J. Meteor. Res.
  Home    About JMR    Author Guide    Editorial Board    Reviewer List    Subscribe    Volunteers    ACTA


JMR has established a new website (http://jmr.cmsjournal.net). The current website will not be updated after 1 December 2019. Visit us at jmr.cmsjournal.net to find out more. Add JMR WeChat (ID: JMR-CMS) to dynamically follow up with JMR publications.  



Current Issue  
2019 Vol. 33, No. 5
Published: 2019-10-28

Forecasting Different Types of Convective Weather: A Deep Learning Approach
Kanghui ZHOU, Yongguang ZHENG, Bo LI, Wansheng DONG, Xiaoling ZHANG
2019, 33(5): 797-809 [Abstract]( 547 ) HTML PDF (3642 KB)  ( 486
Abstract:A deep learning objective forecasting solution for severe convective weather (SCW) including short-duration heavy rain (HR), hail, convective gusts (CG), and thunderstorms based on numerical weather prediction (NWP) data was developed. We first established the training datasets as follows. Five years of severe weather observations were utilized to label the NCEP final (FNL) analysis data. A large number of labeled samples for each type of weather were then selected for model training. The local temperature, pressure, humidity, and winds from 1000 to 200 hPa, as well as dozens of convective physical parameters, were taken as predictors in our model. A six-layer convolutional neural network (CNN) model was then built and trained to obtain optimal model weights. After that, the trained model was used to predict SCW based on the Global Forecast System (GFS) forecast data as input. The performances of the CNN model and other traditional methods were compared. The results show that the deep learning algorithm had a higher classification accuracy on HR and hail than support vector machine, random forests, and other traditional machine learning algorithms. The objective forecasts by use of the deep learning algorithm also showed better forecasting skills than the subjective forecasts by the forecasters. The threat scores (TSs) of thunderstorm, HR, hail, and CG were increased by 16.1%, 33.2%, 178%, and 55.7%, respectively. The deep learning forecast model is currently used in the National Meteorological Center of China to provide guidance for the operational SCW forecasting over China.
A Novel Identification of the Polar/Eurasia Pattern and Its Weather Impact in May
Ni GAO, Cholaw BUEH, Zuowei XIE, Yuanfa GONG
2019, 33(5): 810-825 [Abstract]( 202 ) HTML PDF (3144 KB)  ( 140
Abstract:The Polar/Eurasia (POL) pattern was previously identified based on the empirical orthogonal function method and monthly mean data, in which the positive and negative phases are anti-symmetric in spatial distribution. This paper identifies the positive (POL+) and negative (POL-) phases of the POL pattern through applying a novel approach, i.e., self-organizing maps, to daily 500-hPa geopotential height fields in May over 1948–2017. The POL+, POL1-, and POL2- patterns defined by this method represent actual physical modes. The POL+ pattern features a wave train from the northeastern Atlantic/northern Europe via the subarctic regions of Eurasia to Lake Baikal. The POL1- pattern is characterized by a planetary-scale dipole pattern with a positive anomaly band over subarctic Eurasia and a negative anomaly band from central Asia to the Sea of Okhotsk. The anomaly centers of the POL2- pattern are basically anti-symmetrical to those of the POL+ pattern. The POL+ pattern increases the blocking frequency over the northeastern Atlantic/northern Europe and northeastern Asia, where high-frequency transient eddies are highly recurrent in the north. Accordingly, precipitation increases apparently in the subarctic Asian continent and western Siberia, and decreases around Europe and Lake Baikal. A mimic wave train is also observed in the surface air temperature anomaly field. During the POL1- period, the blocking frequency is abnormally high over Eurasia, whereas high-frequency transient eddies are apparently suppressed over northern Eurasia. Correspondingly, significant precipitation deficits are observed in northern Eurasia. The POL1- pattern also causes a remarkable temperature increase in the subarctic seas of Eurasia and a considerable temperature drop in the midlatitude Asian continent. As the POL2- pattern prevails, the blocking frequency decreases over the North Atlantic/Europe but strengthens over the Asian continent. The POL2- pattern also causes wavelike anomalies of precipitation and surface air temperature over northern Eurasia.
