Vol.33, NO.3, 2022

Display Method:
Daily Maximum Air Temperature Forecast Based on Fully Connected Neural Network
Zhao Linna, Lu Shu, Qi Dan, Xu Dongbei, Ying Shuang
2022, 33(3): 257-269. DOI: 10.11898/1001-7313.20220301

Objective forecast of maximum temperature is an important part in numerical weather prediction(NWP). The forecast uncertainty of near-surface meteorological elements is greater than that of upper atmospheric elements due to the impact of uncertainty in numerical forecasting models for sub-grid and boundary layer schemes.In recent years, meteorological observations expand rapidly, making traditional error correct method difficult to deal with the massive data. As a result, artificial intelligence has an increasingly obvious advantage in processing big data. Based on the fully connected neural network, four sensitivity experiments are designed in order to investigate the importance of auxiliary variable, time-lagged variable and the effectiveness of embedding layer in the neural network. The output products of high resolution(HRES) model of European Centre for Medium-Range Weather Forecasts(ECMWF) and the observations of basic meteorological elements of totally 2238 basic weather stations from 15 January 2015 to 31 December 2020 are employed. The training period is from 15 January 2015 to 31 December 2019, and the rest part is test period.The results show that the forecast error of daily maximum air temperature from the HRES in test period is reduced greatly by the sensitivity experiments, which add auxiliary variables, daily maximum air temperature with 1-2 lag days and embedding layer structures and their combination. The root mean square error is reduced by 29.72%-47.82% and the accuracy of temperature forecast are increased by 16.67%-38.89%, and the effects for Qinghai-Tibet Plateau is especially remarkable where the forecast error of HRES model is very high. It is preliminarily proved that the fully connected neural network with embedding layer has better overall performance than the raw fully connected neural network, and the features also affect the forecast errors and forecast skills of the model. Besides, the prediction error of neural network model with embedding layer is more stable when auxiliary variables and lag time variables are added. Positive forecasting techniques are available for almost all stations in the study, and it is possible to reduce the mean absolute error to less than 1℃ at many stations.

Precipitation Extrapolation Nowcasting in Beijing-Tianjin-Hebei Under Different Weather Backgrounds
Wang Yuhong, Bica Benedikt
2022, 33(3): 270-281. DOI: 10.11898/1001-7313.20220302

Rapid-refresh Multi-Scale Analysis and Prediction System-Integration (RMAPS_IN) is an important tool for Beijing, Hebei and other meteorological departments to make rapid-updated and refined precipitation nowcasting. The precipitation analysis products of the system are based on automatic station observation and radar quantitative precipitation estimation data, while 0-2 h forecast products are obtained by extrapolation based on the analysis products. To study the applicability of different extrapolation methods in RMAPS_IN, the precipitation events of different weather systems from 2019 to 2020 are analyzed, using cross correlation method and optical flow method to conduct a 0-2 h extrapolation nowcasting test based on the RMAPS_IN precipitation analysis products. The cross correlation method uses classic optimal correlation coefficient calculation scheme, while the optical flow method employs the Farneback dense optical flow calculation scheme in the OpenCV function library. According to the characteristics of the regional weather systems, the precipitation events are divided into five types: Low trough cold front precipitation, low vortex precipitation, typhoon precipitation, cyclone precipitation, and warm shear line precipitation. The sample size of each precipitation type is 2108, 1448, 1058, 260, and 140, respectively. The batch test results show that the extrapolated vectors by the cross correlation method and optical flow method have a certain difference in magnitude and direction. The direct difference has a clear correspondence with the position of the weather system that affects precipitation, and is more obviously affected by the geographical location. For typhoon precipitation, the difference in direction is distributed in an arc band, while for other 4 types of precipitation, the difference is large in the northwest and small in the southeast. In terms of forecasting effect, the cross correlation method is generally better than the optical flow method, especially when the forecast time exceeds 30 minutes, and the longer the lead time is, the more obvious the advantage is. But when the forecast time is 10 min, the optical flow method is better in the false alarm rate of low vortex precipitation, typhoon precipitation and warm shear line precipitation. In addition, the nowcasting method based on extrapolation has the best prediction effects on typhoon precipitation in Beijing-Tianjin-Hebei region, followed by warm shear line precipitation, low vortex precipitation, low trough cold front precipitation, and cyclone precipitation. It should be noted that in Beijing-Tianjin-Hebei region, cyclone precipitation and warm shear line precipitation rarely occurred in recent years, and the sample size of these two types of precipitation is significantly smaller than that of other types, so the relevant results are less representative.

