Abstract:After a careful rethink on shortrange forecast of the northern Henan severe rainstorm on 9 July 2016, we found that there existed obvious alternate growing and merging of convective clouds during this rainstorm event involving interaction of mesosmall scale systems. As important guidance, all global models, including deterministic and ensemble ones, failed to provide effective forecasts or hints in advance for this event. High resolution regional models and rapid refresh systems performed better and could help forecasters to predict heavy or very heavy rainfall events in northern Henan 12 h or 6 h earlier. Reasons for the forecast failure of this severe rainstorm process are that forecasters relied too much on global models and were lack of experience on high resolution model application or confidence in products from high resolution models. Only more comprehensive trainings on high resolution model for forecasters before putting them into use could forecasters maximize their operational potential. Applying more probabilistic products gradually in QPF decision process would meet the trend of technological advance and could provide more effective supports for heavy or extreme rainfall forecasting.

Abstract:The capacity of global model in forecasting of warm sector heavy rainfall is still very limited at present, especially for warm sector rainfall in northern China, which is one of the most difficult problems in operational forecasting. A severe regional torrential rainfall case which consisted of both warm sector and frontal precipitation attacked Hebei Province and Tianjin during 1-2 July 2013. As for the severe rainstorm ［409 mm·(24 h)-1］ in the middle part of Hebei Province, the ratio of warm sector rainfall accounted for more than 60% of total precipitation. The heavy rainfall of this event was seriously underestimated in the subjective quantitative precipitation forecast in NMC (National Meteorological Centre) with obvious missing forecast on rainstorm and severe rainstorm. Moreover, none of the numerical weather prediction (NWP) models can provide enough useful information to weather forecasters, which further increased the forecasting difficulty and caused the failure of QPF (quantitative precipitation forecast). In this paper, operational numerical model forecast products, automatic weather station observation, conventional surface and radiosonde observation, and Doppler radar data are employed to investigate the cause of the failure forecast of this warm sector torrential rainfall case. The results show that failure of capturing the subsynoptic and mesoscale small disturbances, such as surface convergence line, gust front, cold pool, and mesoscale vortex, as well as their effects on heavy rainfall under a high temperature and high humidity conditions was one of the important factors leading to the underestimation of the rainfall intensity. Furthermore, insufficient fine analysis on the environment condition in triggering mesoscale convection systems like lowlevel jet and the evolution of jet core also played important roles. Therefore, in terms of forecasting the warm sector rainfall within a very thick warm wet air mass, the triggering caused by interaction of surface convergence line, cold pool and mesoscale vortex and organizational development of local and convective precipitation in the high temperature and high humidity environment must be considered. Shorttime forecast and nowcasting based on surface automatic weather station observation and weather radar data could effectively compensate the lack of ability in finding mesoscale and smallscale systems by global NWP models, which may improve the accuracy of forecasting warm sector rainfalls.

Abstract:The public weather forecasts were not so correct for the time period from winter 2011 to early spring 2012 in Beijing. Of the nine weak precipitation processes (WPPs), four cases were false alarms, two were missed and three were better forecasted. Through the forecast verification, synoptic analysis and comparison of model forecasts for the nine processes, following results are achieved. The synoptic conditions of WPPs in Beijing vary case by case. The weather patterns at 500 hPa can be classified into three categories: patterns of twotroughs oneridge, onetrough tworidges and onetrough oneridge. The surface situation can be classified into four patterns, which are cold front, occluded front in North China, the easterly wind and inverted trough and the eastward returning current. Most WPPs are characterized by poor vapor conditions in lower troposphere or weak dynamic lifting conditions. For the WPP with thinner wet layer, higher vapor saturation layer and higher lifting condensation layer, numerical prediction models tend to give false alarm; for the WPP with favorable vapor condition at low level, lower vapor saturation layer and lower lifting condensation layer, the models are prone to omission. The subjective precipitation forecast errors are mainly because forecasters did not sufficiently understand the structures of the weather systems and their developing mechanisms. Besides, they were lack of correction experiences for the model boundary water vapor and lifting conditions, which are critical factors for rainfalls.

