Abstract Subgrade plain soil filling is a crucial process in road construction prone to particulate matter (PM) emissions. In this study, we conducted a key factor analysis and concentration prediction of PM emissions in a subgrade plain soil filling process based on interpretable machine learning algorithms. Real-time monitoring of particulate pollutant emissions, using a drone and distributed ground-fixed sensors, was conducted to collect data on a new national highway project using spatiotemporal Internet of Things technology. The interpretable machine learning models were developed using Bayesian optimization to predict PM 10 and PM 2.5 concentrations. The results revealed that extreme gradient boosting exhibited the highest predictive performance. Additionally, Shapley additive explanations, partial dependence plot, and individual conditional expectation provided enhanced interpretability. PM 2.5 background concentration and humidity mostly affected the prediction results. Higher PM 2.5 background concentration levels notably increased the predicted values of both PM 10 and PM 2.5 . Moreover, continuous PM emissions from construction activities can mask the natural reduction of PM 10 and PM 2.5 caused by high humidity. This study provides guidance for dust estimation and environmental protection during road construction. Significance Statement Particulate matter (PM) pollution is a major source of air pollution and health problems globally. PM exposure causes many premature deaths worldwide each year according to the World Health Organization. Subgrade plain soil filling is a crucial process in road construction prone to PM emissions. This study contributes to accurately monitor and predict PM by using spatiotemporal Internet of Things and interpretable machine learning, which helps to understand key influencing factors for better dust suppression during road construction. Further, it provides valuable data for environmental protection agencies to make informed decisions, thereby protecting ambient air quality and public health.
Abstract This work is part of the EU-I-CHANGE Living-Labs project in Jerusalem-Israel, including citizens-science data along with cellular providers. Cellular network data enables high spatial resolution humidity monitoring, compared to low-spatial resolution observations from surface stations. Contrary to stations, commercial microwave link (CML) is with high spatial resolution. Humidity is an important variable in atmospheric processes and closely linked to clouds and rain. Humidity above ground is highly influenced by surface characteristics and measuring the near-surface humidity, where most of the sinks and sources of humidity are, can be done via a novel approach of using CMLs, a large part of the cellular networks backhaul. The data used includes Absolute Humidity (AH) measured at four links at and around the Israeli Meteorological Service (IMS) stations. Plots show AH values for each link against IMS stations and ICON model AH values. Correlations, mean, RMSE and index of agreement values were calculated for the comparisons between the two AH values. Results indicate that CML-derived AH captures a substantial portion of the observed humidity variability and shows generally good agreement with both IMS observations and ICON analyses during summer 2021. For example, for Generali IMS station, Correlation values reached 0.75 (CML-IMS) and 0.74 (CML-ICON), with average RMSE ~4 g m −3 . Mean AH for CML, IMS, and ICON; are 13.0, 11.1, 11.2 g m −3 , correspondingly. This is the first study in a city with such a unique climate such as in Jerusalem, with Mediterranean climate in the west vs. arid climate in the east. These findings suggest that CML observations may provide useful high-resolution humidity information in complex urban environments, although further validation across additional seasons and locations is required.
Abstract Accurate near-surface wind forecasts over complex terrain are crucial for various applications, including air quality dispersion, wind energy assessment, and hydrometeorology. However, mesoscale models such as the Weather Research and Forecasting (WRF) Model frequently exhibit a persistent systematic high wind bias in mountainous regions. This study introduces a physically based formulation for the effective roughness length ( ) to better incorporate subgrid-scale (SGS) orographic drag. Derived from large-eddy simulations (LESs), the new formulation utilizes the first three statistical moments (mean, standard deviation, and skewness) of high-resolution topographic data. The performance of the formulation was evaluated through a year-long, 3-km-resolution WRF simulation over the Korean Peninsula for 2016. Validation against 95 surface stations and seven radiosonde sounding stations demonstrates that the new formulation substantially improves the model’s predictive skill. The proposed scheme reduced the annual-mean RMSE for 10-m wind speed from 2.0 m s −1 in the control run to 1.4 m s −1 , representing a 30% reduction in the overall error magnitude. Unlike preexisting empirical drag schemes, this improvement was consistently maintained across both daytime and nighttime periods, and its influence extended vertically within the atmospheric boundary layer (ABL). These results confirm that the LES-based formulation provides a physically robust correction for momentum sinks, offering enhanced fidelity for meteorological and air quality forecasting in complex terrain. Significance Statement Weather forecast models often exhibit a persistent systematic high wind bias over complex terrain, which limits their reliability for critical societal applications. This study introduces a new, physically based method to improve the representation of surface drag from unresolved topography in the Weather Research and Forecasting (WRF) Model by utilizing an effective roughness length ( ) derived from high-resolution topographic statistics. Our evaluation demonstrates that this new parameterization effectively reduces the model’s overestimation of surface wind speeds, decreasing the annual-mean root-mean-square error (RMSE) from 2.0 to 1.4 m s −1 , representing a 30% improvement in overall forecast precision. This substantial increase in forecast accuracy directly benefits applied fields such as air quality dispersion modeling, wind energy assessment, and hydrometeorology, all of which depend on reliable near-surface and atmospheric boundary layer (ABL) wind data.
