Vol.31, NO.4, 2020

Display Method:
Experiments on Improving Temperature and Humidity Profile Retrieval for Ground-based Microwave Radiometer
Zhang Xuefen, Wang Zhicheng, Mao Jiajia, Wang Zhangwei, Zhang Dongming, Tao Fa
2020, 31(4): 385-396. DOI: 10.11898/1001-7313.20200401
Ground-based microwave radiometer (MWR) has a crucial role in scientific researches, weather modification service and climate change studies. MWR adopts passive remote sensing technology which has smaller volume, lower power consumption. Ground-based microwave radiometer detects atmospheric temperature and humidity by receiving atmospheric microwave radiation, which can conduct 24-hour unattended, high-resolution observation. It can detect short-time variation of atmospheric elements. Many studies show that different seasons, different weather conditions, quality control algorithms, and changes in environments have certain effects on retrieval results of MWR. In the case of cloudy condition, the uncertainty of cloud absorption coefficient leads to the increase of retrieval error and incorrect data of MWR especially. In order to improve temperature and relative humidity detection capabilities of MWR, the experiment builds BP neural network algorithm with 6 years(from 2011 to 2016) sounding data. The experiment builds two types of retrieval methods because there are some differences of microwave radiation transfer between clear and cloudy samples. The test uses measured brightness temperature data (requiring correction) and cloud data of millimeter-wavelength cloud radar as model inputs and then uses sounding data to evaluate model outputs (temperature and relative humidity profiles) from 2017 to 2018.Results show that correlation coefficients between outputs of 4 models (clear sky sample temperature model, cloudy sample temperature model, clear sky sample relative humidity model and the cloudy sample relative humidity model) and sounding data are 0.99, 0.99, 0.80 and 0.78. Taking sounding profiles as reference, root mean square errors (RMSE) of retrieval results of 4 models are 2.3℃, 2.3℃, 9%, 16%. Comparing with the MWR original profiles, RMSEs of 4 models are reduced by 0.4℃, 0.3℃, 11% and 9%, accuracies are improved by about 30%, 28%, 64% and 45%. In particular, the deviation of temperature model and humidity model within ±2℃ and ±20% account for 68%, 70% and 95%, 78%, which are 7%, 5% and 27%, 23% higher than MWR original profiles. The bias correction of brightness temperature and the training retrieval model of distinguishing weather samples are helpful to improve the retrieval accuracy of MWR temperature and humidity profiles. The network model combined with cloud radar information has obviously better effects on the retrieval result under cloudy samples.Through these experiments, the quality control of brightness temperature, combination of active and passive retrieval algorithms are well improved. The combination of active and passive retrieval effectively improves the performance of MWR, which will lay a foundation for the development of the comprehensive observation system of atmospheric profiles.
Model Construction of Rainfall Interception by Maize Canopy
Guo Jianping, Luan Qing, Wang Jingxuan, Zhang Limin
2020, 31(4): 397-404. DOI: 10.11898/1001-7313.20200402
Rainfall is the main water source of crops. Crops maintain their normal growth and development by absorbing water from the soil. However, the role of rainfall is often overestimated in water resource evaluation and farmland water balance research because the rainfall interception effect of crop canopy is not considered. It is difficult to calculate crop interception quantitatively, which seriously restricts the impact assessment of rainfall on crops. Therefore, in order to determine the interception effect in different growth stages of maize under different rainfall, the rainfall interception experiment of maize is carried out at Jinzhou Agricultural Meteorological Experimental Station of Liaoning Province in 2018. A total of 10 rainfall levels (rainfall over 20 mm is designed for the measurement of saturated interception) and 8 leaf area indexes (representing 8 different growth periods) are examined in the experiment. The rainfall interception effect of maize canopy is systematically analyzed by simulation. Results show that under certain rainfall, the relationship between the interception of maize canopy and leaf area index conforms to the relationship of quadratic polynomial, exponential function and power function, among which the quadratic polynomial has the highest explanation rate. Under the assumption of fixed leaf area index, the rainfall interception of maize canopy is in accordance with the quadratic polynomial, exponential function, power function and logarithmic function. When leaf area index is above 3, the explanation rate of power function is the highest, and the relationship between saturated interception of maize and leaf area index is in accordance with the quadratic polynomial, exponential function and power function, among which the explanation rate of quadratic polynomial is the highest. The comprehensive leaf area index and rainfall analysis indicate a positive correlation between canopy interception and the square of leaf area index and the logarithm of rainfall. According to the traditional planting mode of maize in China, the maximum leaf area index of high-yield maize is generally about 5-6. Therefore, the maximum interception of a rainfall process is usually 1.5-2.3 mm. When leaf area index is less than 1, the rainfall interception of maize can be ignored. Results are of practical significance for the evaluation of the effectiveness of rainfall resources and the study of farmland water balance.