Abstract:At about 16:00 BT 4 September 2016 the urban area of Hangzhou, West Lake and the surrounding area received a sudden convective shower, which caused great impacts on preparation related to “Hangzhou G20 Summit”. This paper analyzes the mechanism of the shower, and discusses limitations of shorttime forecasting and difficulties of nowcasting. The results indicate that the mainly focusing convection system at that time over Ningbo, eastern Hangzhou area, and Shaoxing did not directly affect Hangzhou City, and eastwardmoving synopticscale upper trough system did not affected this area either. However, under the weakly statically unstable conditions, the shower was produced by a shallow convective system which was triggered by an afternoon sea breeze front with a dry line in the northwest Hangzhou Bay. As this convective system had some characteristics such as small spatial scale, short life, rapid movement, low cloud top, weak reflectivity and intensive rainfall intensity, together with a large number of highlevel cirrus clouds which cannot effectively be discriminated from cumulus over Hangzhou and the surrounding area in the satellite images, the ability of weather radar and stationary meteorological satellite monitoring the system was weakened significantly, so that nowcasting the shower is also very difficult if only using these two types of observations. Therefore, it is crucial to use automatic weather station data with high spatiotemporal resolutions to analyze the environmental conditions for nowcasting the type of shallow convective weather systems.

Abstract:Errors occur for mediumrange forecast of heavy rainfall over North China on 20 July 2016 by subjective forecast of National Meteorological Centre and numerical models. The mediumrange forecast performance of this heavy rainfall event is analyzed based on ECMWF ensemble forecast and T639 ensemble forecast data. The results reveal that: (1) there are obvious changes of location and intensity of rainfall of EC ensemble forecast which is caused by change of forecast for largescale weather system before and after 16 July. Compared to forecast initialized at 20:00 15 July, the ensemble mean circulation on 20 July initialized at 20:00 BT 16 July show that trough at 500 hPa over Hetao Region is deeper whereas downstream ridge is stronger, thus forming the pattern of slowly moving “high pressure to the east and low pressure to the west”. The Yellow River cyclone and midlower level southwest vortex move northward and become stronger. The forecast heavy rainfall is closer to the observed precipitation accordingly. (2) Ensemble sensitivity analysis shows that sensitivity areas are consistent with the spatial pattern of circulation difference before and after the forecast change. The largescale weather systems leading to the change of forecast of heavy rainfall are also sources of uncertainty of mediumrange rainfall forecast. (3) The most sensitive circulation system to precipitation forecast is the Yellow River cyclone. The Yellow River cyclone, the ensemble member is westward and northward and intensity is deeper, the corresponding forecast precipitation amount over BejingTianjinHebei Region is larger.

Abstract:Based on conventional surface and upper air observation data, the synopticscale circulation of the maximum 2 m temperature forecast failure in Shanghai during 17-18 March 2015 is analyzed. Using the realtime observations and numerical weather predictions, the causes of the forecast failure are investigated. The results show that the failure of the cloud cover forecast on 17 March was the main cause for the forecast failure. Meanwhile, the enhanced southeast winds further enlarged the forecast error. On 18 March, the misforecasted occurrence time of cold air advection into Shanghai mainly accounted for the forecast failure. The problems and recommendations in the forecast process were addressed, with the emphasis on the discrepancies between the realtime observations and the forecasts, forecast jumpiness and forecast uncertainty. Forecasters should have paid more attention to the realtime local and upstream observations, and different leadtime forecasts from different models. More important, operational ensemble forecasts in a probabilistic sense were highly recommended instead of traditional deterministic forecasts of a single model. Therefore, it is necessary to systematically assess the ensemble forecast performance, develop new application and interpretation of NWP products, and provide more products of multimodel ensemble forecasts with uncertainty information in public weather services.