Abstract Reliable seasonal drought forecasts are critical for water management in the Intermountain West, yet hydroclimate predictability in this region remains limited because canonical El Niño-Southern Oscillation (ENSO) teleconnections are weak and spatially inconsistent. Here, we show that Utah experiences a robust and coherent drought response during the second dip of multi-year La Niña events, in contrast to the weak and variable responses observed during preceding El Niño and first-dip La Niña winters. Composite analyses of gridded reanalysis products, statewide climate indices, and station-based observations reveal that second-dip La Niña events are characterized by reduced winter–spring precipitation, anomalously warm temperatures, and increased numbers of dry and warm days. These conditions result in diminished April 1 snow water equivalent, suppressed summer streamflow, reduced soil moisture, and strongly negative Palmer Drought Severity Index anomalies across Utah. By contrasting early (1951–1990) and recent (1991–2023) periods, we further show that these drought responses have intensified under a warming climate. While the large-scale circulation anomalies associated with second-dip La Niña events primarily reflect internal ENSO-related variability, rising temperatures amplify drought severity through increased evaporative demand, reduced runoff efficiency, and earlier snowmelt timing. Together, these results indicate that second-dip La Niña winters provide an enhanced and physically grounded source of seasonal drought predictability for Utah. Incorporating second-dip La Niña signals into drought early warning systems may therefore improve water resource planning in snow-dominated, topographically complex regions of the Intermountain West.
Abstract Historical gridded climate datasets of precipitation and temperature can provide different representations of past climate conditions at the same locations and time periods depending on the sources and methods used to produce the datasets. Here, we evaluate and quantify these differences for eight high-resolution (~ 10km) gridded climate datasets over British Columbia (BC), Canada, where sparse station records and complex meteorology produce many challenges in the representation of past climate. The eight datasets considered include four station observation-based sources (PNWNAmet, NRCANmet, Daymet, EMDNA) and four reanalysis-based sources (ERA5-Land, CaSRv21, CaSRv31, CHELSA-W5E5). Each of these is evaluated by comparing 1981–2010 climatologies of seasonal average maximum and minimum temperatures and total precipitation to station observations and BC PRISM climatologies. The analysis identifies cold and dry biases for both CaSR versions, cold and wet biases in ERA5-Land, widespread dry biases in NRCANmet, and warm biases in EMDNA and CHELSA-W5E5 maximum temperatures. Aside from these broader patterns, the datasets display regionally specific differences, particularly for Daymet, EMDNA, and CHELSA-W5E5. Results from the comparisons are summarized to identify which of the datasets should be favoured for different variables or regions. These results highlight points of disagreement between the datasets for BC that can inform choices for dataset selection and identify areas for potential improvement.