Spatial-temporal Distribution of Apples with Different Drought Levels in Northern China
Cheng Xue, Sun Shuang, Zhang Zhentao, Zhang Fangliang, Liu Zhijuan, Wang Peijuan, Huo Zhiguo, Yang Xiaoguang
2020, 31(4): 405-416. DOI: 10.11898/1001-7313.20200403
Taking the major apple producing provinces in northern China as the research area, spatial-temporal distribution characteristics of different levels of apple drought during different growing periods are quantitatively evaluated. The drought index is graded, and historical disaster data collected are used to count disaster samples of different drought levels during different growth stages of apple, which verifies the rationality of the drought index level. Then the spatial-temporal distribution of different levels of drought, drought occurrence frequency, and occurrence range of apples during different eras in northern China are analyzed based on the validated drought grade index. Results show that among 46 total samples, the fully matched and basically matched samples account for 85%. Therefore, the drought grade index constructed can reflect the actual drought conditions reasonably for apples in northern China. Spatial distribution characteristics of drought levels decrease sequentially from north to south. The severe drought during different growing periods for apples is mainly distributed in northern and central of Gansu and northern Ningxia. The degree of drought changes greatly with the change of years in north central Shanxi from fruit tree sprouts to flower buds sprout, in Shanxi, Shandong, north Shaanxi from flower buds sprout to flower full bloom, and in Henan from mature to fallen leaves. The area of severe drought gradually increases with changes of the years from 1981 to 2010 in fruit tree sprouts to flower buds sprout and mature to fallen leaves. The frequency of no drought increases from north to south and the high frequency areas occur one time or above in two years. The frequency of severe drought decreases from north to south and the high frequency areas occur one time or above in two years. The light drought has dominated during the past 36 years in the study area, and the range of severe drought during the period from flower to maturity is larger than the other period. The ratio of drought occurring stations show a significant increasing trend for severe drought in the period from fruit tree sprouts to flower buds sprout and the moderate drought in the period from flower to maturity. The drought is severe and the frequency is high in the northwest. In the production of apples, a drought prevention plan should be prepared on the basis of drought warning, and attention should be paid to the timely response to drought at different growth stages.
Experimental Research on Frost Indexes for Lycium Barbarum Flowering Phase
Duan Xiaofeng, Zhu Yongning, Zhang Lei, Wang Jing, Xu Rui, Wang Jingmei
2020, 31(4): 417-426. DOI: 10.11898/1001-7313.20200404
Based on researches of frost indexes of Lycium barbarum during 2016-2018, tests of frost index are performed using an artificial frost box in 2019, focusing on flower bud, early flowering and full flowering periods. The temperature of treatment is -8—-1℃, and time duration is 1-6 h, which forms 32 combinations of low temperature and time. Frozen symptoms of Lycium barbarum are divided into 3 degrees (P1, P2, P3) according to observation after freezing. The mild frost is defined as 0≤P1≤50%, 0≤P2 < 50% and 0≤P3 < 10% (P1, P2, P3 not to be zero at the same time). The medium frost is defined as 0≤P1 < 50%, P2≥50% and 0≤P3 < 10%, or 0≤P1 < 50%, 0≤P2 < 50% and 10%≤P3 < 50%. The severe frost is defined as 0≤P1 < 50%, 0≤P2 < 50% and P3≥50%. P2 and P3 are decisive for confirming frost levels. The frozen rate of each frozen degree is counted and the judgment standard of frost are formulated. In terms of these studies, frost indexes are established based on low temperature, time of duration, cold level, and rate of cold. Frost indexes are verified by using results of field experiments at 13 sites and investigations into natural frost disaster at 25 Lycium barbarum planting regions. Results indicate that frosts of Lycium barbarum are closely related to the low temperature and its duration. Lower temperature and longer duration lead to more seriously frost disasters. The frost resistance of flower bud period is the strongest, which is significantly stronger than that of early flowering period. The frost resistance of full flowering period is weaker than early flowering to some degree. The result of judgment with frost indexes and the situation of actual disaster have high compliance, coincidence rates are 92.9%, 81.8% and 91.7%, respectively. The indexes have strong practicability which can be used as basics for forecast and assessment of frost damage at different stages of Lycium barbarum flowering phase.