Application of Machine Learning Classification Algorithm to Precipitation-induced Landslides Forecasting
Liu Haizhi, Xu Hui, Bao Hongjun, Xu Wei, Yan Xufeng, Lu Heng, Xu Chengpeng
2022, 33(3): 282-292. DOI: 10.11898/1001-7313.20220303

To address the practical needs of objectively describing the uncertainty of rainfall-based landslides and the existing problems of single warning indicators and subjective forecasting methods in the meteorological disaster early warning business, landslide disaster data from 2014 to 2020 and multi-source used precipitation analysis data are investigated to construct a regional rainfall-induced landslides probability forecasting model. Machine learning classification algorithms is implemented through key steps such as sample construction, model training, parameter optimization and forecast output to explore the feasibility of different types of algorithms in identifying landslides-causing rainfall processes. A training sample set construction method based on the positive samples, the negative samples are obtained by sampling under spatial-temporal limitation. The evaluation of different machine learning classification algorithms using the sample set shows that linear discriminant analysis algorithm has the highest accuracy(0.863) and the best generalization ability(area under the receiver operating characteristic curve is 0.886) without over-fitting problem, followed by the logistic regression algorithm and the K-nearest neighbor algorithm. In the probabilistic forecasting test for the cases of rainfall-induced landslides in 2021, all of three algorithms can extract and learn the conditional features and have certain ability to identify the rainfall processes which induce landslides. K-nearest neighbor algorithms and logistic regression algorithms have a relatively large range of probabilistic forecasting high value areas, which are prone to false alarm results. The probability forecast of the linear discriminant analysis algorithms is more convergent in the range of the high value area, and it can extract local rainfall information better, but it outputs unnecessary low-value probability forecasts in non-rainfall central area. The rainfall-induced landslides probability prediction model based on the machine learning classification algorithm comprehensively considers the coupling effect of the underlying surface factor and the rainfall factor, which is better than the commonly used critical threshold model that assumes the occurrence of landslides in the forecast area is only related to rainfall. The application results show that the machine learning classification algorithm model makes up for the shortcomings of existing forecasting models that are less likely to reflect the influence of the surface environment, so it is an important way to improve the performance of landslides forecasting and warning.

Precipitation Forecast Correction in South China Based on SVD and Machine Learning
Xie Shun, Sun Xiaogong, Zhang Suping, Xiong Zhaohui, Wei Xiaomin, Cui Congxin
2022, 33(3): 293-304. DOI: 10.11898/1001-7313.20220304

Precipitation can be induced by various weather systems and a series of complex physical processes, so its prediction is relatively difficult in weather forecasting. Due to the limitation of numerical model, the prediction error is inevitable. It is a hot topic in meteorological research and operation to explore a more effective method to correct the model product, and to improve the interpretation and applicability. To explore a more effective model product error correction method, a combination of correction methods is put forward, based on singular value decomposition (SVD) and machine learning, including multiple linear regression, LASSO regression and Ridge regression. The results are compared with the traditional matrix coefficient method, and then correction models are tested in pre-flood season precipitation forecast in South China, by correcting European Centre for Medium-Range Weather Forecasts (EC) product. The result shows that the proposed correction models combining SVD and machine learning can effectively reduce the error of EC product. The maximum optimization rate root mean square error is 4.2%, and more than 69% of the stations are optimized to different degrees. These correction models have better robustness to deal with the problem of collinearity between factors, and the correction effect is better than that of the traditional matrix coefficient method. Furthermore, the weighted integration of multiple correction models is carried out by assigning different weights to different models, and the root mean square error by the integrated approach in South China is smaller than EC product and any single correction model. It shows that the weighted ensemble method can better integrate the advantages of multiple correction models and enlarge the advantages. For the weighted ensemble of multiple correction models, it is not only better than the precipitation prediction results of EC product, but also better than any one of the integrated correction models. Its optimization rate of root mean squared error can achieve 5.7%, and more than 77% of the stations are optimized to different degrees.