Abstract Global climate models (GCMs) still show limited skill in reproducing near-surface wind speed (NSWS) over mainland China, especially in regions with complex terrain. In this study, we applied an encoder–decoder super-resolution convolutional neural network as a statistical post-processing framework to reconstruct daily NSWS from 14 CMIP6 models to a finer 0.25° grid, using CN05.1 as the reference dataset. The model was trained on the historical period and evaluated over 2000–2014, and its performance was compared with the raw CMIP6 ensemble and NEX-GDDP-CMIP6. The downscaled ensemble substantially improved the simulation of NSWS over mainland China. During 2000–2014, it achieved a regional mean bias of −0.07 m·s −1 and a spatial correlation coefficient of 0.93 for annual mean NSWS, outperforming both CMIP6 and NEX-GDDP-CMIP6. It also better reproduced seasonal patterns, interannual variability, and the spatial distribution of recent changes. Applied to future SSP-RCP scenarios, the downscaled projections indicate a weak but statistically significant decline in area-mean NSWS over the 21st century, with larger decreases under stronger forcing. However, the projected China-wide mean during the next two decades remains close to the 1980–2014 baseline, while localized increases persist in parts of eastern and northeastern China. These results show that super-resolution CNN methods can provide useful post-processing for CMIP6 NSWS over China, although future work should assess the added value of static predictors such as topography and land cover and further evaluate extreme wind behavior.
Abstract This study investigates the physical mechanisms driving the systematic positive bias in surface solar irradiance (SSI) within the high-resolution contiguous United States for 40 years at 4-km grid spacing (CONUS404) simulation, using the satellite-based National Solar Radiation Database (NSRDB) as a reference. Analysis of hourly data under mutually clear-sky conditions for 2018 reveals that the average SSI bias decreases by 99% (from 41.8 to 0.5 W m −2 ), indicating that irradiance overestimation is primarily attributable to cloud-related processes. To examine these discrepancies without relying on complex, multilayer atmospheric profiles, a quantitative relationship was established to derive the model’s equivalent cloud optical thickness (ECOT) solely from surface SSI variables. This framework enables a consistent, gridpoint-level classification of cloud conditions into mutually exclusive categories: optically thin, medium, and thick. A key finding is that regional SSI overestimation is driven by intercategory misclassification and a resulting deficit of optically medium clouds. Rather than omitting these clouds as false clears, the simulation systematically misclassifies them as optically thin, allowing excessive solar radiation to reach the surface. This misclassification pathway displays clear geographic dependence. In cloud-prone regions like the mid-Atlantic and Pacific Northwest, this occurs in 8.6% and 9.3% of total observations, respectively, leading to notable positive SSI biases. Conversely, in relatively clear regions like California, a high baseline frequency of clear-sky conditions buffers the radiation bias. Furthermore, in the Texas Gulf, a lower net SSI bias masks underlying classification discrepancies through an offsetting error mechanism: Solar radiation overestimation is structurally countered by overpredicting thick convective systems associated with transient tropical cyclones. Significance Statement Surface solar irradiance (SSI) is heavily regulated by atmospheric cloud properties, transitioning from minor attenuation under thin cloud layers to widespread scattering and minimal surface energy transmission beneath thick, precipitating systems. The contiguous United States for 40 years at 4-km grid spacing (CONUS404) dataset provides a high-resolution (4 km), multidecadal regional climate reanalysis over the CONUS. Evaluation of this dataset reveals a widespread positive bias in surface solar irradiance relative to the satellite-based National Solar Radiation Database (NSRDB) reference. This discrepancy is primarily driven by an underrepresentation of nonprecipitating, optically medium clouds within the simulation configuration. Characterizing this systematic cloud-layer deficit establishes a key diagnostic reference for dataset users, provides guidance for future regional model developments, and supports the evaluation of simulated cloud–radiation interactions.
Abstract Flash droughts, defined by their rapid onset and potential for severe agricultural impacts, remain understudied in humid regions like the southeastern United States. This study presents a multi-indicator assessment of flash drought climatology from 2001 to 2023 across the Southeast using four metrics: the U.S. Drought Monitor (DM), evaporative stress index (ESI), soil moisture volatility index (SMVI), and the lawn and garden index (LGI). Flash droughts most commonly emerged during early summer and autumn, aligning with seasonal transitions in evapotranspiration and precipitation. SMVI identified the most events (mean = 24), while DM was the least sensitive (mean = 4). Regional hotspots, including central Georgia and eastern North Carolina, were consistently detected across indicators. To assess agricultural vulnerability, the study integrates crop-specific critical growth stages with flash drought detections using conditional probabilities. Corn showed the highest exposure, with 21% of years when a flash drought occurred during the crop’s most moisture critical stage, while DM-defined events aligned with critical stages in over 40% of cases. These results demonstrate that vulnerability is shaped not only by drought frequency but also by timing relative to crop development. The findings underscore the importance of multimetric approaches and regionally tailored frameworks to inform climate-resilient agricultural strategies in the Southeast. Significance Statement Flash droughts—rapid-onset droughts with severe impacts—occur nearly every year across the southeastern United States, yet their agricultural relevance is often overlooked. Using four widely used detection metrics, this study identifies a distinct flash drought climatology with peaks in early summer and fall. Timing and duration are key to understanding impacts. We show where and when flash droughts are most likely to harm crops in the Southeast by identifying how often they occur during key stages of crop growth—times when even short dry spells can seriously affect yields. By linking flash drought patterns to crop development, this work supports improved monitoring and informs adaptation strategies tailored to southeastern agriculture.