Comprehensive Climatic Index and Grade Classification of Cold Damage for Taiwan Green Jujube in Fujian
Yang Kai, Chen Binbin, Chen Hui, Chen Fuzi, Yang Xiqiong, Li Lirong, Li Lichun, Lin Jing
2020, 31(4): 427-434. DOI: 10.11898/1001-7313.20200405
Taiwan green jujube is widely planted in Fujian, Guangdong, Guangxi and Hainan because of its high planting efficiency. However, these areas locate in the northern edge of planting areas of tropical fruit trees, therefore blindly planting may aggravate the loss of cold damage. In order to achieve a reasonable distribution and reduce the loss of cold damage, it is urgent to establish objective and quantitative comprehensive climatic index for cold damage of Taiwan green jujube in these areas.The classification standard of cold damage symptoms is established by interview of experts and field disaster investigation. According to field observation and results of geographical transplantation experiment, combined with the investigation of cold damage from December 2014 to February 2015 and December 2015 to February 2016, 32 samples of cold damage are obtained, and the critical temperature of cold damage for Taiwan green jujube is determined to be 5.0℃. Referring to each process of the cold damage examples, the disaster grade of each cold damage example is obtained by comparing the classification standard of cold damage symptoms, and the correlation between meteorological factors and disaster grade is analyzed. The disaster-inducing factors for cold damage of Taiwan green jujube are determined, including the extreme minimum temperature, sustained days of cold damage process with the temperature below 5.0℃, the process of harmful cold accumulation for the extreme minimum daily temperature below 5.0℃, and the cooling range of cold damage process for the temperature below 5.0℃. After normalizing each factor, principal component analysis is used to simplify four disaster-inducing factors, and the comprehensive climatic index of cold damage for Taiwan green jujube is constructed. The formula is Ih=-1.255X1+1.688X2+0.53X3+2.52X4. K-means cluster analysis method is used to determine the index classification combined with the cold damage grade. The cold damage is graded with Ih as follows:0.02≤Ih < 0.72, light; 0.72≤Ih < 1.76, moderate; 1.76≤Ih < 2.72, severe; Ih≥2.72, most severe. According to the method of comparative validation for typical years, the comprehensive climatic grade index of cold damage for Taiwan green jujube is in good agreement with the actual situation, which verifies the reliability of the index. Results have practical reference value for the evaluation of cold damage, introduction and expansion of Taiwan green jujube.