Multiscale Characteristics of Two Supercell Tornados of Heilongjiang in 2021
Xu Yue, Shao Meirong, Tang Kai, Zhang Libao, Du Jing, Wang Yongchao
2022, 33(3): 305-318. DOI: 10.11898/1001-7313.20220305

Two strong supercell tornados hit Shangzhi Acheng of Harbin and Meilisi of Qiqihar, Heilongjiang Province on 1 June ("6·1" Tornado) and 9 June ("6·9" Tornado) in 2021. Using the conventional meteorological observations and Doppler weather radar data, the multiscale characteristics of two events are analyzed.Both events occur in the southeast quadrant of northeastern cold vortex. The left outlet of the upper-level jet stream and the southerly jet stream at lower-level are conducive to the development of vertical movement and the transport of warm-wet air. The temperature difference between 850 hPa and 500 hPa exceeds 30℃. Two storms are both triggered by mesoscale dry-lines and convergence lines. The pseudo-cold fronts, which generate from the mesoscale warm front and the cold pool coming from the thunderstorm outflow, are beneficial to the development and maintenance of tornados. Tornados appear on the wet part of the junctions between the pseudo-cold front and dry line, and in the front of cold pool. The parent storms of tornados rapidly develop into supercells as they pass over water bodies such as reservoirs and wetlands. The warm-wet inflow gap indicates the development of hook echo. Medium to strong mesocyclones firstly appear at about 3 km high, and then go upwards and downwards, touchdown 5-10 minutes later. The tornados occur when hook echoes and mesocyclones appear simultaneously.There are also some differences between them. Short-time heavy rainfall occurs on 1 June and thunder-gust occurs on 9 June with typical sounding layer structures, but "6·1" Tornado is stronger. The atmospheric instabilities are dominated by cold advection at upper-level for "6·1" Tornado but warm advection at lower-level for "6·9" Tornado. Water vapor and vertical velocity of "6·1" Tornado is more beneficial to the development of supercell than those of "6·9" Tornado. For the vertical wind shears of 0-1 km and 0-6 km, the corrected lifting condensation level and convective available potential energy (CAPE) are 12 m·s-1, 18 m·s-1, 770 m and 420 J·kg-1 for "6·1" Tornado, and 10 m·s-1, 33 m·s-1, 1100 m and 2500 J·kg-1 for "6·9" Tornado. The stronger 0-1 km wind shears and the lower corrected lifting condensation level show the possibility of intense tornado. CAPE may be underestimated because of the spatiotemporal resolution limitation for soundings.The main cause for the long duration of "6·1" Tornado is that the mesoscale vortex at 3 km altitude maintains due to the continuous warm-wet inflow. However, the strong mesocyclone of "6·9" Tornado doesn't last that long.