Asymmetry of Atmospheric Responses to Two-Type El Niño and La Niña over Northwest Pacific
Mingcheng CHEN, Tim LI, Xiaohui WANG
2019, 33(5): 826-836 [Abstract]( 172 ) HTML PDF (2122 KB)  ( 123
Abstract:The mechanism for asymmetric atmospheric responses to the central Pacific (CP) El Niño and La Niña over the western North Pacific (WNP) is studied in this paper. The negative anomalies of rainfall over the key region of WNP are explained by diagnosing the column-integrated equations of moisture and moist static energy (MSE). It is revealed that the nonlinear advection of moist enthalpy is critical to introduce negative rainfall anomalies over the region. The anomalous easterly (westerly) in La Niña (CP El Niño) causes negative advection of anomalous moist enthalpy, inducing negative heating anomaly and an anticyclone anomaly in the WNP, which weakens (strengthens) the cyclone (anticyclone) in La Niña (CP El Niño). The MSE budget analysis shows a larger nonlinear term in CP El Niño than in eastern Pacific (EP) El Niño, inconsistent with the amplitudes of sea surface temperature anomalies. The reason is that the nonlinear term transforms to positive above 700 hPa in EP El Niño, offsetting the negative advection below 700 hPa and thus making the nonlinear term smaller. The nonlinear term is negative at low levels in CP El Niño, resulting in a larger nonlinear term. The stronger precipitation anomalies in the WNP during EP El Niño can be attributed to the linear moist enthalpy advection. The mean easterly wind at mid levels causes a larger (smaller) positive moist enthalpy advection in CP (EP) El Niño, due to a larger (smaller) moist enthalpy gradient, resulting in a positive (negative) linear moist enthalpy advection, which weakens (strengthens) the negative precipitation anomalies in the key region.
Seasonal Climate Prediction Models for the Number of Landfalling Tropical Cyclones in China
Baoqiang TIAN, Ke FAN
2019, 33(5): 837-850 [Abstract]( 211 ) HTML PDF (5317 KB)  ( 157
Abstract:Two prediction models are developed to predict the number of landfalling tropical cyclones (LTCs) in China during June–August (JJA). One is a statistical model using preceding predictors from the observation, and the other is a hybrid model using both the aforementioned preceding predictors and concurrent summer large-scale environmental conditions from the NCEP Climate Forecast System version 2 (CFSv2). (1) For the statistical model, the year-to-year increment method is adopted to analyze the predictors and their physical processes, and the JJA number of LTCs in China is then predicted by using the previous boreal summer sea surface temperature (SST) in Southwest Indonesia, preceding October South Australia sea level pressure, and winter SST in the Sea of Japan. The temporal correlation coefficient between the observed and predicted number of LTCs during 1983–2017 is 0.63. (2) For the hybrid prediction model, the prediction skill of CFSv2 initiated each month from February to May in capturing the relationships between summer environmental conditions (denoted by seven potential factors: three steering factors and four gene-sis factors) and the JJA number of LTCs is firstly evaluated. For the 2- and 1-month leads, CFSv2 has successfully reproduced these relationships. For the 4-, 3-, and 2-month leads, the predictor of geopotential height at 500 hPa over the western North Pacific (WNP) shows the worst forecasting skill among these factors. In general, the summer relative vorticity at 850 hPa over the WNP is a modest predictor, with stable and good forecasting skills at all lead times.