Abstract:Conventional observations, encryption automatic weather station, Doppler weather radar, windprofiling radar and NCEP reanalysis data are identified and examined to analyze the extensive severe convection falling area and lifting trigger mechanism in Shandong on 11 September 2016. The results show that under the influence of upper trough, unstable atmospheric stratification occurred over the regions whether or not convections, severe convective cloud cluster spread everywhere in the environments characterized by large convective available potential energy (CAPE) and little convective inhibition (CIN), thus the trigger of lifting become the key factor for severe convection’s occurrence. Lifting trigger mechanism was organized by surface convergence line, dry line, seabreeze front and gust flow. Because of the little convective inhibition, lifting force could be relatively weak causing various thunderstorm lifting mechanisms in different regions. Surface convergence line lifting caused the severe convection in Northwest Shandong, and the combination of seabreeze front and cold front caused the severe convection in Shandong Peninsula. Gust front of the preexisting thunderstorm cold pool boundary was the reason of severe convection in midland Shandong, while the interaction of dry line and surface convergence line caused the severe convection in Southeast Shandong. The magnitude of convergence line lifting force which can be measured by boundary’s divergence was a consequential element. Under the condition of the surface convergence line, mesoscale boundary of mass in different temperature and humidity became the determining factor of thunderstorm trigger. The omissive forecast of the southeast lowlevel flow in shortterm forecast which was a key mesoscale system caused the deviation of the forecast falling area. So, adjustment trend of model forecast in different initial times and seasons are the major reasons for the less intense forecast. Experience indicates that improving the correction ability of numerical model by analyzing a large number of cases is an effective method to raise the forecasting accuracy.

Abstract:By using conventional and nonconventional meteorological data, numerical prediction model, GFS reanalysis data (0.5°×0.5°), the environment features and model errors of extremely heavy rain under northwest flow which happened on 19 June 2016 in northern Jiangxi are analyzed. Compared with the similar situation, some possible causes of the prediction error are found out, such as the deviation of southward northwest dryandcold air, the speed of the subtropical high, the strengthening and convergence of the southwest lowleveljet, shortwave trough from upstream and abnormal moisture conditions. Even more, the prediction of rain belt by ECMWF was northward and much lighter than observation. Finally the forecasting methods of severe heavy rain are given.

Abstract:In midnight on 31 July 2015, an extreme flashrain event was happened in Taihang Mountain. The rainfall was more than 100 mm at 5 automatic weather stations (AWSs) in mountain areas of Shijiazhuang and Xingtai, of which Yuantou town of Shijiazhuang City and Nanzhonggao Town of Xingtai City were caught by extreme shorttime intense rainfall with rainfall amount more than 50 mm·h-1 and more than 100 mm·(3 h)-1. However, the numerical prediction, the superior guidance forecast and meteorological observatory all failed to forecast this rainstorm. Based on conventional upperlevel and surface observations, AWS data, Doppler weather radar data and numerical forecast test, this paper analyzed the prediction ideas and the reason of failure. The event occurred under weak synopticscale background in northeast of Tibetan high. The key of forecasting severe precipitation was to grasp the effects of east wind in Taihang Mountain and terrain. The middlelevel northwest flow strengthened vertical wind shear, lowlevel shear line moved eastward, southwest wind strengthened warm moist air flows, and dry air layer overlaid on the thicken wet layer, strengthening the convective instability. The primary reason of the forecast error is that the lack of analyzing the maximum CAPE led to underestimating the unstable conditions and the development of thunderstorm at night in Shanxi Province. In addition, forecasters failed to use the unconventional data for realtime testing and correct numerical results, thus failing to predict the enhancement of thunderstorm enhance when it went down the hill. There is no effective conceptual model for severe precipitation conditions in the weak weather background at present. Forecast capacity of numerical model for such weather is poor. So forecasters need to improve the comprehensive analysis ability on observational data and the interpretation capacity of numerical prediction products in the future, and also to develop effective forecasting conceptual model through studying a large number of cases.

Abstract:A sudden rainstorm process happened in Tianjin on 4 September 2015. The analysis based on various observation data showed that the southerly flow met with the northeast wind at the bottom of the high pressure, forming the convergence in the mountain front and triggering thunderstorm. After the thunderstorm was formed, the “backward propagation” mechanism formed, matching up the southerly wind in the midlow level, and forming the southnorth “train effect” which led to the happening of heavy rainfall. The precipitation caused by upper trough developed into the cool pool near the surface layer. The divergence flow met the easterly flow, triggering the new thunderstorm in the downstream of precipitation area. This “forward propagation mechanism” made the rain belt move faster than that of upper trough. And when the divergence flow met with the easterly flow, smallscale convergence cyclonic circulation formed, making the precipitation increase in front of the convergence line. After the uppertrough precipitation moved across the urban area, due to the mesoγ scale convergence effect, which triggered the formation of mesoγ scale convective cloud, the heavy precipitation once again appeared in the urban area of Tianjin City. During the shortterm forecasting, besides the 〖JP2〗analysis of largescale environmental background,〖JP〗 forecasters only considered the precipitation field from mesoscale model, ignoring the analysis of mesoscale environmental field. The results showed that although there is deviation between the precipitation from model and observed precipitation, the prediction of the mesoscale environmental field could provide reference for the prediction of sudden heavy rainfall in the short period of time (24 h).

