously different from that of the seeded hailstorm; and 2) from the microscopic perspective, the decrease in low-density graupel and supercooled water and the increase in ice crystals and aggregates in the seeded region mainly occurred in the seeded period, which was consistent with the time required for the “benefit competition” after the silver iodide completed nucleation. When compared with the seeded region, the hydrometeors in the unseeded region had an opposite trend (or an in-phase trend with an obviously lower changing rate), which further indicated the impact of the artificial ice nuclei on microphysical processes in the seeded region. For hailstorms with a high content of supercooled droplets and graupel, the key mechanisms of hail suppression are “cloud water glaciation,” “beneficial competition,” and “early rainout.”
Abstract
This study analyzed the macro- and microphysical response characteristics of a typical multicell hailstorm after seeding on 27 April 2019 in Weining, China, using X-band dual-polarization radar (YLD1-D) data. An improved X-pol hydrometeor identification algorithm was employed for hydrometeor identification. According to the diffusion of the seeding agents, the seeded hailstorm was graded into three study regions, and the evolution of the seeded hailstorm was divided into four periods. The response characteristics of the seeded hailstorm in each region and period were compared and analyzed. The results show that 1) macroscopically, the decrease in the reflectivity and the height of strong echo mainly occurred in the seeded period, whereas the decrease in the echo top and the height of the storm was mainly in the postseeded period; the echo height variation of the unseeded hailstorm is obviously different from that of the seeded hailstorm; and 2) from the microscopic perspective, the decrease in low-density graupel and supercooled water and the increase in ice crystals and aggregates in the seeded region mainly occurred in the seeded period, which was consistent with the time required for the “benefit competition” after the silver iodide completed nucleation. When compared with the seeded region, the hydrometeors in the unseeded region had an opposite trend (or an in-phase trend with an obviously lower changing rate), which further indicated the impact of the artificial ice nuclei on microphysical processes in the seeded region. For hailstorms with a high content of supercooled droplets and graupel, the key mechanisms of hail suppression are “cloud water glaciation,” “beneficial competition,” and “early rainout.”
Abstract
Recent worldwide heatwaves have shattered temperature records in many regions. In this study, we applied a dynamical downscaling method on the high-resolution version of the Max Planck Institute Earth System Model (MPI-ESM-1-2-HR) to obtain projections of the summer thermal environments and heatwaves in the Pearl River Delta (PRD) considering three shared socioeconomic pathways (SSP1-2.6, SSP2-4.5, and SSP5-8.5) in the middle and late twenty-first century. Results indicated that relative to the temperatures in the 2010s, the mean increases in the summer (June–September) daytime and nighttime temperatures in the 2040s will be 0.7°–0.8°C and 0.9°–1.1°C, respectively. In the 2090s, the mean difference will be 0.5°–3.1°C and 0.7°–3.4°C, respectively. SSP1-2.6 is the only scenario in which the temperatures in the 2090s are expected to be lower than those in the 2040s. When compared with those in the 2010s, hot extremes are expected to be more frequent, intense, extensive, and longer-lasting in the future in the SSP2-4.5 and SSP5-8.5 scenarios. In the 2010s, a heatwave occurred in the PRD lasted for 6 days on average, with a mean daily maximum temperature of 34.4°C. In the 2040s, the heatwave duration and intensity are expected to increase by 2–3 days and 0.2°–0.4°C in all three scenarios. In the 2090s, these values will become 23 days and 36.0°C in SSP5-8.5. Moreover, a 10-yr extreme high temperature in the 2010s is expected to occur at a monthly frequency from June to September in the 2090s.
Significance Statement
Pearl River Delta (PRD) has been experiencing record-shattering heatwaves in recent years. This study aims to investigate the future trends of summer heatwaves in the PRD by modeling three future scenarios including a sustainable scenario, an intermediate scenario, and a worst-case scenario. Except for the sustainable scenario, summer temperatures in the intermediate and worst-case scenarios will keep increasing, and heatwaves will become more frequent, intense, extensive, and longer-lasting. In the worst-case scenario, extreme heat events that occurred once in 10 years in the 2010s will shorten to once a month in the 2090s. A better understanding of heatwave trends will benefit implementing climate mitigation methods, urban planning, and improving social infrastructure.