Development of Land Surface Model BCC_AVIM2.0 and Its Preliminary Performance in LS3MIP/CMIP6
Weiping LI, Yanwu ZHANG, Xueli SHI, Wenyan ZHOU, Anning HUANG, Mingquan MU, Bo QIU, Jinjun JI
2019, 33(5): 851-869 [Abstract]( 213 ) HTML PDF (3385 KB)  ( 178
Abstract:The improvements and validation of several parameterization schemes in the second version of the Beijing Climate Center Atmosphere–Vegetation Interaction Model (BCC_AVIM2.0) are introduced in this study. The main updates include a replacement of the water-only lake module by the common land model lake module (CoLM-lake) with a more realistic snow–ice–water–soil framework, a parameterization scheme for rice paddies added in the vegetation module, renewed parameterizations of snow cover fraction and snow surface albedo to accommodate the varied snow aging effect during different stages of a snow season, a revised parameterization to calculate the threshold temperature to initiate freeze (thaw) of soil water (ice) rather than being fixed at 0℃ in BCC_AVIM1.0, a prognostic phenology scheme for vegetation growth instead of empirically prescribed dates for leaf onset/fall, and a renewed scheme to depict solar radiation transfer through the vegetation canopy. The above updates have been implemented in BCC_AVIM2.0 to serve as the land component of the BCC Climate System Model (BCC_CSM). Preliminary results of BCC_AVIM in the ongoing Land Surface, Snow, and Soil Moisture Model Intercomparison Project (LS3MIP) of the Coupled Model Intercomparison Project Phase 6 (CMIP6) show that the overall performance of BCC_AVIM2.0 is better than that of BCC_AVIM1.0 in the simulation of surface energy budgets at the seasonal timescale. Comparing the simulations of annual global land average before and after the updates in BCC_AVIM2.0 reveals that the bias of net surface radiation is reduced from -12.0 to -11.7 W m-2 and the root mean square error (RMSE) is reduced from 20.6 to 19.0 W m-2; the bias and RMSE of latent heat flux are reduced from 2.3 to -0.1 W m-2 and from 15.4 to 14.3 W m-2, respectively; the bias of sensible heat flux is increased from 2.5 to 5.1 W m-2 but the RMSE is reduced from 18.4 to 17.0 W m-2.
Relationship between Extreme Precipitation and Temperature in Two Different Regions: The Tibetan Plateau and Middle–East China
Rui WANG, Tao XIAN, Mengxiao WANG, Fengjiao CHEN, Yuanjian YANG, Xiangdong ZHANG, Rui LI, Lei ZHONG, Chun ZHAO, Yunfei FU
2019, 33(5): 870-884 [Abstract]( 203 ) HTML PDF (4271 KB)  ( 182
Abstract:The change of extreme precipitation with temperature has regional characteristics in the context of global warming. In this study, radiosonde data, co-located rain gauge (RG) observations, and Tropical Rainfall Measuring Mission (TRMM) precipitation radar (PR) products are used to explore the relationship between extreme precipitation intensity and near-surface temperature in Middle–East China (MEC) and the eastern Tibetan Plateau (TP) during 1998–2012. The results show that extreme precipitation intensity increases with increasing temperature at an approximate Clausius–Clapeyron (C–C) rate (i.e., water vapor increases by 7% as temperature increases by 1℃ based on the C–C equation) in MEC and TP, but the rate of increase is larger in TP than in MEC. This is probably because TP (MEC) is featured with deep convective (stratiform) precipitation, which releases more (less) latent heat and strengthens the convection intensity on a shorter (longer) timescale. It is also found that when temperature is higher than 25℃ (15℃) in MEC (TP), the extreme precipitation intensity decreases with rise of temperature, suggesting that the precipitation intensity does not always increase with warming. In this case, the limited atmospheric humidity and precipitable water could be the primary factors for the decrease in extreme precipitation intensity at higher temperatures.
