Abstract:Previous studies based on observation stations have not fully revealed the spatiotemporal characteristics of PM2.5 concentrations.This paper evaluates the accuracy of PM2.5 remote sensing inversion data in the Pearl River Delta(PRD)urban agglomeration and further explores the spatiotemporal patterns and influencing factors of annual,seasonal,and monthly average PM2.5 concentrations based on the remote sensing inversion data.Validation from 2015 to 2022 confirms the high accuracy of the remote sensing data(R2:0.964-0.998).Analysis based on the inversion data from 2000 to 2022 reveals a"rise-stagnation-decline"trend in annual average PM2.5 concentrations,with key turning points in 2004 and 2013(R2=0.943 6).The primary influencing factors are government policies.Spatially,PM2.5 concentrations exhibit a distinct"high in the center,low in the periphery"pattern.The Guangzhou-Foshan border forms the core pollution zone for PM2.5,while the regions with the lowest pollution levels are identified in Zhuhai,southeastern Huizhou,and southern Jiangmen.Seasonally,PM2.5 concentrations follow a"high in autumn and winter and low in spring and summer"pattern.Taking the high pollution period(2003-2014)as an example,monthly average concentrations of PM2.5 showed a trend of"falling first and then rising"throughout the year,with the highest in January(62.55 μg·m-3)and the lowest in July(22.53 μg·m-3),with precipitation being the main influencing factor.Spatially,from October to January,the area of high PM2.5 concentrations expands southward to encompass Jiangmen and Zhuhai,with significant regional variations.The findings of this study provide theoretical basis for the joint prevention and control of air pollution in urban agglomerations.

Abstract:Typhoons are natural disasters that severely impact coastal areas,and accurately monitoring their intensity is crucial for disaster prevention and mitigation.Combining satellite observations with deep learning technology has become a pomising new method for typhoon intensity monitoring.However,the accuracy of different deep learning methods remains unclear.This study evaluates the performance of six convolutional neural network models based on deep learning in monitoring typhoon intensity in the Western Pacific.Using Himawari-8/9 satellite cloud product data and the China Meteorological Administration's best track data from 2015 to 2024,we analyzed the computational effectiveness of LeNet-5,AlexNet,DenseNet-121,VGG-16,ResNet-50,and GoogleNet-InceptionV3 models.This study not only explores the applicability and performance of these models in different scenarios but also visualizes the feature extraction steps of the models to clarify the differences between them and their working principles.LeNet-5 and AlexNet show the largest biases in extremely weak(TD)and extremely strong(SuperTY)categories;DenseNet-121 maintains relatively uniform bias distribution across all intensity levels;ResNet-50,VGG-16,and GoogleNet-InceptionV3 demonstrate stable performance in medium intensity ranges(TS,STS,TY,STY).Overall,the GoogleNet-InceptionV3 model achieves the highest accuracy with an R² of 0.89,while ResNet-50,with an R²of 0.87,offers faster computational speed.

Abstract:In order to improve the accuracy of air quality models for predicting O3 and PM2.5 concentrations in Guangzhou,a LightGBM algorithm was employed to establish correction models for O3 and PM2.5 concentration forecasts in three time periods of 0-24 h,24-48 h,and 48-72 h,using the observations from Guangzhou national control stations and hourly forecast data.The SHAP method was then applied to interpret and analyze the correction models.The results show that the LightGBM correction model can significantly improve the O3 and PM2.5 concentration forecasts for each station and forecast time period.The error of each station significantly reduced and its distribution is more consistent after correction.As the forecast time period increases,the effect of the correction models on forecast improvement slightly decreases.Overall correction effects across all time period showed the root mean square error(RMSE),average absolute deviation(AAD),and correlation coefficient of O3 concentration from the model and observation changed from 39.5 μg·m-3,30.3 μg·m-3,and 0.61 to 23.3 μg·m-3,16.9 μg·m-3,and 0.86,respectively.For PM₂.₅ concentrations,the RMSE,AAD,and correlation coefficient changed from 18.3 μg·m-3,12.7 μg·m-3,and 0.26 to 9.7 μg·m-3,6.9 μg·m-3,and 0.76,respectively.Key features among different correction models showed little variation.For the O3 concentration correction model,the most important features mainly include O3 concentration,temperature,relative humidity,NO2,shortwave radiation,wind direction and speed.For the PM2.5 concentration correction model,the most important features mainly include air pressure,air quality index,temperature,relative humidity,air pressure,PM10,O3,wind direction and speed.The influence of each feature on the correction models conforms to the generation and accumulation mechanism of O3 and PM2.5.The case study further proves the effectiveness and practicality of the correction model.