Abstract:Using NCEP-FNL data, we analyzed a continuous heavy rain process in Hangzhou Bay, and found that the constant intersection of cold and warm flows in Zhejiang Province caused the continuous rainfall. The increase of lowlevel cold air and southeast warm and humid air resulted in the development of baroclinic atmosphere disturbance, the enhancement of rainfall in the frontal area and the occurrence of heavy rains. Due to the deviation of the numerical model forecast, a large error occurred in the 24 h forecast of heavy rainfall in Hangzhou Bay during this process. The test results showed that the GFS model is better for the prediction of the cold and warm system in 36 h lead time, but has an error for the wind forecast in the low-level front, which affected the convergence, horizontal frontogenesis and the transportation of water vapor in the lowmiddle layer, leading to the deviation of falling area of heavy rain and rainstorm. In addition, it is also an important factor in the rainstorm forecast error that the model is weak in forecasting cold air in the south bank of Hangzhou Bay. Therefore, for rainfall stability forecast, forecasters need to attach importance to model adjustment in the prediction at adjacent time, and judge the influence of cold and warm air flow according to actual monitoring and the movement and evolution of rain belt, so as to timely adjust the rainfall zone and influence time and make an effective correction to forecasts.

Abstract:Based on the NECP/NCAR reanalysis data, highlevel and surface observation data and radar data, this paper analyzes the three stages of the heavy rainfall that occurred in Zhejing Province during 18-20 August 2014. The water vapor transmission, vertical movement and convective instability of the rainstorm are investigated from horizontal and vertical directions, so the lesson of a failure forecast could be summarized. The results are as follows. (1) The first stage of the heavy rainfall in northern Zhejiang was mainly caused by the cold front, but the existence of the tropical depression played a positive role in strengthening warm wet flow coming from southwest, making the north cold air meet with the warm wet flow, which induced the frontogenetic function in the deformation field. So, the neglected existence and influence of the tropical depression are the main inaccurate reason of the less precipitation forecast. In addition, the duration of the rainstorm was also underestimated. (2) The second stage of the heavy rainfall happened in the central and southern Zhejiang. The southerly and southwesterly flow around the eastern or southern side of the low pressure drew abundant water vapor. There was a slowmoving cold front that resulted in the maintenance of the mesoscale convergence line in the lower atmosphere for a long time when the tropical depression landed at Fujian Province. There existed the coupling of the two systems, generating strong upward motion, and the instability energy in the lower level released. Finally, the severe downpour occurred. So, the underestimated influences of the tropical depression (moisture, dynamic and instability energy conditions) are the main inaccurate reason for the less precipitation forecast. (3) The big error of numerical forecasting is a crucial reason for overestimating the precipitation in the third stage. As a result, we find that the duration of the rainfall, heat condition and correction of numerical forecasting played important roles in predicting the precipitation of such rainstorm. Besides, there was strong upward motion caused by the coupling of tropical weather system and midlatitude weather system, and meanwhile, severe precipitation area maintained at the same place when the two systems confronted each other. What is more, plenty of water vapor and energy brought by jet caused the heavy rainfall frequently. Therefore, the actual rainfall was bigger than the forecasted rainfall.

Abstract:This paper analyses the forecast deviation of the first snow in 2016 in Beijing using EC reanalysis data and wind profiling radar from 19 to 21 November 2016. The conclusions are as follows. (1) The deficiency of “cold wedge”, the location of shear line at 850 hPa by south and the lack of whole layer precipitable water caused the total snowfall amount less than that of the forecast. (2) There are four reasons for the less snowfall amount in urban areas of Beijing. That is, the convergence of water vapor flux was weak, the time for rain to turn into snow was 3 hours later than that of the forecast, the saturation in the layer from -18 to -9℃ was insufficient, and the precipitation weather system had passed through Beijing after rain turned into snow. (3) The slow falling of the 0℃ layer height was due to the failing formation of “cold wedge”, and the strong warm advection in low levels. At the same time, the similar weather examples are compared in this paper. When it refers to the phenomenon of rain into snow and the insufficient saturation at high levels, the 0℃ layer height should still be considered in forecasting the time of rain into snow. When the 0℃ layer height drops to lower than 100 m, rain can turn into sleet and snow.