The Effect of Solar Cycle on Climate of Northeast Asia
Yan SONG, Zhicai LI, Yu GU, Kuo-Nan LIOU, Xiaoxin ZHANG, Ziniu XIAO
2019, 33(5): 885-894 [Abstract]( 238 ) HTML PDF (3974 KB)  ( 169
Abstract:The impact of solar activity on climate system is spatiotemporally selective and usually more significant on the regional scale. Using statistical methods and solar radio flux (SRF) data, this paper investigates the impact of the solar 11-yr cycle on regional climate of Northeast Asia in recent decades. Significant differences in winter temperature, precipitation, and the atmospheric circulation over Northeast Asia are found between peak and valley solar activity years. In peak years, temperature is higher over vast areas of the Eurasian continent in middle and high latitudes, and prone to producing anomalous high pressure there. Northeast Asia is located to the south of the anomalous high pressure, where the easterlies prevail and transport moisture from the western Pacific Ocean to the inland of East Asia and intensify precipitation there. In valley years, temperature is lower over the Eurasian continent and northern Pacific Ocean in middle and high latitudes, and there maintain anomalous low pressure systems in the two regions. Over the Northeast Asian continent, north winds prevail, which transport cold and dry air mass from the high latitude to Northeast Asia and reduce precipitation there. The correlation coefficient of winter precipitation in Northeast China and SRF reaches 0.4, and is statistically significant at the 99% confidence level based on the Student's t-test. The latent heat flux anomalies over the Pacific Ocean caused by solar cycle could explain the spatial pattern of abnormal winter precipitation of China, suggesting that the solar activity may change the climate of Northeast Asia through air-sea interaction.
Improved Calculation of Turbulence Parameters Based on Six Tropical Cyclone Cases: Implication to Wind Turbine Design in Typhoon-Prone Areas
Binglan WANG, Zhiqiang HE, Lili SONG, Wenchao CHEN
2019, 33(5): 895-904 [Abstract]( 151 ) HTML PDF (1017 KB)  ( 97
Abstract:In view of the absence or insufficiency of tropical cyclone (TC) turbulence parameters in current design standards of wind turbines, in this paper, TC turbulence parameter models with roughness length involved are developed based on six landfall TCs observed from meteorological towers located on various underlying surfaces, so as to provide references for the wind turbine design under TC conditions. Firstly, the roughness length values are examined in order to reduce the effect on turbulence parameters of the various underlying surfaces. On this basis, the reference turbulence intensity is normalized by the roughness length. The related turbulence parameters are parameterized, including the turbulence standard deviation and the turbulence spectrum; and the turbulence parameters available under TC conditions for turbulence turbine design are presented finally. Comparisons of the wind parameter models presented in this paper with those used in current turbine design standards suggest that the former can represent TC characteristics more accurately. In order to withstand TCs, we suggest that the turbulence parameter models recommended in this paper be included in future wind turbine design standards under TC conditions.
Influence of High Relative Humidity on Secondary Organic Carbon: Observations at a Background Site in East China
Linlin LIANG, Guenter ENGLING, Yuan CHENG, Xiaoye ZHANG, Junying SUN, Wanyun XU, Chang LIU, Gen ZHANG, Hui XU, Xuyan LIU, Qianli MA
2019, 33(5): 905-913 [Abstract]( 126 ) HTML PDF (991 KB)  ( 109
Abstract:To investigate the impacts of relative humidity (RH) on secondary organic aerosol (SOA) concentrations and chemical reactions, the carbonaceous aerosol components [i.e., organic carbon (OC) and element carbon (EC)] were quantified in daily PM2.5 samples collected at a background site in East China during summer 2015. Based on the method of EC-tracer, the concentration of secondary organic carbon (SOC) demonstrated an obvious negative relationship with RH higher than 60%. Moreover, the ratio of SOC/EC also exhibited obvious decreasing trends with increasing RH, indicating negative effects for chemical production of SOA under high RH conditions. Due to high RH, photochemistry was weakened, gaseous oxidant concentrations was lowered (e.g., significantly decreased O3 levels), and the production rates of SOA were relatively low. On the other hand, because of more water uptake under higher RH conditions, the aerosol droplet acidity was reduced and enhancement of SOA formation by acidity was accordingly absent. In addition, high RH also plays an important role in changing viscosity of pre-existing aerosol coatings, which can affect reactive uptake yield of SOA. Overall, the results from this study imply that SOA production may be more associated with photochemical processes, while aqueous-phase chemistry is not very important for some SOA formation in a moist ambient environment. In the ambient atmosphere, oxidant concentrations, reaction rates, airborne species, etc., are highly variable. How do these factors affect SOA yields under given ambient environment warrants further detailed investigations.