Abstract:Based on new-type observational data from dual-polarization radars with different bands and wind profile radars in Guangzhou,this study investigates the environmental conditions and storm structural characteristics of the deadly supercell tornado that occurred in Baiyun District,Guangzhou,on April 27,2024.This tornado event deveoped under a persistently strong southwesterly wind on the southern side of the Tibetan Plateau and the influence of the Western Pacific Subtropical High.The atmospheric stratification profile exhibited a"trumpet-shaped"profile,characterized by dry upper levels and moist lower levels.The lifted condensation level(LCL)was remarkably low in the morning(approximately 180 m),accompanied by abundant energy(Convective Available Potential Energy(CAPE)of approximately 2500 J·kg-1 and Convective Inhibition(CIN)of 0 J·kg-1).Additionally,strong 0-3 km wind shear(17 m·s-1)further indicated favorable thermodynamic and dynamic conditions for the development of a severe tornado.The parent supercell responsible for the tornado exhibited a long lifecycle and underwent multiple intensification periods.The tornado occurred during a relatively quiescent period between two intensity peaks,posing significant challenges for forecasting and early warning.The parent storm displayed distinct hook echo and mesocyclone characteristics,with a"descending-type"developmental pattern.Due to minimal influence from surface friction,the storm developed more vigorously.The downdraft of the parent storm formed a cold pool,with the strongest center temperature dropping to 22℃and a radius of approximately 30 km.A gust front at the leading edge of this cold pool enhanced low-level convergence and uplift,intensifying the mesocyclone as it approached the ground.Prior to the tornado,the vertical vorticity of the mesocyclone was 0.026 s-1,while the vertical vorticity of the Tornado Vortex Signature(TVS)reached 0.46 s-1.Furthermore,compared to the CINRAD/SA-D,the XPAR-D captured clearer tornado signatures,such as the TVS and the Tornado Debris Signature(TDS),as well as more detailed evolution of the low-elevation hook echo.In operational applications,these radars demonstrate complementary advantages,enabling more precise characterization of the development of local severe convective storms and providing crucial technical support for nowcasting and early warning.

Abstract:To address the issues of insufficient accuracy and poor timeliness in traditional wind field prediction methods,this study introduced the Informer model to enhance the forecast accuracy of the long-term meteorological data at Xiamen Gaoqi International Airport.The paper details the unique advantages of the Informer model in handling wind field time series data,including its probabilistic sparse self-attention mechanism and self-attention distillation technology.These features enable the model to efficiently capture long-term dependencies and complex characteristics within the data.Compared with traditional Artificial Neural Networks(ANNs)and Long Short-Term Memory(LSTM)models,the Informer model demonstrates higher prediction accuracy across different time scales.In the 60-minute predictions and seasonal variations,the Informer model demonstrated high robustness and efficiency.Additionally,a comparison of the effects of different wind field variations on the model's wind field predictions revealed that the Informer model consistently maintained stable predictive performance under varying wind field conditions,further validating its broad applicability and robustness.By enhancing prediction accuracy and timeliness,this research not only provides more accurate wind speed and direction forecasts for aviation meteorological services,aiding in flight safety,optimizing flight scheduling,and improving energy efficiency,but also has a positive impact on short-term weather forecasting and offers new research ideas and solutions.It has significant implications for advancing the application of deep learning in meteorological forecasting.