Abstract Southwestern warm–moist air intrusions exert strong control on boundary layer temperature and humidity over East China and play an important role in regional weather and precipitation processes. However, their multiscale structures and model-dependent variability remain insufficiently understood. This study investigates an intrusion event on 13–15 January 2024 using Raman lidar, radiosondes, in situ observations, ERA5 reanalysis, Weather and Research Forecasting (WRF) Model mesoscale simulations, and parallelized large-eddy simulation model (PALM) large-eddy simulations (LESs) at grid spacings from 28 km to 40 m. All models reproduce the general evolution of the event but simulate the moist-air arrival approximately 1 h later than observed by lidar. PALM resolves much finer vertical and horizontal structures, capturing water vapor standard deviation up to 0.58 g kg −1 and potential temperature standard deviation up to 0.24 K. These fine structures reflect turbulence-driven mixing and moisture redistribution within the boundary layer, which are often smoothed out in mesoscale models. Horizontal variability is strongly modulated by stability: Large fluctuations occur under inversion layers, while mixed-layer turbulence produces near-homogeneous distributions. Spectral analysis shows that only PALM explicitly resolves the inertial subrange and turbulence-driven moisture filaments, whereas WRF lacks subkilometer variability. A lidar-constrained PALM simulation significantly reduces the moist-air timing bias (from 1 h to nearly 0 h) and improves ground-level moisture representation during daytime when convective mixing efficiently transports assimilated moisture signal downward. These findings demonstrate the importance of high-resolution observations and LES for representing moisture transport and boundary layer adjustment during warm–moist air intrusions. The results also highlight key deficiencies in mesoscale parameterizations and provide guidance for improving humidity and temperature predictions in operational models such as WRF. Significance Statement Warm–moist air intrusions strongly influence wintertime weather over East Asia, yet their fine-scale structure and boundary layer impacts remain poorly represented in models. By combining Raman lidar observations with multiresolution large-eddy simulations, this study shows how turbulence and model grid spacing control the timing and intensity of moistening and inversion erosion. The results demonstrate that high-resolution LES is essential for capturing key boundary layer processes and for improving humidity and temperature forecasts during synoptic intrusion events.
Abstract Weather Research and Forecasting (WRF) Model simulations at a “gray-zone” horizontal grid spacing of Δ x = 200 m are compared with an upscaled large-eddy simulation (LES) with the FastEddy model and observations over coastal central California to understand and characterize high-resolution turbulent flows for offshore wind resource assessment. A case study of a strong coastal low-level jet is examined, resulting from the typical spring and summertime conditions of a broad subtropical ridge in the northeastern Pacific Ocean. The WRF Model simulations use three different planetary boundary layer (PBL) parameterizations: two one-dimensional (1D) parameterizations and a three-dimensional (3D) parameterization, ideally suited for simulating flows at gray-zone horizontal resolutions. Similarities and differences between the WRF Model simulations and upscaled LES are assessed. Overall, the WRF Model simulations compare reasonably well with the upscaled LES in simulating the coastal flows, and both the upscaled LES and WRF Model simulations compare favorably with observations. The WRF 3D PBL simulation compares most favorably with the upscaled LES, in particular in simulating distributions of vertical velocity over the ocean and land and in capturing energetic small-scale structures, due to its more realistic treatment of horizontal mixing. Significance Statement The coastal environment in central California is complex, with a stable marine boundary layer interacting with a boundary layer over steep, heterogeneous terrain, which can be rapidly destabilized with daytime heating. It is not well observed or simulated with current computer models that use grid spacings on the order of 1 km. Motivated by the need to understand the small-scale heterogeneous flow features relevant to the evaluation of offshore wind resources in this region, a detailed computer modeling study is carried out. A computer model simulation that explicitly resolves turbulence from large eddies is used as a reference in which to compare simulations that do not or only partially resolve this turbulence. The main finding is that a computer model simulation with more realistic treatment of horizontal mixing most closely matches the simulation that explicitly resolves turbulence, when compared at equivalent resolutions.