Fengyun-3D MERSI True Color Imagery Developed for Environmental Applications
Xiuzhen HAN, Feng WANG, Yang HAN
2019, 33(5): 914-924 [Abstract]( 174 ) HTML PDF (1056 KB)  ( 108
Abstract:Many techniques were developed for creating true color images from satellite solar reflective bands, and the so-derived images have been widely used for environmental monitoring. For the newly launched Fengyun-3D (FY-3D) satellite, the same capability is required for its Medium Resolution Spectrum Imager-II (MERSI-II). In processing the MERSI-II true color image, a more comprehensive processing technique is developed, including the atmospheric correction, nonlinear enhancement, and image splicing. The effect of atmospheric molecular scattering on the total reflectance is corrected by using a parameterized radiative transfer model. A nonlinear stretching of the solar band reflectance is applied for increasing the image contrast. The discontinuity in composing images from multiple orbits and different granules is eliminated through the distance weighted pixel blending (DWPB) method. Through these processing steps, the MERSI-II true color imagery can vividly detect many natural events such as sand and dust storms, snow, algal bloom, fire, and typhoon. Through a comprehensive analysis of the true color imagery, the specific natural disaster events and their magnitudes can be quantified much easily, compared to using the individual channel data.
Radiometric Cross-Calibration for Multiple Sensors with the Moon as an Intermediate Reference
Lu ZHANG, Peng ZHANG, Xiuqing HU, Lin CHEN, Min MIN, Na XU, Ronghua WU
2019, 33(5): 925-933 [Abstract]( 124 ) HTML PDF (791 KB)  ( 97
Abstract:The instrument cross-calibration is an effective way to assess the quality of satellite data. In this study, a new me-thod is proposed to cross-calibrate the sensors among satellite instruments by using a RObotic Lunar Observatory (ROLO) model and Apollo sample reflectance in reflective solar bands (RSBs). The ROLO model acts as a transfer radiometer to bridge between the instruments. The reflective spectrum of the Apollo sample is used to compensate for the difference in the instrument's relative spectral responses (RSRs). In addition, the double ratio between the observed lunar irradiance and the simulated lunar irradiance is used to reduce the difference in instrument lunar viewing and illumining geometry. This approach is applied to the Moderate Resolution Imaging Spectroradiometer (MODIS), the Sea-Viewing Wide Field-of-View Sensor (SeaWiFS), and the Advanced Land Imager (ALI) on board three satellites, respectively. The mean difference between MODIS and SeaWiFS is less than 3.14%, and the difference between MODIS and ALI is less than 4.75%. These results indicate that the proposed cross-calibration method not only compensates for the RSR mismatches but also reduces the differences in lunar observation geometry. Thus, radiance calibration of any satellite instrument can be validated with a reference instrument bridged by the moon.