Abstract:Based on daily temperature,precipitation,wind speed and relative humidity data from national meteorological observation stations from 1990 to 2020,this study analyzes the temporal and spatial characteristics of meteorological environment in the Pear River Delta(PRD).The effects and contributions of urban expansion to meteorological elements variations were quantitatively analyzed based on land use classification data in the PRD.The results showed that from 1990 to 2020,urban expansion was significant in the PRD,with the proportion of urban construction land area increasing from 5.71%to 14.99%.Regional mean annual temperature and annual precipitation showed an increasing trend,while annual strong wind days,annual low wind days,and annual relative humidity showed a weak declining trend.Urban expansion led to the increase of mean annual temperature,with a change rate of 0.05℃·(10 a)-1 and a contribution rate of 20.80%.Urban expansion led to the increase of annual precipitation,with a change rate of 28.80 mm·(10 a)-1 and a contribution rate of 46.24%.Urban expansion led to the increase of annual strong wind days,with a change rate of 1.16 d·(10 a)-1 and a contribution rate of 100%.Urban expansion led to the decrease of annual relative humidity,with a change rate of-0.84%·(10 a)-1 and a contribution rate of-100%.Urban expansion had no obvious effect on annual low wind days.Seasonally,the contribution of urban expansion was the highest for mean temperature in winter.Its contribution to precipitation in summer and autumn was higher than that in spring and winter.Contribution to strong wind days were relatively higher in spring,autumn and winter,and the contributions to relative humidity were high across all four seasons,and the contribution to low wind days was highest in autumn.

Abstract:Based on a machine learning(ML)algorithm,LightGBM,an optimized forecast model of visibility was established to correct the numerical weather prediction(NWP)at Shanghai stations.The model was trained based on the historical station observations(2019-2023)and hourly NWP output data from the numerical weather forecasting(CMA-SH9)and integrated weather-air quality forecasting(WARMS-CMAQ)models.In order to alleviate the problem of extremely unbalanced observational samples and improve the prediction skill for fog and other low-visibility events,visibility was classified into different levels,with the ML task framed as a classification problem.The influence of low-visibility samples was emphasized through data pre-cleaning and differentiated weight coefficients for different grades.Finally,the recall rate,precision,and comprehensive TS scores of the first two grades(≤1 km and 1-3 km)were used as evaluation criteria.The evaluation of the test dataset and subsequent independent operational phase show that the ML-based model significantly improves visibility forecasting skills compared to numerical models.In particular,the hit rate for low visibility events(≤1 km)increased from approximately 20%to nearly 60%,and the TS score improved to 0.3.In addition,the analysis of typical cases since December 2023 shows that the LGBM model performs good in forecasting heavy fogs,with better agreement with observations in terms of fog onset and dissipation.It's indicated that this ML-based model obviously alleviates the serious underprediction of low-visibility events(especially for fog)in the original numerical model,which proves the algorithm's feasibility and superiority.

Abstract:The"7.20"extreme rainfall event in Henan Province was characterized by prolonged duration and intense precipitation,causing disastrous damage.To investigate the key sources and transport pathways of water vapor during this extreme rainfall event,this study utilized multi-source meteorological data and the FLEXPART model to conduct trajectory simulations and analyze the water vapor transport processes.Additionally,cluster analysis was performed on the transport trajectories.The results revealed two primary water vapor pathways that significantly contributed moisture to the extreme precipitation event.The first pathway,originating from the Northwest Pacific typhoon"In-Fa",provided continuous water vapor transport toward Henan between 1 000 hPa and 700 hPa,with the core between 1 000 hPa and 900 hPa during July 19-21.The second pathway,from the South China Sea typhoon"Cempaka",transported water vapor on July 20,primarily between 950 hPa and 800 hPa.Topographic effects resulted in significant height differences between the two water vapor transport channels.The continuous accumulation of water vapor over Henan Province was an important factor in the formation of the extreme precipitation.The persistent transport of large amounts of vapor by the typhoons played a crucial role in sustaining the heavy rainfall,significantly contributing to the extreme precipitation event.

Abstract:To address the increasing challenge of persistent pollution and rising ozone(O₃)levels in Guangzhou,this study developed advanced air quality forecasting models using machine learning techniques.Based on environmental monitoring and meteorological observation data,the Liang-Kleeman information flow was used to conduct causal analysis on factors affecting the concentrations of atmospheric pollutants such as CO,NO2,O3,PM2.5,PM10,and SO2.Using the Random Forest(RF),Extreme Gradient Boosting(XGBoost),and Long Short-Term Memory Neural Network(LSTM)algorithms for integrated modeling,five distinct pollutant concentration forecasting models(RF,XG,LSTM,and the integrated models MIX1 and MIX2)were constructed.These models forecast pollutant concentrations,which were then used to calculate the Air Quality Index(AQI)and identify the primary pollutant.The results show that the integrated models(MIX1 and MIX2)generally outperform the single ones(RF,XGBoost,and LSTM models).For pollutant concentration forecasting,the MIX1 model was optimal for CO,NO2,and O3,while the MIX2 model performed best for PM10,PM2.5,and SO2.For air quality forecasting,the MIX2 model was superior for 1-2 day forecasts,whereas the MIX1 model was optimal for 3-7 day forecasts.The accuracy rates for primary pollutant by the MIX1 and MIX2 models for 1-7 day forecast were 71.26%-83.33%and 73.71%-81.11%,respectively.The models showed high reliability,with accuracy rates for primary pollutant identification ranging from 71.26%to 83.33%for MIX1 and 73.71%to 81.11%for MIX2 across the 1-7 day forecast,providing valuable tools for environmental authorities to implement targeted air pollution control measures.