Abstract Lake Victoria (LV), the world’s largest tropical lake, significantly influences East Africa’s regional climate through diurnally varying lake–land-breeze circulations. In particular, lake-breeze fronts (LBFs) often trigger cloudiness, deep moist convection, and heavy precipitation, leading to weather-related hazards. To systematically study LBFs over LV, we developed an observation-based lake-breeze detection algorithm (OLBDA), which objectively identifies LBF passages using 15-min data from 45 automatic weather stations across Uganda’s extended coastal region over 6 years (2017–22). Focusing on daytime (0900–1900 LT) during dry months (December–February and June–August), the algorithm applies thresholds based on the lowest (highest) 30% of temperature drops (dewpoint and wind speed increases), enabling detection of subtle meteorological changes associated with LBF passage. Validation against visible satellite imagery shows strong performance with high accuracy (ACC = 0.76), detection rate [probability of detection (POD) > 0.7], and low false alarms [probability of false alarm (POFA) = 0.2]. Application of the OLBDA revealed that LBFs predominantly occur from early afternoon to evening, peaking at 1300 LT near the coast and shifting inland with distance. LBFs were observed on 48% of days, with December–February and January having the highest seasonal and monthly frequencies, respectively. LBF propagation speed varied in relation to the background wind in different hinterland sectors of LV, with the fastest movement westward at 8 m s −1 and the slowest northward at 5.8 m s −1 . LBFs penetrate up to ∼130 km inland. The OLBDA offers a robust framework for analyzing LBFs, providing valuable insights for disaster preparedness and climate mitigation in the Lake Victoria basin.
Abstract The response of Earth’s climate system to greenhouse gas (GHG) forcing appears largest in the Arctic, both in observations and models. However, in multimodel evaluations (CMIP3 through CMIP6), most models appear to underestimate the amount of warming. Here, we examine the role of turbulent parameterizations of the stable boundary layer (SBL) in impacting the magnitude of warming and the spread in model responses. Turbulent parameterization of the SBL has long been a challenge in models. Forms used in fine vertical grid research boundary layer models often do not perform well in coarser-grid operational or climate models. Global climate model (GCM) vertical grids often deteriorate to a few layers in the SBL and 500 m or more in deeper stable layers (SLs). This grid spacing may not capture well the correct strength/depth of the Arctic inversion, the entrainment warming as the SBL is destabilized by GHG forcing or the correct energy budgets. It seems likely that the large spread in GCM Arctic simulations is in part due to the differences and misapplications of these grid-dependent parameterizations. In this investigation, we propose a new approach to the SBL parameterization problem by explicitly incorporating the grid vertical spacing in the parameterization. This will be carried out by analytically recovering a stability correction function that depends on the grid spacing. Initial tests of an analytically recovered correction function indicate that the correction function provides longer-tailed stability functions for coarse-grid models. Applying this correction to a simplified Arctic profile made coarser-grid models agree better with fine-scale results. Significance Statement This article addresses how small-scale turbulence may impact large-scale climate change response in coarse-grid climate models. The impact of climate change in models depends on the vertical grid spacing. The present investigation proposes and tests a model correction that makes model results less grid dependent. It may help bring climate and small-scale boundary layer modeling communities together.
Abstract Supercells may contain a dominant mesoanticyclone and produce substantial severe weather. In this study we examine the spatial and temporal characteristics of 885 manually identified and quality-controlled left-moving (LM) supercell storms and present fundamental characteristics of the severe weather reports produced by these storms, including comparisons with a recent, large algorithm-derived dataset. Findings indicate that LM supercells are most common on the central and southern Great Plains, though their occurrence centroid shifts from the Great Plains in the warm season (June–September) to the Southeast in the cool season (November–April). Long-tracked LM supercells are most common in the Southeast and during December-January-February, and storm speed correlates strongly with the mean 0–6 km wind speed. Storms follow a diurnal distribution with most storms occurring from 21–00 UTC. The proportion of storms associated with at least one severe report increases sharply with mesoanticyclone intensity, and significantly severe hail reports are strongly associated with strong mesoanticyclones. Faster-moving LM storms do not produce a greater proportion of severe reports, while longer-tracked storms produce a larger proportion of severe reports. The likelihood of an LM storm producing a severe report does not depend on time of day.