Abstract:Conventional meteorological data, Yinchuan CD Doppler radar and ECMWF, T639, WRF, NCEP/NCAR data are used to analyze the causes, predictability and significant forecast error margin of the rare severe rainstorm along the eastern Helan Mountain on 21 August 2016, which was missed by Ningxia Meteorological Observatory. The results show that the ECMWF, T639, WRF models predicted that there would be precipitation below 15 mm in the midnorth of Ningxia, but the magnitude was significantly smaller. Among the three models, the effect of ECMWF forecast was obviously superior to the other two. In the mid and late August 2016, the subtropical high was stronger continuously, located by north and getting to the strongest on the 21st. Under the control of the 588 dagpm line of the subtropical high, when the 592 dagpm line rapidly retreated eastward in southern Ningxia, the interaction of boundary layer shear line, lowlevel convergence and jet as well as the terrain of Helan Mountain triggered this severe rainstorm. Lack of experience, the local forecasters failed to capture the evolution of the circulation. The southeast and southerly lowlevel jet were formed when the stream in lower and middle layers were westbound, which was originated from the East China Sea, and established a warm convergence line in the eastern foot of Helan Mountain. The terrain friction convergence, the uplift of the wind and the movement of the mountain blocked the flow of Helan Mountain, and extended the rainfall time, contributing to the intensification of the rainstorm. Unfortunately, forecasters neglected the impact of Helan Mountain terrain on the rainfall.They didn’t grasp the forecast indicators, such as K index≥38℃, LI≤-3, Q≥12 kg·kg-1.By the analysis, the predictability of the severe rainstorm on 21 August was put forward.

Abstract:In contrast to the observation, there was an obvious deviation in tack and rainfall distributions in operational Typhoon Kujira (1508) forecasting in Hainan. It brought a great challenge to the operational typhoon warning service in Hainan. In this paper, the reasons of the track and rainfall forecast deviations for Typhoon Kujira are investigated by using conventional observation data, FY-2G satellite data and ECMWF ERAinterim reanalysis data (0.25°×0.25°) and operational deterministic model products. Meanwhile, the reason of the reverse rainfall distributions, which was caused by two groups of tropical cyclones in summer with the similar tracks to Typhoon Kujira, is analyzed. The results show that (1) the larger deviation in operational positioning of Typhoon Kujira for its weak intensity, and a weaker and more eastwardlocated subtropical high predicted by ECWMF are the main reasons for the failure of Typhoon Kujira’s track and landing point forecastings. (2) An enhanced subtropical westerly jet and upper-level north-easterly over South China Sea are well forecasted by ECMWF. However their roles in typhonic unsymmetrical structure via strong vertical shear are neglected, thus leading to the forecast deviation of Typhoon Kujira (1508) in track and rainfall distributions. (3) The evolution of the subtropical westerly jet and upper tropospheric westerly trough should be paid more attention in operational typhoon track and rainfall forecastings. When the subtropical westerly jet intensifies and moves southeastward, the vertical wind shear becomes greater accompanying the accelerating of upper-level northeasterly near the typhonic center. Severe convection tends to develop in the south side of the typhoon, which facilitates a westwardmoving of typhoon with heavy rainfall occurring at the left side of its track. However when the westerly trough moves southeastward in the vicinity to the typhoon circulation, the vertical wind shear near the typhonic center weakens significantly. At this time, the convection tends to develop in the north side of typhoon, which leads to a northwardmoving of typhoon with heavy rainfall occurring at the right side of its track.