Abstract
Recent worldwide heatwaves have shattered temperature records in many regions. In this study, we applied a dynamical downscaling method on the high-resolution version of the Max Planck Institute Earth System Model (MPI-ESM-1-2-HR) to obtain projections of the summer thermal environments and heatwaves in the Pearl River Delta (PRD) considering three shared socioeconomic pathways (SSP1-2.6, SSP2-4.5, and SSP5-8.5) in the middle and late twenty-first century. Results indicated that relative to the temperatures in the 2010s, the mean increases in the summer (June–September) daytime and nighttime temperatures in the 2040s will be 0.7°–0.8°C and 0.9°–1.1°C, respectively. In the 2090s, the mean difference will be 0.5°–3.1°C and 0.7°–3.4°C, respectively. SSP1-2.6 is the only scenario in which the temperatures in the 2090s are expected to be lower than those in the 2040s. When compared with those in the 2010s, hot extremes are expected to be more frequent, intense, extensive, and longer-lasting in the future in the SSP2-4.5 and SSP5-8.5 scenarios. In the 2010s, a heatwave occurred in the PRD lasted for 6 days on average, with a mean daily maximum temperature of 34.4°C. In the 2040s, the heatwave duration and intensity are expected to increase by 2–3 days and 0.2°–0.4°C in all three scenarios. In the 2090s, these values will become 23 days and 36.0°C in SSP5-8.5. Moreover, a 10-yr extreme high temperature in the 2010s is expected to occur at a monthly frequency from June to September in the 2090s.
Significance Statement
Pearl River Delta (PRD) has been experiencing record-shattering heatwaves in recent years. This study aims to investigate the future trends of summer heatwaves in the PRD by modeling three future scenarios including a sustainable scenario, an intermediate scenario, and a worst-case scenario. Except for the sustainable scenario, summer temperatures in the intermediate and worst-case scenarios will keep increasing, and heatwaves will become more frequent, intense, extensive, and longer-lasting. In the worst-case scenario, extreme heat events that occurred once in 10 years in the 2010s will shorten to once a month in the 2090s. A better understanding of heatwave trends will benefit implementing climate mitigation methods, urban planning, and improving social infrastructure.
Abstract
This study investigated the uncertainty involved in statistical downscaling of hydroclimatic time series obtained by artificial neural networks (ANNs). Phase 6 of the Coupled Model Intercomparison Project (CMIP6) general circulation model (GCM) Canadian Earth System Model, version 5 (CanESM5), was used as large-scale predictor data for downscaling temperature and precipitation parameters. Two ANNs, feed forward and long short-term memory (LSTM), were utilized for statistical downscaling. To quantify the uncertainty of downscaling, prediction intervals (PIs) were estimated via the lower upper bound estimation (LUBE) method. To assess performance of proposed models in different climate regimes, data from the Tabriz and Rasht stations in Iran were employed. The calibrated models via historical GCM data were used for future projections via the high-forcing and fossil fuel–driven development scenario shared socioeconomic pathway (SSP) 5-8.5. Projections were compared with the Canadian Regional Climate Model 4 (Can-RCM4) projections via the same scenario. Results indicated that both LSTM-based point predictions and PIs are more accurate than the feedforward neural network (FFNN)-based predictions, with an average of 55% higher Nash–Sutcliffe efficiency (NSE) for point predictions and 25% lower coverage width criterion (CWC) for PIs. Projections suggested that Tabriz is going to experience a warmer climate with an increase in average temperature of 2° and 5°C for near and far futures, respectively, and a drier climate with a 20% decrease in precipitation until 2100. Future projections for the Rasht station, however, suggested a more uniform climate with less seasonal variability. Average precipitation will increase by up to 25% and 70% until near and far future periods, respectively. Ultimately, point predictions show that the average temperature in Rasht will increase by 1°C until the near future and then be a constant average temperature until the far future.