Simulation of the Northern and Southern Hemisphere Annular Modes by CAMS-CSM
Sulan NAN, Junli YANG, Yan BAO, Jian LI, Xinyao RONG
2019, 33(5): 934-948 [Abstract]( 211 ) HTML PDF (5253 KB)  ( 99
Abstract:As leading modes of the planetary-scale atmospheric circulation in the extratropics, the Northern Hemisphere (NH) annular mode (NAM) and Southern Hemisphere (SH) annular mode (SAM) are important components of global circulation, and their variabilities substantially impact the climate in mid-high latitudes. A 35-yr (1979-2013) simulation by the climate system model developed at the Chinese Academy of Meteorological Sciences (CAMS-CSM) was carried out based on observed sea surface temperature and sea ice data. The ability of CAMS-CSM in simulating horizontal and vertical structures of the NAM and SAM, relation of the NAM to the East Asian climate, and temporal variability of the SAM is examined and validated against the observational data. The results show that CAMS-CSM captures the zonally symmetric and out-of-phase variations of sea level pressure anomaly between the midlatitudes and polar zones in the extratropics of the NH and SH. The model has also captured the equivalent barotropic structure in tropospheric geopotential height and the meridional shifts of the NH and SH jet systems associated with the NAM and SAM anomalies. Furthermore, the model is able to reflect the variability of northern and southern Ferrel cells corresponding to the NAM and SAM anomalies. The model reproduces the observed relationship of the boreal winter NAM with the East Asian trough and air temperature over East Asia. It also captures the upward trend of the austral summer SAM index during recent decades. However, compared with the observation, the model shows biases in both the intensity and center locations of the NAM's and SAM's horizontal and vertical structures. Specifically, it overestimates their intensities.
Convectively Coupled Equatorial Waves Simulated by CAMS-CSM
Lu WANG, Tianjun ZHOU, Jian LI, Xinyao RONG, Haoming CHEN, Yufei XIN, Jingzhi SU
2019, 33(5): 949-959 [Abstract]( 210 ) HTML PDF (1980 KB)  ( 148
Abstract:The Chinese Academy of Meteorological Sciences developed a Climate System Model (CAMS-CSM) to participate in the upcoming Coupled Model Intercomparison Project phase 6 (CMIP6). In this study, we assessed the model performance in simulating the convectively coupled equatorial waves (CCEWs) by comparing the daily output of precipitation from a 23-yr coupled run with the observational precipitation data from Global Precipitation Climatology Project (GPCP). Four dominant modes of CCEWs including the Kelvin, equatorial Rossby (ER), mixed Rossby-gravity (MRG), tropical depression-type (TD-type) waves, and their annual mean and seasonal cycle characteristics are investigated respectively. It is found that the space-time spectrum characteristics of each wave mode represented by tropical averaged precipitation could be very well simulated by CAMS-CSM, including the magnitudes and the equivalent depths. The zonal distribution of wave associated precipitation is also well simulated, with the maximum centers over the Indian Ocean and the Pacific Ocean. However, the meridional distribution of the wave activities is poorly simulated, with the maximum centers shifted from the Northern Hemisphere to the Southern Hemisphere, especially the Kelvin, MRG, and TD waves. The seasonal cycle of each wave mode is generally captured by the model, but their amplitudes over the Southern Hemisphere during boreal winter are grossly overestimated. The reason for the excessive wave activity over the southern Pacific Ocean in the simulation is discussed.
Improved Assimilation of Fengyun-3 Satellite-Based Snow Cover Fraction in Northeastern China
Shuai ZHANG, Chunxiang SHI, Runping SHEN, Jie WU
2019, 33(5): 960-975 [Abstract]( 211 ) HTML PDF (6107 KB)  ( 211
Abstract:Assimilation of snow cover is an important method to improve the accuracy of snow simulation. However, the effects of snow assimilation are poor because satellite observed snow cover data contain erroneous information, such as cloud contamination. In this paper, an improved approach is proposed to reduce the effects of observational errors during assimilation of snow cover fraction acquired by the Fengyun-3 (FY-3) satellite in northeastern China. A snow depth constraint was imposed on quality control of a snow depth product from a microwave radiation imager. The assimilation experiments were carried out before and after quality control (denoted as SCFDA and SCFDA_WSD, respectively). The snow cover fraction results were evaluated against the Moderate Resolution Imaging Spectroradiometer (MODIS) snow cover products. When assimilating the snow cover fraction with the snow depth constraint (i.e., SCFDA_WSD), substantially larger improvement was obtained than that without such a constraint/quality control (SCFDA), and the deviation and root mean square error of the snow cover fraction were significantly reduced. The assimilation performance was also evaluated against in-situ snow depth observations. The SCFDA_WSD also showed greater improvements during the snow accumulation and snowmelt periods than the SCFDA. The SCFDA_WSD improvements in woodland and shrubland were the most obvious. At different altitudes, the effects of the SCFDA_WSD were basically equivalent, and the deeper the snow depth was, the better the effect. In addition, the SCFDA_WSD method was found in close agreement with the observations during a sudden snowfall event.