Comparison of Characteristics and Environmental Factors of Thunderstorm Gales over the Sichuan-Tibet Region
Wang Hong, Li Ying, Song Lili, Shen Yun
2020, 31(4): 435-446. DOI: 10.11898/1001-7313.20200406
Characteristics, environmental factors and synoptic situations of thunderstorm gales over the Sichuan-Tibet Region from 2010 to 2017 are analyzed based on significant weather report, surface observations and sounding data from China Meteorological Administration and ERA-Interim reanalysis data from European Centre for Medium-Range Weather Forecasts(ECMWF). Distinct properties are revealed through comparison of characteristics and environmental parameters of thunderstorm gales over highland(1 km above sea level) and basin(1 km below sea level). Results show that thunderstorm gales occur over the highland during a full year except winter, with two peaks in May-June and September, respectively. Their diurnal variation shows a major peak at 2000 BT. However, thunderstorm gales over the basin are active both in the afternoon and in the evening mainly in summer. The annual station-averaged frequency of thunderstorm gales over the highland is about 2 times per station, proportions of which to thunderstorms and gales are about 4.5% and 8%, respectively. It is only 0.4 times per station for thunderstorm gales over the basin, which account for 1.5% of the thunderstorms but 60% of gales. The atmospheric water vapor content, convective available potential energy and downdraft convective potential energy over the highland are significantly lower than those over the basin. The mean vertical temperature lapse rate in the middle and lower troposphere over the highland is larger than that over the basin. Usually, there is a shallow moist layer in the middle troposphere overlaid on a drier air layer over the highland. However, there is usually significant dry air in the middle troposphere and a moist layer at low level over the basin. Synoptic situations of thunderstorm gales over the Sichuan-Tibet Region are composited during two peaks in May-June and September, respectively. During May and June, the vertical wind shear of the environment is strong, with the middle level affected by a westerly trough transporting weak cold advection at 500 hPa, and the upper level located on the right side of a jet entrance at 200 hPa. However, in September, the middle level over the Sichuan-Tibet Region is at the north edge of subtropical high pressure at 500 hPa, with significant dry air in the mid-upper troposphere and remarkable warm moist air flow at low level. Though synoptic situations are different in two seasons, both of them can provide favorable condition to the formation of thunderstorm gales.
Spatio-temporal Characteristics of Boundary Layer Height Derived from Soundings
Liang Zhihao, Wang Donghai, Liang Zhaoming
2020, 31(4): 447-459. DOI: 10.11898/1001-7313.20200407
Using K-means cluster method, the whole country is divided into four regions (Qinghai-Tibet region, northwest region, central region and eastern region) by boundary layer height (BLH) derived from potential temperature gradient method based on L-band radar sounding secondly data of 119 stations from January 2010 to December 2018. Characteristics of BLH and frequency of different boundary layer state are investigated, including convective boundary layer (CBL), neutral boundary layer (NBL) and stable boundary layer (SBL), through their interannual, annual and diurnal variations respectively. Results show that there is no significant difference in the annual average BLH and the frequencies of different boundary layer states in four regions at 0800 BT and 2000 BT from 2010 to 2018. At 0800 BT, the annual average BLH is around 200-600 m and mainly in SBL. At 2000 BT, the annual average BLH in Qinghai-Tibet region is the highest (about 1500 m), followed by northwest region and central region (about 1000 m and 500 m), and that of eastern region is the lowest (about 400 m). Qinghai-Tibet region and northwest region are mainly with CBL and NBL, while central region and eastern region are mainly with NBL. Besides, the annual variation of BLH in four regions is similar at 0800 BT, but it's significantly different at 2000 BT. At 0800 BT, the difference of one-year monthly average BLH in four regions are not obvious, and there is no clear difference among these regions. But at 2000 BT, the monthly average BLH in each region reaches maximum in spring and summer, and minimum in autumn and winter. As for corresponding annual variation of different boundary layer state frequencies, SBL's frequency first increases then decreases while the frequencies of CBL and NBL first decrease then increase overall at 0800 BT and 2000 BT. And their turning point is in May to July. In general, the variation range of monthly average BLH and boundary layer state frequencies gradually descend from Qinghai-Tibet region, northwest region, central region to eastern region. The diurnal variations of BLH are different in four regions. In particular, the diurnal variations of Qinghai-Tibet region, northwestern region and central region show distinct seasonal difference. In Qinghai-Tibet region, the amplitude of diurnal variation can reach 2000-2300 m in spring and summer, but relatively weaker in autumn and winter. The diurnal variation in eastern region is similar in all seasons, and amplitudes are around 600 m.