Refined Risk Assessment of Tropical Cyclone Disasters in Fujian
Zhuang Yao, Bao Ruijuan, Zhang Rongyan, Gao Shiyan, Pan Hang, Chen Si, Lin Xin
2022, 33(3): 319-328. DOI: 10.11898/1001-7313.20220306
Tropical cyclones have brought huge economic losses to Fujian Province. In order to achieve dynamic monitoring and early warning of wind and rain disaster risks caused by tropical cyclones, the disaster-causing mechanism of tropical cyclones is analyzed. A refined risk assessment method is developed to meet the needs of real-time decision-making on disaster prevention, mitigation, and reducing the economic losses caused by tropical cyclones.Through multiple rounds of rationality tests, 7 rain-induced disaster factors and 4 wind-induced disaster factors are picked out based on the tropical cyclone wind and rain data of 66 national meteorological stations from 1981 to 2021. And then, the risk assessment model of tropical cyclone disaster factors is established using the range standardization and correlation coefficient objective weighting method, and the risk level is divided by the natural breakpoint method and the disaster impacts.The results show that the risk assessment index system of disaster factors is reasonable, and the spatial distribution of disaster risk is investigated. The high rain risk areas are located along the coast, and the rain risk of Nanping and Sanming areas is low; the high wind risk area is significantly narrower than the high rain risk area, and the risk level decreases fast inland. Among them, the coastal areas from Luoyuan Bay to Chongwu are protected by the terrain barrier of Taiwan, and the risk is one level lower than that of the north and south parts of the coast. In addition, after the tropical cyclone lands on the east coast of Guangdong and moves northward, it often stays in the low-pressure cloud over the west of Fujian, resulting in a high risk area in the northwest of Fujian. Based on the spatial distribution of a single tropical cyclone, the disaster situation and the encrypted wind and rain data of regional stations, using the function of GIS and combining several typical tropical cyclone cases, a reasonable threshold for hazard classification is designed. It is targeted, especially urban waterlogging and mountain torrent disasters, which are basically consistent with the disaster situation, and provide more valuable reference information for meteorological disaster decision-making services.
Evaluation of GHMLLS Performance Characteristics Based on Observations of Artificially Triggered Lightning
Zhang Yue, Lü Weitao, Chen Lüwen, Wu Bin, Qi Qi, Ma Ying, Zhang Yang, Zheng Dong, Yan Xu, Meng Qing
2022, 33(3): 329-340. DOI: 10.11898/1001-7313.20220307
Artificially triggered lightning refers to the lightning that is artificially triggered to the ground under appropriate thunderstorm conditions. The location of artificially triggered lightning can be determined; the occurring time can be precisely stamped, and the channel-base current can be measured directly. Therefore, it's one of the effective methods to evaluate the performance of lightning location system (LLS). From the observations of artificially triggered lightning experiment conducted at the Field Experiment Base on Lightning Sciences, China Meteorological Administration from 2014 to 2019, 50 lightning flashes are selected to evaluate and analyze the performance characteristics of Guangdong-Hongkong-Macau Lightning Location System (GHMLLS).The results show that the lightning detection efficiency and stroke detection efficiency are about 96% (48/50) and 88% (233/265), respectively. The arithmetic mean, geometric mean and median values of location error are 279 m, 193 m and 202 m, respectively. The results show that there is a systematic deviation to the southwest in GHMLLS observations around the triggered lightning experiment site, which is about 170 m to the west and 50 m to the south. After correction, the arithmetic mean, geometric mean and median values of location error are reduced to 198 m, 108 m and 103 m, respectively. The linear fitting result with intercept of 0 shows that the LLS-inferred peak current of GHMLLS is about 65% of the direct measurement value of the channel-base current. Meanwhile, the arithmetic mean (median) value of the LLS-inferred peak current error is -37% (-36%). However, there is a strong positive correlation and the correlation coefficient is 0.93. The arithmetic mean (median) value of the absolute value of the LLS-inferred peak current error is reduced to 15% (12%) when the ratio of 65% is used to correct them. Among 233 return strokes of triggered lightning flashes, 16 return strokes are mistakenly classified as intra-cloud lightning, so the return stroke classification accuracy of GHMLLS is 93%. The peak currents of these mistakenly classified return strokes are lower, the stations available for locating are fewer, and the errors of location and LLS-inferred peak current are larger.In conclusion, GHMLLS have good detection efficiency and location accuracy. The return stroke classification accuracy of GHMLLS is at a high level as well. Nevertheless, there is an obvious systematic deviation in the LLS-inferred peak current of GHMLLS. In order to obtain more reliable analysis results, it's recommended to divide it by 0.65 when using the LLS-inferred peak current of GHMLLS.