Abstract:In order to study the dynamic and microphysical characteristics of wind and hail events in Zhejiang using phased array dual polarization radar,an analysis of the wind and hail event in Zhejiang on 10 July 2023 was conducted based on X-band phased array radar observation data,wind field reversal,particle recognition and other products.The results indicated that the storm occurred in an environment characterized by high energy,vertical instability,and weak vertical wind shear.Hail storm was stronger than extreme wind storm,which developed at higher altitudes,and exhibited a backward-propagating storm pattern.During the hail event,the mass center of the horizontally polarized reflectivity factor(ZH)and differential phase shift(KDP)sank,while the low-level differential reflectivity(ZDR)decreased.The region of low correlation coefficient(CC)first extended upwards before sinking as the ice crystal particles and supercooled water drops fell from the layer above 0℃,reaching the ground as rain and hail.Extreme gale storm developed in a basin,with the terrain convergence enhancing its intensity.As the extreme wind approached,the center of mass for ZH and KDP rapidly dropped to the near-surface and falling precipitation particles strengthened the downward airflow through entrainment.Besides,the horizontal contraction and vertical stretching of the radial convergence velocity in the middle layer,along with the divergence velocity near the surface indicated the occurrence of localized downburst.Meanwhile,the downslope terrain and the basin venturi effect further accelerated the near-surface winds.

Abstract:Based on simulations from 18 climate models participating in the Coupled Model Intercomparison Project Phase 6(CMIP6),this study uses the piControl experiment as a reference for a base climate state and the abrup-4×CO2 experiment to simulate the climate response under an extreme greenhouse gas forcing scenario.The Climate Feedback Response Analysis Method(CFRAM)was employed to quantitatively evaluate the contributions of external forcings,various radiative feedback processes,and non-radiative processes to the warming over Guangdong Province and the associated uncertainties.The results show that the projected warming over Guangdong varies across models with an estimated range of 3.71℃to 7.07℃,and a robust estimate centered between 4.42℃and 5.40℃.The warming is most pronounced in the western and northern regions of the province.A quantitative attribution analysis of the multi-model ensemble mean indicates that water vapor feedback and CO2 forcing are the primary drivers of the projected warming,contributing 4.92℃and 2.42℃,respectively.Surface heat storage and cloud feedback also provide significant positive contributions to the warming,while surface sensible heat flux and latent heat flux contribute to a cooling effect.Uncertainty analysis reveals that cloud shortwave radiative effects,surface sensible heat flux,and surface heat storage are the main sources of uncertainty,stemming primarily from differences in model representations of cloud physics,surface energy balance,and heat redistribution.Future research to improve the accuracy of regional climate projections should focus on better constraining these key processes.