Significance Statement
The downscaling of hydroclimatic parameters is subject to uncertainty. The best way is to provide an area with the highest contingency of them as a prediction interval. The reduction width of such an interval leads to increased confidence in explaining and predicting these processes. We proposed and applied a deep learning–based machine learning method for both point prediction and prediction interval estimation of temperature and precipitation parameters for the future over two different climatic regions. The results show the superiority of such a machine learning–based prediction interval estimation for quantification of the downscaling uncertainty.
Abstract
This study investigated the uncertainty involved in statistical downscaling of hydroclimatic time series obtained by artificial neural networks (ANNs). Phase 6 of the Coupled Model Intercomparison Project (CMIP6) general circulation model (GCM) Canadian Earth System Model, version 5 (CanESM5), was used as large-scale predictor data for downscaling temperature and precipitation parameters. Two ANNs, feed forward and long short-term memory (LSTM), were utilized for statistical downscaling. To quantify the uncertainty of downscaling, prediction intervals (PIs) were estimated via the lower upper bound estimation (LUBE) method. To assess performance of proposed models in different climate regimes, data from the Tabriz and Rasht stations in Iran were employed. The calibrated models via historical GCM data were used for future projections via the high-forcing and fossil fuel–driven development scenario shared socioeconomic pathway (SSP) 5-8.5. Projections were compared with the Canadian Regional Climate Model 4 (Can-RCM4) projections via the same scenario. Results indicated that both LSTM-based point predictions and PIs are more accurate than the feedforward neural network (FFNN)-based predictions, with an average of 55% higher Nash–Sutcliffe efficiency (NSE) for point predictions and 25% lower coverage width criterion (CWC) for PIs. Projections suggested that Tabriz is going to experience a warmer climate with an increase in average temperature of 2° and 5°C for near and far futures, respectively, and a drier climate with a 20% decrease in precipitation until 2100. Future projections for the Rasht station, however, suggested a more uniform climate with less seasonal variability. Average precipitation will increase by up to 25% and 70% until near and far future periods, respectively. Ultimately, point predictions show that the average temperature in Rasht will increase by 1°C until the near future and then be a constant average temperature until the far future.
Significance Statement
The downscaling of hydroclimatic parameters is subject to uncertainty. The best way is to provide an area with the highest contingency of them as a prediction interval. The reduction width of such an interval leads to increased confidence in explaining and predicting these processes. We proposed and applied a deep learning–based machine learning method for both point prediction and prediction interval estimation of temperature and precipitation parameters for the future over two different climatic regions. The results show the superiority of such a machine learning–based prediction interval estimation for quantification of the downscaling uncertainty.
Abstract
This study investigates the characteristics and long-term trends of southwesterly flows around southern Taiwan (SWs) during mei-yu seasons (15 May–15 June) from 1979 to 2022. The results show that the number of SWs in general exhibited an increasing trend over this 44-yr period, with a decadal oscillation starting from a relatively small number in the 1980s and reaching a relative peak in the 2000s. This tendency posts a potential threat to Taiwan because of the increasing trend of heavy rainfall associated with the higher moisture flux of the SWs events. The SWs activity was influenced by the long-term increasing trend of geopotential height gradients and their decadal variability near Taiwan. When the intraseasonal oscillation was evident, the weather system mainly affecting the occurrence of SWs was the low pressure system to the north of Taiwan; when it was weak, the intensity and location of the western North Pacific subtropical high to the south of Taiwan was relatively more important. In addition, the SWs index, which was highly correlated with the precipitation during mei-yu seasons, can effectively reflect the interannual variability of precipitation in Taiwan in periods of different lengths. These findings indicate that the SWs index can be used as a monsoonal precipitation index for Taiwan, especially southern Taiwan.