Radar Echo Characteristics and Recognition of Warm-sector Torrential Rain in Sichuan Basin
Luo Hui, Xiao Dixiang, Kuang Qiuming, Qing Quan, Kang Lan
2020, 31(4): 460-470. DOI: 10.11898/1001-7313.20200408
Based on data of real-time precipitation and weather radar during 28 torrential rain events in warm regions of Sichuan Basin, radar echo characteristics of torrential rain in early and mature stages are analyzed. Feature vectors for identifying early and mature stages of torrential rain in warm regions are constructed and the selected samples are studied by random forest machine learning method. According to the influence range, duration, and cumulative amount of precipitation, the thunderstorm group is the main part of the rainstorm in the warm region, and its development can be divided into three types.According to the burst of short-term heavy precipitation, thunderstorm groups are divided into primary and mature stages. In the early stage of the thunderstorm to the mature stage, the "in situ development type" is dominant, the "individual development type" and the "in front side trigger type" are the second. With the evolution after the maturity, the "in situ development type" and the "front side trigger type" are the main types. Convective precipitation is the main type of heavy rain in the warm area. After the first type of thunderstorm group, the combination of mature thunderstorms is the main source of thunderstorms, which moves slowly and is conducive to the generation of heavy precipitation. In front of mature thunderstorms, new thunderstorm cells are continuously generated and merged to continue spreading northward, forming a large range of precipitation. Individual development thunderstorm groups have the longest duration and a large influencing range. They are accompanied by long-term merge when moving. Among 28 processes, a large proportion appears in the northwest and has the longest duration. Echoes of these processes are in the southwest-northeast direction, which is basically consistent with the trend of the Longmen mountains in the western part of the Basin. The uplift of the topography (generating easterly wind) plays a key role in the occurrence and development of warm rainstorms. In the primary stage, the average core height and average top height of "in situ development type" thunderstorm group has a bimodal structure. Similar structure is found for "front side trigger type" in the mature stage. Multiple parameters of three types of thunderstorm groups show a unimodal distribution in the nascent and mature stages. To identify heavy rains in the warm area, feature vector is constructed using multiple parameters of the thunderstorm group, and random forest machine learning is also applied, leading to satisfying results.
Estimating the Inversion Accuracy of Atmospheric Temperature and Water Vapor Profile Under Limb Sounding
Zong Xuemei
2020, 31(4): 471-481. DOI: 10.11898/1001-7313.20200409
Profiles of atmospheric temperature and water vapor are important for studying atmospheric state and play an important role in the energy balance of earth-atmosphere system. Limb remote sensing is an important means to obtain the profile of atmospheric parameters. The atmospheric radiation ultra-high spectral detector developed by Shanghai Institute of Technical Physics, Chinese Academy of Sciences, has a detection band range of 650-3050 cm-1 and the spectral resolution on limb view is as high as 0.015 cm-1, which will be the highest spectral resolution that the world's Fourier spectral detector can achieve. A method by using information and weighting function linearization are proposed to evaluate the inversion accuracy of the research instrument in advance. Weighting functions of atmospheric temperature and water vapor at 16 different tangent points are simulated and calculated by RFM model. The degree of signal freedom and the entropy reduction are also calculated by the information content method, and the optimal number of inversion channels is determined to be 200 by the stepwise iterative algorithm. Combined with the threshold (0.3 K) of detectable brightness temperature and the linearized weighting function of the instrument, the available spectral channel numbers of atmospheric temperature and volume mixing ratio of water vapor profiles under different inversion accuracy of six atmosphere models (US standard atmosphere, tropical atmosphere, middle-latitude summer atmosphere, middle-latitude winter atmosphere, subarctic summer atmosphere, subarctic winter atmosphere) are calculated and analyzed, and the inversion accuracy is estimated theoretically. On the demanded optimal 200 channels, the inversion accuracy of the whole temperature profile is 0.6 K, but if the inversion accuracy of the temperature profile is required to be 0.5 K, the number of channels available for inversion at a higher tangent height is smaller. Except the tropical atmosphere model, there are enough channels for the other five atmosphere models meeting 5% accuracy demands of the inversion of water vapor volume mixing ratio profiles. However, the inversion of the water vapor profile of the tropical atmosphere has barely enough channels at 16-20 km for 10% relative inversion accuracy of volume mixing ratio. The number of channels usable for atmospheric parameters retrieving increases by the decreasing of inversion accuracy. Among six atmosphere models, the tropical atmosphere is relatively special and its inversion accuracy is lower, which may be related to the unique temperature profile of the tropical atmosphere. There is no isothermal layer in the tropical atmosphere, which may lead to fewer atmospheric parameter inversion channels near the height of sharp temperature transition.