Abstract:Typhoon Haikui(2012)was influenced by a saddle-point flow field,resulting in nearly stationary movement in southern Anhui and causing extreme rainfall.To examine the sensitivity of the track and precipitation forecast of Haikui to the model's initial conditions after entering the saddle-point region,ensemble forecast experiments and sensitivity experiments were conducted for Haikui using the Weather Research&Forecasting(WRF)model coupled with an Ensemble Kalman Filter(EnKF)assimilation system.The results show that:(1)After Typhoon Haikui made landfall in Zhejiang and moved northwestward into Anhui,typhoon tracks forecast by ensemble forecasting experiments can mainly be divided into the stagnation type,the eastward-turning type,and the westward-moving type.The stagnation tracks are most similar to the observed track of Haikui.(2)Compared with the circulation of the typhoon itself,differences in the environmental field of initial conditions are the key factor causing variations in typhoon motion in the experiments.The typhoon's movement is more sensitive to the continental high-pressure system and mid-latitude trough in the northwestern part of the simulation domain.(3)When the intensities of the continental high and trough are comparable in the initial field,the weakened typhoon circulation remains between the continental high and the subtropical high,leading to stagnation over southern Anhui.This,combined with cold air interaction,results in heavy rainfall in southern Anhui and northern Jiangxi.When the continental high is weak and the trough is strong in the initial field,the typhoon is forecasted to be in a saddle-point field in the 24-hour forecast.However,both the continental high and the subtropical high are relatively weak.As the trough moves southward,the continental high weakens and retreats westard,and the subtropical high shifts eastward.The typhoon then turns northeastward,resulting in heavy rainfall in Jiangsu.Conversely,when the continental high is strong but the trough is weak,the typhoon continues moving westward.The influence of the trough and cold air shifts to the north.Cold air is blocked by the high pressure and fails to directly interact with the typhoon's circulation,resulting in relatively weaker precipitation in southern Anhui.

Abstract:The WRF model was used for numerical sensitivity simulations to analyze the influence of short-wave radiation diurnal variation on the landing and weakening stage of Typhoon In-fa(2106).The results show that the CTRL experiment successfully reproduced the weakening process of the typhoon.In the daytime experiment,the simulated typhoon weakened more slowly and wasmaintained for a longer time,Whereas it weakened rapidly in the nighttime experiment.Compared with the CTRL experiment,the daytime experiment had stronger short-wave radiation,leadingwhich led to increased surface temperature,latent heat flux from surface evaporation,and water vapor from the ocean to the atmosphere,keeping the lower atmosphere highly unstable.Sustained radiative heating and latent heat release in the middle layer produced strong vertical motion.Meanwhile,increased relative humidity in the upper layer promoted upper-level cloud formation,and upper-level stability decreased due to cloud radiative forcing.The continuous release of high-intensity latent heat prevented the breakdown of the typhoon's warm core structure,which is the main reason the typhoon intensity was maintained.In the nighttime experiment,however,surface radiative cooling reduced the energy supply to the lower layer.The typhoon thus released its unstable energy quickly,reached a stable state rapidly,and consequently weakened swiftly.

Abstract:Based on observation and forecast data of 15-min wind power output of all wind farms in Shanxi Province from March 2022 to March 2023,this study analyzed the errors of"power prediction curves"in the provincial wind power trading market.The forecasting effectiveness of wind power ramp events was evaluated by developing identification,matching,and scoring methods.The ERA5 reanalysis data were used to analyze the weather background and evolution of large-scale wind power ramp-up events.The results are shown as follows:(1)For all seasons,the periods around 08:00(Beijing Time,same below)and 20:00 are the two daily peaks of forecast error.Meanwhile,the error dispersion for the two periods is relatively large,with a maximum value of up to 4 000 MW.(2)The occurrence frequency for weak ramp events is significantly higher than strong ramp events.The forecasting skill score for weak ramp events is 0.23,which is higher than strong ramp events(0.05).(3)Wind power ramp events mainly occur in winter and spring.The periods around 8:00 and 19:00 each day are the concentrated occurrence times for ramp-down and ramp-up events,respectively.Ramp-up events mainly occur from afternoon to the first half of the night,while ramp-down events are dominant from the second half of the night to the next morning.(4)When Shanxi Province is located behind a high surface pressure or in front of an upper-air trough,the intensification of the surface high pressure or the establishment of a pre-trough low-level jet will lead to province-wide wind power ramp-up events.

Abstract:To explore the value of adaptive cooperative observations from multi-band radar in capturing tornado features and improving operational warnings,and to construct a tornado warning index for operational reference,this study analyzes the rare EF3-level tornado case on 19 September 2023 in Funing,Jiangsu,which is a rare tornado event for September in the history of Jiangsu.The analysis is conducted from the multi-band radar echo characteristics,tornado nowcasting,and early warning verification.The key findings are as follows.(1)The S-band and X-band weather radars cooperate to provide meso-and micro-scale features of the tornado and its parent storm.Funing EF3-Level tornado convection pattern conforms to the tornadic supercell conceptual model and the mature mesocyclone conceptual model,exhibiting mesoscale features such as Tornado Vortex Signature(TVS),Tornadic Debris Signature(TDS),Zdr arc,Kdp arc,Zdr column,Kdp column,CC column,W column,and Reflectivity Core(RC).(2)Both the multi-band radar TDS,which is associated with strong updrafts,and the mesocyclone,which is directly linked to the tornado,can be used for tornado targeted warning.The lead warning time of Funing tornado using TDS is 13 min and 33 min,while the lead warning time of Funing tornado using mesocyclone is 19 min and 39 min.In addition,the height of TDS offers a certain indication of the intensity and maintenance of the tornado.(3)According to early warning test of Jianhu EF1-level tornado on 19 September,the combined warning of multi-band radar TDS and mesocyclonic characteristics reached a lead time of 15 min.These findings provide a reference for the multi-band weather radar monitoring and warning of tornadoes in Jiangsu.