High-resolution Model for Seasonal Prediction of Surface Shortwave Radiation in China
Liu Bo, Ma Libin, Rong Xinyao, Su Jingzhi, Yan Yuhan, Hua Lijuan, Tang Yanli
2022, 33(3): 341-352. DOI: 10.11898/1001-7313.20220308
Based on the global high-resolution climate model CAMS-CSM developed by Chinese Academy of Meteorological Sciences, the seasonal prediction skill of downward short-wave radiation flux (DSWRF) in China and three key regions is evaluated during the period of 2011-2020. The results show that the high-resolution version of CAMS-CSM can well predict the seasonal and interannual variability of DSWRF, but the predicted intensity is relatively weaker in spring and summer, while slightly stronger in autumn and winter compared to the observation. The prediction of the climate mean state doesn't change much with the lead time, indicating the systematic bias of the DSWRF is formed steadily in the early stage of model integration. However, there are obvious diversities in the prediction skill of the DSWRF anomalies in different seasons and different regions. From the anomalous spatial and temporal correlation coefficients, it can be noted that the prediction skill is higher in Inner Mongolia and Northwest China in autumn and winter, while lower in some areas of Beijing-Tianjin-Hebei in summer and autumn. From the perspective of comprehensive assessment of trend anomalies (P index), the model can score more than 70 points for all seasons in China at 0-month lead time, and the best performance can be close to 80 points for summer and autumn in Northwest China. Overall, the high-resolution version of CAMS-CSM climate model has certain prediction capability for DSWRF at 0-1 month ahead in China, especially in northwest regions where the solar-radiation is rich all year, which can provide specific scientific guidance for the future DSWRF short-term prediction and the solar energy site selection. In addition to the systematic bias of the model, there is a significant negative correlation between the predicted DSWRF bias and the total cloud cover bias, indicating that the bias of DSWRF prediction mainly comes from the simulation bias of total cloud cover, especially in spring and summer, as well as in autumn and winter in South China. In order to improve the prediction accuracy of DSWRF, it is an effective way to reduce the uncertainty of the model cloud microphysical processes. However, it is difficult to meet the demand of practical application with only high-resolution climate model, and its results still need to be processed with methods such as dynamic downscaling and bias revision to further improve the prediction skills.
Monitoring Characteristics of Hydrogen and Oxygen Isotopes in Precipitation of Nanjing
Zhu Xuan, Xiao Wei, Wang Jingyuan, Chu Haoran, Hu Yongbo, Xie Chengyu, Zheng Youfei
2022, 33(3): 353-363. DOI: 10.11898/1001-7313.20220309
With the climate change, extreme precipitation events become more and more frequent. It is particularly important to explore the characteristics of different precipitation types for the study of local precipitation process and water cycle. In order to explore the isotopic characteristics of different precipitation types, especially the relationship between the isotopic characteristics of tropical cyclone precipitation and its moving path, the characteristics of precipitation in Nanjing from July 2018 to June 2019 are analyzed based on the precipitation isotopic composition data and the meteorological data from China Meteorological Administration. The results indicate that, in general, the stable hydrogen and oxygen isotopic compositions of precipitation in Nanjing are more depleted in monsoon wet season and more enriched in non-monsoon wet season. After classifying the precipitation types, it is found that the intensity of tropical cyclone and Meiyu precipitation is high, and the hydrogen and oxygen isotopic compositions are seriously depleted; while the intensity of other precipitation are relatively weak, and the hydrogen and oxygen isotopic compositions are relatively rich. The deuterium excess value of tropical cyclone precipitation is usually less than the global average (10‰), which may be caused by the fast speed and short path of tropical cyclones from the eastern sea surface. The deuterium excess value of Meiyu precipitation is slightly higher than the global average, which may be influenced by persistent stationary fronts, the long and slow water vapor transport distance, and strong land evaporation. Other precipitation deuterium excesses are much larger than the global average, and this may be due to the more complex effects of land surface evaporation. Detailed analysis also indicates that, among the tropical cyclone influencing the precipitation of Nanjing, those directly landing in China from the ocean lead to lower deuterium excess values from 7.5‰ to 8.6‰. But for those landing on the southern Japan first and then in China, like Typhoon Jongdari (1812), the deuterium excess value of precipitation is much greater than 10‰, indicating that it is significantly affected by the land.
A Daily Meteorological Impact Index of Maize Yield Based on Weather Elements
Liu Wei, Song Yingbo
2022, 33(3): 364-374. DOI: 10.11898/1001-7313.20220310