Abstract Accurate road surface temperature forecasts are essential for planning road salting and keeping roads free of ice. However, these forecasts often contain systematic errors due to various factors. This study presents a postprocessing methodology aimed at improving forecast accuracy, especially for temperatures near 0°C, where even small errors can be critical for road safety. The methodology was tested to correct systematic errors in the Finnish Meteorological Institute’s road surface temperature forecasts. An extreme gradient boosting (XGBoost) model was trained using data from several years and approximately 400 road weather stations to learn patterns between forecasted weather variables and road surface temperature errors. A distinctive approach in this study was to apply sample-specific weights during model training, which gave more emphasis to cases where the surface temperature was near 0°C. Another XGBoost model without weights was trained to perform well across all temperatures. The models were evaluated using data from September 2024 to May 2025. Based on scores calculated over the entire period, the unweighted model reduced the mean absolute error (MAE) by 3%–23% compared to original road weather model results, depending on the forecast lead time that extended up to 62 h. For cases where the observed surface temperature was between −3° and 3°C and the lead time was 12 h or less, the weighted model achieved a MAE reduction of 17%–34%. In conclusion, both models performed well, with the weighted model showing particular effectiveness in improving forecast accuracy at near 0° temperatures. Significance Statement This study aims to improve the accuracy of road surface temperature forecasts using machine learning. Accurate forecasts are important because they help in timing road salting operations effectively, keeping roads free of ice and preventing unnecessary salting. The machine learning model developed for forecasting near 0°C temperatures was able to reduce forecast errors by 17%–34% in that temperature range. These models have strong potential to enhance future road weather forecasts in Finland, and the same methodology can be applied in other countries as well.
Abstract The assimilation of land surface temperature (LST) data from geostationary satellites into land surface models is a promising avenue for improving weather and climate predictions, yet it has been met with mixed success. Geostationary satellite sensors support frequent measurement of land surface temperature during cloud-free periods, potentially providing multiple measurements over the course of a single day capturing a temperature diurnal range, something not available from low-Earth-orbiting sensors. This study provides a comprehensive comparison of geostationary LST from the International Satellite Cloud Climatology Project (ISCCP) H-series dataset with in situ soil temperature measurements at 5- and 10-cm depths from the International Soil Moisture Network (ISMN) across the United States from 2010 to 2015. The analysis, stratified by land-cover type (cropland, grassland, shrubland) and season, reveals a strong correlation between LST and in situ soil temperatures, particularly during nighttime hours where the correlation coefficient often exceeds 0.9. However, a significant and systematic diurnal bias was identified, with LST being colder than soil temperature at night (bias up to −10 K) and substantially warmer during the day (bias up to +20 K). While this diurnal pattern is consistent across different land-cover types and seasons, the magnitude of the bias presents a major challenge for current data assimilation systems. These findings highlight both the potential of high-frequency geostationary LST data and the critical need for advanced assimilation techniques that can account for predictable, large-magnitude diurnal biases to unlock the full value of these observations. Significance Statement Integrating observations from satellites into Earth system models offers the potential to increase model accuracy and relevancy if the observations are well correlated and observationally relevant to the model parameter being updated. This study shows that International Satellite Cloud Climatology Project land surface temperature estimates are well correlated with in situ 5- and 10-cm soil temperature observations from the International Soil Moisture Network with correlations often exceeding 0.9; however, the bias between them is considerable, shifting from strongly negative to strongly positive over the course of a day. Though well correlated, the strong diurnal pattern and high biases increase the assimilation complexity when assimilating the observations into a land data assimilation system.