Abstract:The Pearl River Basin is a climate-sensitive area where climate warming leads to frequent heatwave disasters.The Comfort Index of Human Body integrates multiple meteorological elements to better assess the impact of heatwaves on human health.Based on the index for defining heatwaves,meteorological observations and five CMIP6 climate model data containing seven SSPs were used to analyse the characteristics of summer heatwaves and exposed vulnerable population in the Pearl River Basin during the historical period(1961-2022)and the 21st century's near-term(2021-2040),mid-term(2041-2060),and end-term(2081-2100).The results show:(1)From 1961 to 2022,the maximum number of summer heatwave days in the Pearl River Basin can be up to 42 days,with the longest duration lasting up to 16 days.The average annual heatwave-affected area is approximately 165,000 km²,with the Pearl River Delta experiencing the most severe heatwaves.(2)In the near,mid,and end terms of the 21st century,compared to the baseline period(1995-2014),the basin-wide average number of summer heatwave days will increase by approximately 7-9 days,11-29 days,and 5-75 days,respectively,with the maximum duration increasing by about 2-4 days,5-10 days,and 2-50 days.The increase is more significant in the high-altitude areas upstream of the Pearl River Basin;the heatwave-affected area continues to expand,and the heatwaves in the late 21st century may influence the entire basin.(3)From 1961 to 2022,the annual average exposed vulnerable population in summer was about 28.6 million person-days,and the maximum in the Pearl River Delta can reach 108 million person-days in 2019.Future vulnerable population exposure will increase significantly,about 6.4-7.8 times,18.5-38.4 times,and 11.9-117.7 times of the baseline period in the near,mid,and end terms,respectively,especially in the Eastern Pearl River Basin.(4)The combined effect of climate and population change is the dominant factor determining the future changes in the exposure of vulnerable populations in the Pearl River Basin,and its contribution rate gradually increases over time.There is an urgent need to strengthen heatwave prevention measures in the future.

Abstract:Aiming at the scarcity of offshore observations in the Guangdong sea area and the demand for high-precision,long-term data in wind power development,this study collected gradient wind data from 7 observation sites(comprising wind measurement towers and lidars)during 2012-2021.It developed a three-dimensional Barnes objective analysis method and a multi-source data fusion and assimilation technology,and constructed a high-resolution three-dimensional grid dataset(1 000 m horizontal resolution;10 m and 30 m vertical levels)and hourly wind field data at 20 reference points based on ERA5 reanalysis data.Error analysis shows that the correlation coefficient between the fused-assimilated wind speed and the observed wind speed are all≥0.82(with an average of 0.913),the average root mean square error(RMSE)is 1.20 m·s-1,and the wind speed error at heights above 30-40 m is≤1 m·s-1.The accuracy is significantly higher than that of ERA5 and power-law fitting.The average RMSE of wind direction is 20.2 °.The dataset clearly presents the spatiotemporal distribution characteristics of wind speed,such as increasing with the offshore distance and exhibiting annual dual peaks in winter and summer(winter>summer).It can also effectively depict the diurnal variation of wind fields,the passage of key weather systems(including typhoons),and climate system features such as monsoons and land-sea breezes.This dataset can provide reliable data support for wind farm planning,site selection,and wind energy resource assessment in the Guangdong sea area.