Abstract:By using conventional data and simulation results from models of European Centre for MediumRange Weather Forecasts (ECMWF) and Chinese Unified Atmospheric Chemistry Environment Dust (CUACE/Dust), the weather situation, causes and forecasting results of the sand and dust weather process from 27 to 30 March 2015 were analyzed. This sand and dust weather in North China and Huanghuai Region of China resulted from the southward moving of cold air caused by rapid changes of northeast cyclone. And then, active warm air from southern regions brought the dust back to the north. Forecas ters’ inadequate estimation to the rapid changes of weather situation and the missing forecasting of the dust from the northern path by numerical models led to the underestimation of the strength and coverage area of the dust storm in North China and Huanghuai Region of China on 28 March. Moreover, the back transport of dust from south to north was not fully considered by both forecasters and models, especially in haze days. Thus, dusty, hazy, and the mixed weathers should be identified by observing the concentrations of PM10 and PM2.5 in spring, and pollution types could be judged by whether the primary pollutants were PM10 or not.

Abstract:This paper reviews the predictive information provided by dynamic models and precursors consi-dered in the operational monthly prediction of precipitation in October 2016. The average amount of precipitation over China in October 2016 is the largest compared to the precipitation in the corresponding period since 1951, and the spatio-temporal characteristics of precipitation distributions and atmospheric circulation changes significantly during the month. There exists considerable difference between operational forecasts of precipitation anomalies issued in late September and observational results in the south of North China, Huanghuai, Jianghuai and Jianghan regions. Meanwhile, the forecasts underestimate the adjustment of general circulation within month and the transition of precipitation anomaly distributions. Analysis of model output and the impact of external forcing signals indicate that the estimation of general atmospheric circulation is basically consistent with observation, while significant difference remains between the forecast and observation results of the intensity, position of ridge line of West Pacific subtropical high (WPSH), etc. Lagged and simultaneous correlation analysis between atmosphere and sea surface temperature (SST) indicates that, the WPSH tends to be intensified and northward in the autumn of decay years of medium and strong El Nino events. In addition, the anomalous meridional circulation induced by active convection in the West Pacific warm pool, which leads to the anticyclonic circulation anomaly around Japan, also contributes the intensification and northward of WPSH. The significant negative phase of tropical Indian Ocean dipole (TIOD) in September and October, which is in favor of strong Indian-Myanmar trough and vapor transport, combined with cold air from north, resulting in more precipitation in east and north China. Moreover, the active convection of tropical West Pacific, and increased number of generated and landed typhoons, is the main reason for heavy rains in the southeast coastal regions of China.

Abstract:During the summer of 2017 (June to August), the average precipitation over China was 348.6 mm, which is 8.1% more than the climate mean (332.6 mm). Two rainfall belts were observed over eastern China. The average temperature in summer was 21.7℃, which is 0.8℃ higher than the climate mean (20.9℃). The summer of 2017 is the second warmest since 1961. The East Asian summer monsoon (EASM) was weaker than normal, while the West Pacific subtropical high (WPSH) was significantly stronger than normal with southward ridge position. The anomalous Philippine Sea anticyclone controlled the lower troposphere. The “-+-” circulation pattern was located in Eurasian midhigh latitude, with negative anomalous over Ural Mountain and near Japan and positive anomalous over Baikal Region. Circulation configuration of low latitudes and midhigh latitudes resulted in anomalous convergence of moisture flux and more precipitation over the south of the Yangtze River. Further analysis indicated that the interdecadal strengthening of West Pacific subtropical high and the sea surface temperature anomalies over the central and eastern tropical Pacific changing from negative to positive are the important cause for the significantly more precipitation over the south of the Yangtze River.

Abstract:The main characteristics of the general atmospheric circulation in October 2017 are as follows. The polar vortex showed a dipole pattern. The circulation presented a fourwave pattern in middlehigh latitudes. The northwestern Pacific subtropical high was stronger and remarkably more westward extended than normal. The monthly mean precipitation amount was 46.2 mm, which is 29% more than normal (35.8 mm), ranking the top 8th since 1961. The monthly mean temperature was 10.6℃, 0.3℃ higher than normal. There were six largerange heavy rainfall processes and three cold air processes over China, of which the rainfall process from 8 to 10 October was accompanied by cold airs, resulting a rarelyseen largerange heavy rainfall in Autumn over North China. On 16 October, Typhoon Khanun (1720) made landfall at Leizhou Peninsula, Guangdong, bringing gale and rainfall to South China. In late October northern China suffered a foghaze event.