Abstract Cold fronts play an important role in influencing climate and weather patterns across eastern Colorado, a region positioned between the U.S. Great Plains and the Rocky Mountains. In particular, cold fronts substantially contribute to cool-season precipitation and often accompany hazardous weather events, such as extreme winds, rapid temperature fluctuations, and dangerous fire weather conditions. This study identifies and examines a subset of strong cold frontal passages in eastern Colorado between September and May 1950–2022. Using ERA5, we identify 723 strong cold front events based on both the magnitude and spatial extent of postfrontal temperature changes across eastern Colorado. Front locations are detected using a thermal front parameter, which identifies strong gradients in equivalent potential temperature that coincide with cold-air advection. Cold front events are further categorized into low-precipitation (LP) and high-precipitation (HP) fronts to explore dynamical and thermodynamic processes that influence precipitation variability during frontal passages. Our findings indicate that cold fronts are most frequent during the transition seasons, often occur overnight, and produce the largest daily maximum temperature decreases across far eastern Colorado, away from significant topography. Additionally, cold front frequency decreases during stronger El Niño years and phases 1–2 of the Madden–Julian oscillation (MJO) and increases during MJO phase 7. Composite analyses of LP and HP fronts reveal distinct dynamical and thermodynamic patterns associated with these events, providing valuable insights into their development and attendant surface weather impacts. Significance Statement Cold fronts are a frequent weather feature across eastern Colorado during the cool season and are often linked with impactful weather, including sharp temperature drops and precipitation. This study identifies significant cold frontal passages based on the magnitude and spatial extent of their attendant temperature changes and further categorizes them by how much precipitation they produce. Our findings show that stronger and more frequent cold fronts tend to occur in far eastern Colorado and during the overnight hours. The fronts that produce the most precipitation are typically deeper, characterized by stronger upward motion, and occur within moist environments. These insights can help operational forecasters better anticipate strong frontal passages, improve hazard communication, and more accurately predict precipitation types at a regional level in eastern Colorado.
Abstract Lake evaporation, as a key hydrological process on the Tibetan Plateau (TP), is highly uncertain in estimation and lacks a systematic evaluation. Taking Siling Co, the largest lake in Tibet, as an example, this study comprehensively evaluated 30 lake evaporation models from five groups (Combination, Solar radiation–temperature, Dalton, Temperature–daylength, Temperature) based on eddy covariance observations to determine their applicability and rank them. Results show that the Dalton group models outperformed others, with the mass transfer (4) model achieving the lowest root-mean-square deviation (RMSD) (0.16 mm day −1 ) and highest Nash–Sutcliffe efficiency (NSE) (0.71). The Combination group models, despite their theoretical robustness, significantly overestimated evaporation (e.g., Penman–Brutsaert model overestimated by 76.7%) when the lake heat storage term G was calculated from net radiation. However, when G was replaced with observed values, Combination models improved drastically (RMSD < 1.0 mm day −1 , NSE > 0.86). We propose a data-driven selection framework 1) with water temperature profiles ( G available), using Combination models; 2) with only wind/humidity, using Dalton models; and 3) Temperature/Solar radiation–temperature models should be used with caution in deep TP lakes. Although these findings are derived solely from observations during the 2014 open-water season at Siling Co, they still provide valuable insights for evaporation estimation in large, deep lakes across the TP and other sparsely instrumented high-elevation regions.
Abstract Sea fog, as a significant marine meteorological hazard, has a pronounced impact on maritime transportation and port operations. This study employs Himawari-9 satellite data and multichannel Raman polarization lidar (MCRPL) to analyze the causes and evolution characteristics of a typical advection fog event occurring off Qingdao from 22 to 25 May 2024. The results indicate that this sea fog formed as advection fog through rapid lower-level condensation when a warm, moist air mass passed over the “cold pool” region in the central-western Yellow Sea under southerly wind conditions. A strong temperature inversion layer, high relative humidity (RH), and low wind speeds were key factors sustaining the fog over an extended period. From the initial to mature stages, the fog layer’s brightness temperature difference (BTD) decreased from −4.6 to −2.4 to −5.7 to −2.9 K, while its thickness increased from 50 to 150 m to over 300 m, locally reaching 400 m. During the mature stage, the fog top reached 150–210 m. Total backscatter intensity (TBI) at 355 and 1064 nm exhibited saturated high values, while the volume depolarization ratio (VDR) at the fog base decreased to below 0.1. High-value zones with a color ratio (CR) of 0.50 appeared. This study comprehensively characterizes the three-dimensional structural features of sea fog, deepening the understanding of the formation and evolution mechanisms of advection fog. Significance Statement This study aims to elucidate how a dense fog formed and persisted near the coast of Qingdao. Results indicate that warm, moist air cooled over colder sea surfaces under southerly winds, leading to fog formation. Low wind speeds and a stable atmospheric structure allowed the fog to persist for an extended period, during which it alternated between thinning and thickening. Understanding this process aids in early prediction of low visibility risks at sea, thereby enhancing shipping and port safety. Future applications of these observational methods can extend to monitoring and early warning systems for additional marine regions and weather phenomena.