Ten-day and monthly climate suitability has been widely used in agrometeorological research and operation, but it will underestimate the impact of short-term meteorological disasters on crops. In order to dynamically reflect the effect of meteorological conditions on crop yields, the daily precipitation suitability is optimized by calculating a weighed 10-day mean of daily precipitation including previous 9 days. Daily climate suitability is constructed consideringthe suitability of temperature, sunshine and precipitation. The correlation coefficient and Euclidean distance of daily climate suitability before the forecast period is used to identify three similar years and the comprehensive similar year. The whole growth climate suitability sequence of crop is established based on daily climate suitability before the forecast period and daily climate suitability in the similar years after the forecast period. And the climate suitability index is integrated from daily climate suitability sequence. The yield forecast model is established by using crop meteorological yields and the daily climate suitability. Daily crop meteorological yield impact index is designed to indicate the effect of meteorological conditions on crop yields. A daily yield forecast model is constructed to analyze the accuracy of daily yield forecast in the main maize-producing provinces in Northeast China and to indicate the accuracy of the meteorological impact index on crop yield. The results show that the use of comprehensive similar years can improve the accuracy of forecasts. The interannual fluctuation of daily forecast accuracy in Heilongjiang smaller than that in the other three provinces. The forecast accuracy is the lowest in Liaoning. Under the comprehensive similar years in monthly time scale, the advancement of the maize fertility process and the access of real-time meteorological data will improve the accuracy of monthly average forecasts. The accuracy on 31 August is generally higher that on 31 July. The daily forecast can provide a reference for yield forecast. The daily scale forecast in Liaoning varies greatly.Daily forecast yield and the announced yield are gradually approaching with the advancement of the maize fertility process and the accuracy of daily forecast yield is also improved. The impact index based on daily meteorological data can quantitatively assess the effect of meteorological conditions on crop yields at different time scales. To a certain extent, the daily meteorological impact index on crop yield can improve the quantitative evaluation level of agrometeorology operation.
Modification of Leaf Water Content for the Photosynthetic and Biochemical Mechanism Model of C4 Plant
Feng Xiaoyu, Zhou Guangsheng
2022, 33(3): 375-384. DOI: 10.11898/1001-7313.20220311
The accurate simulation of leaf photosynthesis is of great significance to the study of terrestrial ecosystem model and understanding the impact of global change on vegetation. To improve the description of photosynthesis from phenomenon to mechanism, empirical model is gradually replaced by photosynthetic biochemical mechanism model, in which the photosynthetic biochemical mechanism model proposed by Farquhar is widely recognized and used. The effect of CO2 concentration on plant photosynthesis is considered by the model, but the response of plant photosynthetic parameters to temperature and light intensity is studied without considering water stress. Water is one of the important raw materials of photosynthesis, which directly affects leaf stomatal conductance, transpiration rate and photosynthetic rate, and then affects plant photosynthesis. Therefore, many experiments are carried out and the water response function is gradually established. However, these studies mostly focus on soil water content rather than leaf water content that directly affects photosynthesis, limiting the accurate simulation of photosynthesis.Taking maize from North China as the research object, drought simulation data are studied based on six water gradient tests which are carried out at Gucheng Ecological and Agro-meteorological Experimental Station of Chinese Academy of Meteorological Sciences from June to October in 2014. Different from the previous empirical model of photosynthesis, a C4 plant photosynthetic biochemical mechanism model developed from the biochemical mechanism model proposed by Farquhar and modified by von Caemmerer is applied. The sampling blades and the environmental factors such as temperature, CO2, relative humidity, and light intensity are kept consistent, and the temperature difference between different observation times are adjusted to quantitatively study the relationship between leaf water content and maximum carboxylation rate accurately. The results show that the relationship between them can be expressed in a quadratic curve significantly (passing the test of 0.01 level), and the determination coefficient of the fitting equation is up to 0.88. With different parameters, the values of the maximum carboxylation rate are different, but the normalized leaf water content correction function is independent of the parameters. Through calculation, when the leaf water content is about 80%, the value of correction function is 1, and when the leaf water content drops to about 70%, the value is 0. This result perfects the photosynthetic biochemical mechanism model of C4 plant from the perspective of leaf water content, which provides a reference for further improving the accuracy of photosynthesis simulation, drought monitoring and early warning of maize.