Comparison of Solar Ultraviolet Irradiance Measurements at Zhongshan Station, Antarctica
Zheng Xiangdong, Cheng Haixuan
2020, 31(4): 482-493. DOI: 10.11898/1001-7313.20200410
A comparative analysis is presented on surface solar ultraviolet B (UVB) band and ultraviolet A (UVA) irradiance measured by 3 UV broadband pyranometers: FSUVA(315-400 nm), FSUVB(280-315 nm) provided by Jiangsu Radio Scientific Institute Co. LTD (CJRSI), American Yankee UVB(280-320 nm), and Brewer ozone spectrophotometer, placed at Zhongshan Station, Antarctica. Using data of Brewer ozone spectrophotometer in 2017 as a reference, results show that, for UVB(280-315 nm) irradiance, the error of FSUVB is averagely (55±75)% but its irradiance is lower during the "ozone hole" period, indicating that domestic made FSUVB broadband radiometer is less sensitive to the ozone layer thinning. Furthermore, the irradiance relative error of FSUVB shows a certain upward trend with the increase of total atmospheric ozone, indicating an over-measured UVB irradiance by the FSUVB pyranometer in regions with normal ozone concentration, such as the area of middle-low latitudes including China. As a contrast, the error of Yankee UVB (280-320 nm) is averagely (-31±22)% lower than that of Brewer measured, however, the relative error and total ozone variation are unrelated. The under-measured UVB irradiance from the Yankee UVB pyranometer is attributed to the system calibration error. Since Brewer spectral UV measurement is limited within 286.5-363 nm, a so-called UV correction factor, on the basis of empirical ratio of spectral irradiance at the wavelength longer than 363 nm to the Brewer measured irradiance at 363 nm, is applied to make up Brewer spectral irradiance gap of 363.5-400 nm with wavelength resolution of 0.5 nm for constructing Brewer entire spectral UVA (315-400 nm) irradiance. The error of FSUVA is averagely (23±59)% when Brewer UVA irradiance is used as the reference. With tropospheric Ultraviolet visible (TUV) radiation model calculations under cloud-free and the sun zenith angle (SZA)less than 80° as references, irradiance errors measured by FSUVB, Yankee UVB and FSUVA are (30±37)%, (-22±19)% and (27±6.4)%, respectively, in 197 cases, while the average of differences between Brewer and TUV calculations are respectively (3.4±8.5)% in UVB band of 286.5-315 nm, (2.9±6.8)% in UVB band of 286.5-320 nm and (3.4±4.5)% in UVA band of 315-400 nm, proving the method of Brewer UVA correction factor is rational. Again, only the relative error of FSUVB measurements referenced to TUV calculations displays an evident increasing trend with the growth of total ozone. Mechanisms of over-measured solar irradiances from both domestic made broad-band UV pyranometers are not fully identified. Calibration methods needed to be improved with consideration of variable SZA and total ozone. In addition, the stray light at longer wavelengths should have a significant influences on the pyranometer's performances and this stray light need to be eliminated in further instrument improvement. For FSUVB, its less sensitiveness to the solar UVB irradiance during the period of "ozone hole" still needs to be resolved.