Abstract:Data from four two-dimensional raindrop spectrum at the Fengyun Satellite South China Precipitation Validation Station,along with co-located minute rain gauge data from Apirl to May 2024,were used.The consistency of precipitation observations between the two instruments was analyzed by comparing cumulative rainfall and instantaneous rain rates.The results indicate that at three of the four stations,the cumulative rainfall from the disdrometer exceeded that from the rain gauge by 10%to 13%.At a certain station,the disdrometer's cumulative rainfall was 0.5%lower.The correlation coefficient for cumulative rainfall was high,with an average above 0.99.When comparing instantaneous rain rate,applying a cubic spline interpolation to the rain gauge data significantly reduced the bias and mean absolute error between the the two instruments.This was particularly effective for weak precipitation(<10 mm·h-1)with discontinuous observations,where the correlation coefficient can be increased from 0.37 to approximately 0.8.This indicates that the interpolation method is reasonable and feasible for generating high-temporal-resolution instantaneous precipitation rate datasets from conventional rain gauge.The evaluation demonstrated good consistency between the two-dimensional raindrop spectrum and the rain gauge,suggesting these datasets can support the ground validation and algorithm improvement of Fengyun satellite precipitation retrieval products.

Abstract:Numerical weather forecast is the mainstream technique in modern weather forecasting,which has been developing towards higher resolution in recent years,while forecast errors remain unavoidable.This paper proposes a numerical forecast bias correction model,termed LSTM-GAM-UNet(LGU-Net),which introduces the Long Short-Term Memory(LSTM)structure and Global Attention Mechanism(GAM)based on the CU-Net model.The model further integrates various meteorological elements,terrain features derived from"Jilin-1"satellite data,and satellite cloud images,thereby constructing a multi-element fusion correction model.This model is also specifically optimized for meteorological forecasting.An experiment was conducted in northeastern China to correct the biases of the Global Forecast System(GFS)for the 2 m temperature(T2),2 m dew point temperature(D2),10 m wind components(U10,V10),and precipitation.Different models were compared and analyzed for bias correction experiments.By comparing with the original GFS forecasts,as well as the correction results from Anomaly Numerical-correction with Observation(ANO)method and the CU-Net method,it was shown that the LGU-Net model effectively improved the bias correction performance.In addition,the addition of cloud imagery data has a significant positive impact on precipitation correction,with an 80.76%and 76.04%improvement in RMSE and MAE compared to GFS,respectively.This paper provides new technical support for high-precision meteorological element forecasts.

Abstract:Based on the radiation observations and the basic meteorological elements from the Dianbai National Climate Reference Station(Dianbai station)and the Integrated Observation Platform for Marine Meteorology of the Marine Meteorological Science Experiment Base(MMSEB)at Bohe(offshore platform)of China Meteorological Administration(CMA)in 2019,this study comparatively analyzes the annual average monthly variation,daily variation,and monthly average daily variation of radiation at two stations.The relationships between radiation and other meteorological elements are also analyzed.The main conclusions are as follows.(1)The monthly average maximum values for each radiation components at the two sites generally occur in summer and autumn(with the exception of upward shortwave radiation at the offshore platform,which peaks in December),while the minimum values occur in winter and spring.Due to differences in the underlying surface,the monthly average upward short-wave radiation at Dianbai station is significantly higher than that at the offshore platform,resulting in a notably lower annual net radiation of 103.8 W·m-2 at Dianbai station compared to 145.8 W·m-2 at the offshore platform.(2)The comparison of the daily variation of net radiation at the two stations reveals significant differences between the land and sea.During the daytime,the net radiation at the offshore platform is,on average,35.8%higher than that at Dianbai station,with an average value of 92.5 W·m-2.At nighttime,the net radiation at the offshore platform is,on average,33.1%lower than that at Dianbai station,with an average value of-7.34 W·m-2.(3)The downward shortwave radiation of offshore platforms is greater than that at Dianbai station,with its peak-values concentrated around noon from June to November.The upward short-wave radiation at Dianbai station is obviously higher than that at offshore platform,with the largest differences between the two stations occurring between summer and winter.The short-wave albedo at Dianbai station is generally higher than that at the offshore platform station,and both stations exhibit a high value area in the evening from September to January,while the offshore platform also shows a secondary peak in the morning.(4)The spatial distribution of the upward long-wave radiation at Dianbai station during the daytime are similar to the downward short-wave radiation.In contrast,the upward long-wave radiation at offshore platform reflects the seasonal characteristics of the overall sea temperature.The downward long-wave radiation at Dianbai station is generally greater than that at the offshore platform in the daytime,but smaller in the rest of the time,especially in winter.