Abstract The C vector, which is the extension of the Q vector in the three-dimensional space, restores the information of the ageostrophic motion lost by the latter. A comprehensive system has been established for the Q-vector diagnosis, but few studies have focused on the C vector. The vertical component of the C vector has dual physical interpretations: It describes both the rotation of the ageostrophic wind and the horizontal geopotential gradient. Based on the quasigeostrophic (QG) approximation and the vertical C-vector component in the p -coordinate system ( C p ), the C p tendency equation describing its local variations is derived. This equation has also dual physical interpretations, describing the local variation of both the ageostrophic vertical relative vorticity and the shape of the two-dimensional geopotential surface. The advection of C p , deformation, and the vorticity–divergence interactions determine the C p tendency. Based on the fifth generation European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis (ERA5) data and the objective cyclone identification method, the C p tendency equation is applied to diagnose the evolution of a Huang–Huai cyclone, which occurred during the extreme precipitation over Henan in July 2021. Compared with the commonly used vertical relative vorticity, C p has a better performance in both tracking the cyclone center and featuring its intensity variation. The results show that the C p tendency plays a major role in the evolution of the cyclone: Its distribution and variations drive the cyclone’s motion and predict that the cyclone will strengthen (decay), respectively. The deformation and the vorticity–divergence interactions collectively lead to the distributions and variations of the C p tendency, and the advection of C p can be neglected. Significance Statement The C vector is an improvement over the Q vector, with its vertical C-vector component ( C p ) depicting the barotropic dynamics of the ageostrophic motion that is lost by the Q vector. Nevertheless, few researchers have focused on the C vector. We developed the C-vector diagnosis by deriving the C p tendency equation. Theoretical discussions show that the equation contains complete information about the quasigeostrophic (QG) kinematics, and a case study proves its usefulness in describing the evolution of the synoptic-scale cyclone. This study can provide a new look at the application and development of the C-vector diagnosis.
Abstract Weather-based decision-support tools for in-orchard management are derived from open-field weather station data with an assumption that there is minimal to no difference between open-field and in-orchard weather. Orchard training system, growth stage, and various management practices can cause different weather conditions inside the orchard, that may lead to bias and uncertainty in the weather-based models. This study quantified orchard training and management effects (e.g., irrigation and overhead cooling for ameliorating heat stress to maturing fruits and canopy) on air temperature, relative humidity, solar radiation, and wind speeds at different time scales using two seasons data from six commercial apple orchards. A paired t test revealed significant differences ( p < 0.05) between the seasonal means of the open-field and in-orchard weather, indicating substantial orchard effects. Typically, orchards have 1.9°–4.4°C daily maximum cooler air temperature and 9.2%–27.5% higher relative humidity due to the evident impacts of canopy transpiration. Also, in-orchard stations recorded lower solar radiation (267.8–483.2 W m −2 ) and wind speed (2.2–3.7 m s −1 ). Monthly averages revealed the dependence of orchard effects on the phenological stages of apple canopies. Wind resistance and reduced air mixing caused drier microclimate during winter [relative humidity (RH) offset: 8.7%] and spring (RH offset: 7.7%) seasons. The use of overhead sprinklers for heat stress mitigation reduced air temperature (5.2°C) and increased relative humidity (up to 31%) inside the orchard, and the effects lingered during evening and night hours. Significance Statement The purpose of this study is to quantify the in-orchard microclimatic differences compared to an open-field weather station during the entire year. The study revealed that modern orchards have 1.9°–4.4°C cooler daily maximum air temperature and 9.2%–27.5% higher relative humidity. These offsets provide growers with a benchmark on using open-field weather data meaningfully to decide up on timely management actions without needing to install private weather station inside the orchard. Further research should focus on (i) correcting or redoing the weather-based models to incorporate orchard microclimates and (ii) correcting the open-field weather forecasts to in-orchard specific weather.