Ensemble Learning for Bias Correction of Station Temperature Forecast Based on ECMWF Products
Chen Yuwen, Huang Xiaomeng, Li Yi, Chen Yue, Tsui Chi, Huang Xing
2020, 31(4): 494-503. DOI: 10.11898/1001-7313.20200411
To improve the accuracy of numerical weather prediction (NWP) and its ability for extreme weather event forecast, a hybrid model based on ensemble learning is proposed and tested by post-processing one of the most successfully predicted variables, temperature at 2 m height. The NWP dataset used is provided by The International Grand Global Ensemble (TIGGE) project in the European Centre from Medium-Range Weather Forecasts (ECMWF), with a horizontal resolution of 0.125°×0.125° and lead times from 6 to 168 h (with a 6 h increment, 28 lead times totally). The observation is collected from 301 stations covering China expect for Xizang and Qinghai, including 4 variables, temperature, pressure, relative humidity and wind speed every 3 hours. The ECMWF product and observation span a period of 6 years ranging from 1 January 2013 to 31 December 2018. Data from 2013 to 2017 are used for machine learning and model training, and data in 2018 are used for testing. The hybrid model named ALS consists of 2 stages. Stage 1 trains two separate models, a long short-term memory combined with a fully connected neural network (LSTM-FCN) and an artificial neural network (ANN). Stage 2 blends the output of LSTM-FCN and ANN using a linear regression (LR) model. The correction result is the output of LR. ALS model is then applied to correct the station temperature forecast with lead time from 6 to 168 h. Outcomes are verified by observations from stations, while LR model is used as control model. ALS model reduces the average root mean square error (RMSE) of the station temperature forecast by 0.61℃ (19.6%), and by 0.23℃ (8.4%) compared with the LR model. ALS model reduces RMSE at more stations compared with LR model (252 vs. 186). ALS model is particularly effective in areas where the accuracy of station temperature forecast is low, such as Guizhou and Yunnan. Forecasts for stations in these areas are significantly improved with an average RMSE reduction over 40%. Moreover, case analysis of high temperature show that ALS model improves the forecast accuracy of high temperature events significantly, with a RMSE reduction of 30.5% at 4 stations compared to station temperature forecast. It demonstrates that ensemble learning can be used to supplement weather forecast.
Bias Correction of Summer Extreme Precipitation Simulated by CWRF Model
Dong Xiaoyun, Yu Jinhua, Liang Xinzhong, Wang Chen
2020, 31(4): 504-512. DOI: 10.11898/1001-7313.20200412
The accurate forecast of extreme precipitation plays an important role in guiding the national economy and people's livelihood. The newly developed Climate-Weather Research and Forecasting model (CWRF) integrates a comprehensive ensemble of alterable parameterization schemes for each of the key physical processes, including surface (land, ocean), planetary boundary layer, cumulus (deep, shallow), microphysics, cloud, aerosol, and radiation. This facilitates the use of an optimized physics ensemble approach to improve weather and climate prediction. Evaluating the simulation performance and correcting the error can effectively improve the operational prediction level of extreme precipitation in CWRF model.Daily rainfall data simulated by CWRF model and observed at 2416 meteorological stations in China from June to August during 1980-2015 are used to compare correcting effects of Q-lin, Q-tri, RQ-lin, RQ-tri, SSP-lin and CDFt on extreme precipitation of control scheme simulated by CWRF in eastern China. Based on the simulation performance ranking of 14 parameterization schemes in CWRF model, effects of the top 4, the latter 4 and the ensemble of 14 parameterization schemes are compared. Correcting effects of two approaches are compared: Revising after the collection of members and revising before the collection of members. Main results show that the error of the extreme precipitation simulation of C1 scheme can be obviously reduced by using six error correction methods, among which the RQ-lin correction method is the best. Although there are great differences between parameterization schemes in the simulation of extreme precipitation index, CWRF model shows good ability for extreme precipitation index in eastern China. The first four parametric schemes with good extreme precipitation simulation ability are C13, C14, C12 and C1, while the C6, C4, C3 and C10 schemes perform worse, respectively. Different parameterization schemes are revised to ensure that it is the closest to the average value of observed extreme precipitation after each of 14 members of the parameterization scheme being revised. Results have important application value for improving outputs of model and improving its prediction ability.Error correction can only be used as a supplementary means to improve extreme precipitation prediction. The precision of model physical process and the improvement of model resolution are the key to improve extreme precipitation prediction.