Differences in the Rainfall Characteristics between Mount Tai and Its Surrounding Areas
Yuting GAN, Nina LI, Jian LI
2019, 33(5): 976-988 [Abstract]( 210 ) HTML PDF (3017 KB)  ( 186
Abstract:As a typical small-scale, isolated topography, Mount Tai exhibits great differences in precipitation characteristics from the surrounding areas. It was found that the amount of rainfall occurring over Mount Tai is significantly larger than what is observed in the surrounding areas. Based on hourly rain gauge records for the warm season (May to September) of 1996–2015, differences between Mount Tai and its surrounding areas were further revealed in terms of rainfall diurnal variation, spatial scale, and evolution process. The diurnal variation of the enhancement on rainfall amount exhibit “dual peaks” occurring in the early morning and afternoon, and the dual peaks are mainly attributable to rainfall frequency. The diurnal phase of the rainfall amount in the surrounding areas lags 1 h behind that over Mount Tai. Regarding differences in rainfall spatial scale, compared to those of surrounding areas, precipitation over Mount Tai is characterized by a smaller coverage, especially in the early morning. Mount Tai also tends to have a kind of unique, small-scale rainfall in the afternoon and at night. Based on statistical analysis of precipitation events, differences in rainfall evolution process were identified as well. Rainfall over Mount Tai often starts earlier in the afternoon and ends later at night than it does in the surrounding areas. Furthermore, nocturnal rainfall events over Mount Tai are prone to peaking over a shorter period and enduring for a longer period after reaching their maximum intensity, compared with nocturnal rainfall events occurring in the surrounding areas. Rainfall events over Mount Tai always last longer, especially those occurring in the early morning. In general, Mount Tai has a large enhancement effect on rainfall.
How Much Can AI Techniques Improve Surface Air Temperature Forecast?—A Report from AI Challenger 2018 Global Weather Forecast Contest
Lei JI, Zaiwen WANG, Min CHEN, Shuiyong FAN, Yingchun WANG, Zhiyuan SHEN
2019, 33(5): 989-992 [Abstract]( 125 ) HTML PDF (324 KB)  ( 95
Abstract:In August 2018, the Institute of Urban Meteorology (IUM) in Beijing co-organized with Sinovation Ventures a Weather Forecasting Contest (WFC)—one of the AI (artificial intelligence) Challenger Global Contests. The WFC aims to take advantage of the AI techniques to improve the quality of weather forecast. Across the world, more than 1000 teams enrolled in the WFC and about 250 teams completed real-time weather forecasts, among which top 5 teams were awarded in the final contest. The contest results show that the AI-based ensemble models exhibited improved skill for forecasts of surface air temperature and relative humidity at 2-m and wind speed at 10-m height. Compared to the IUM operational analog ensemble weather model forecast, the most notable improvements of 24.2% and 17.0% in forecast accuracy for surface 2-m air temperature are achieved by two teams using the AI techniques of time series model, gradient boosting tree, depth probability prediction, and so on. Meanwhile, it is found that reasonable data processing techniques and model composite structure are also important for obtaining better forecasts.
  Paper Search
Search DOI Adv Search
  Featured Collections
  Conference Links
  Journal Links
Copyright © Journal of Meteorological Research
Tel:86-10-68407634, 86-10-58993104    Zip/Postal Code: 100081    E-mail: cmsams@cms1924.org
Supported by: Beijing Magtech