Abstract
This study investigates the characteristics and long-term trends of southwesterly flows around southern Taiwan (SWs) during mei-yu seasons (15 May–15 June) from 1979 to 2022. The results show that the number of SWs in general exhibited an increasing trend over this 44-yr period, with a decadal oscillation starting from a relatively small number in the 1980s and reaching a relative peak in the 2000s. This tendency posts a potential threat to Taiwan because of the increasing trend of heavy rainfall associated with the higher moisture flux of the SWs events. The SWs activity was influenced by the long-term increasing trend of geopotential height gradients and their decadal variability near Taiwan. When the intraseasonal oscillation was evident, the weather system mainly affecting the occurrence of SWs was the low pressure system to the north of Taiwan; when it was weak, the intensity and location of the western North Pacific subtropical high to the south of Taiwan was relatively more important. In addition, the SWs index, which was highly correlated with the precipitation during mei-yu seasons, can effectively reflect the interannual variability of precipitation in Taiwan in periods of different lengths. These findings indicate that the SWs index can be used as a monsoonal precipitation index for Taiwan, especially southern Taiwan.
Abstract
North Atlantic Ocean hurricane activity exhibits significant variation on multiannual time scales. Advance knowledge of periods of high activity would be beneficial to the insurance industry as well as society in general. Previous studies have shown that climate models initialized with current oceanic and atmospheric conditions, known as decadal prediction systems, are skillful at predicting North Atlantic hurricane activity averaged over periods of 2–10 years. We show that this skill also translates into skillful predictions of real-world U.S. hurricane damage. Using such systems, we have developed a prototype climate service for the insurance industry giving probabilistic forecasts of 5-yr-mean North Atlantic hurricane activity, measured by the total accumulated cyclone energy (ACE index), and 5-yr-total U.S. hurricane damage (given in U.S. dollars). Rather than tracking hurricanes in the decadal systems directly, the forecasts use a relative temperature index known to be strongly linked to hurricane activity. Statistical relationships based on past forecasts of the index and observed hurricane activity and U.S. damage are then used to produce probabilistic forecasts. The predictions of hurricane activity and U.S. damage for the period 2020–24 are high, with ∼95% probabilities of being above average. We note that skill in predicting the temperature index on which the forecasts are based has declined in recent years. More research is therefore needed to understand under which conditions the forecasts are most skillful.
Significance Statement
The purpose of this article is to explain the science and methods behind a recently developed prototype climate service that uses initialized climate models to give probabilistic forecasts of 5-yr-mean North Atlantic Ocean hurricane activity, as well as 5-yr-total associated U.S. hurricane damage. Although skill in predicting North Atlantic hurricane activity on this time scale has been known for some time, a key result in this article is showing that this also leads to predictability in real-world damage. These forecasts could be of benefit to the insurance industry and to society in general.
Abstract
North Atlantic Ocean hurricane activity exhibits significant variation on multiannual time scales. Advance knowledge of periods of high activity would be beneficial to the insurance industry as well as society in general. Previous studies have shown that climate models initialized with current oceanic and atmospheric conditions, known as decadal prediction systems, are skillful at predicting North Atlantic hurricane activity averaged over periods of 2–10 years. We show that this skill also translates into skillful predictions of real-world U.S. hurricane damage. Using such systems, we have developed a prototype climate service for the insurance industry giving probabilistic forecasts of 5-yr-mean North Atlantic hurricane activity, measured by the total accumulated cyclone energy (ACE index), and 5-yr-total U.S. hurricane damage (given in U.S. dollars). Rather than tracking hurricanes in the decadal systems directly, the forecasts use a relative temperature index known to be strongly linked to hurricane activity. Statistical relationships based on past forecasts of the index and observed hurricane activity and U.S. damage are then used to produce probabilistic forecasts. The predictions of hurricane activity and U.S. damage for the period 2020–24 are high, with ∼95% probabilities of being above average. We note that skill in predicting the temperature index on which the forecasts are based has declined in recent years. More research is therefore needed to understand under which conditions the forecasts are most skillful.
Significance Statement
The purpose of this article is to explain the science and methods behind a recently developed prototype climate service that uses initialized climate models to give probabilistic forecasts of 5-yr-mean North Atlantic Ocean hurricane activity, as well as 5-yr-total associated U.S. hurricane damage. Although skill in predicting North Atlantic hurricane activity on this time scale has been known for some time, a key result in this article is showing that this also leads to predictability in real-world damage. These forecasts could be of benefit to the insurance industry and to society in general.
