Using naive Bayes classifier for classification of convective rainfall intensities based on spectral characteristics retrieved from SEVIRI

This paper presents a new algorithm to classify convective clouds and determine their intensity, based on cloud physical properties retrieved from the Spinning Enhanced Visible and Infrared Imager (SEVIRI). The convective rainfall events at 15 min, 4 × 5 km spatial resolution from 2006 to 2012 are analysed over northern Algeria. The convective rain classification methodology makes use of the relationship between cloud spectral characteristics and cloud physical properties such as cloud water path (CWP), cloud phase (CP) and cloud top height (CTH). For this classification, a statistical method based on ‘naive Bayes classifier’ is applied. This is a simple probabilistic classifier based on applying ‘Bayes’ theorem with strong (naive) independent assumptions. For a 9-month period, the ability of SEVIRI to classify the rainfall intensity in the convective clouds is evaluated using weather radar over the northern Algeria. The results indicate an encouraging performance of the new algorithm for intensity differentiation of convective clouds using SEVIRI data.     

High temporal and spatial resolution observations of meso-scale features of pre- and mature summer monsoon cloud systems over Bangladesh

Meso-scale characteristics of disturbances that bring about atmospheric disasters in pre- and mature monsoon seasons in Bangladesh are analyzed. Several types of meteorological instruments capable of observations with high temporal and spatial resolutions were introduced for the first time in this area to capture the meso-scale structure of rainfall systems. We installed an automatic weather station (AWS) and several automatic raingauges (ARGs) and utilized the weather radar of Bangladesh Meteorological Department (BMD). From the radar image in the summer of 2001 (16–18 July), a striking feature of the systematic diurnal variation in this area was elucidated. In these 3 days, the diurnal evolutions of convective activity were remarkably similar to each other, implying that this pattern can be understood as a typical response of local cloud systems to the diurnal variation of insolation under some summer monsoon situations. The ARG data show the difference in characteristics of rainfall between pre- and mature monsoon seasons. The short intense downpour tends to occur more frequently in the pre-monsoon season than in the mature monsoon season. The pre-monsoon rainfall also has clear diurnal variation with a peak that is more strongly concentrated in time. In the northern part the rainfall peak is found in between midnight and early morning, while it is observed in the daytime in central to western parts of the country. Two disaster cases caused by meso-scale disturbances are analyzed. Although they occurred in the same season, the structures of the cloud systems were largely different from each other. The disturbance brought about tornadoes on 14 April 2004, consisting of many spherical cloud systems of approximately 20 km size. On the other hand, another one that caused the tragic river water transport accident on 23 May 2004 had meso-scale rain band structure. The latter case was captured by the AWS located at Dhaka. Sudden changes in temperature, wind and pressure were observed clearly, showing the typical structure of convective rain bands.     

Numerical simulation of deep convective cloud seeding using liquid carbon dioxide

In this study, a one-dimensional transient cumulonimbus cloud is modeled to be seeded by liquid CO₂. The model includes microphysical and dynamical processes associated with glaciogenic seeding by homogenous ice nucleation and two thermal terms associated with seeding by ?90 ºC liquid CO₂. For this model, the study concentrates on five types of hydrometeors, namely, cloud droplet, cloud ice, snow, hail/graupel, and rain. Point and horizontal seeding methods are implemented to observe their implications for rainfall enhancement, amount of hail/graupel production, vertical cloud extension, and radar’s reflectivity. In addition, the seeding temperature effects on the rainfall and microphysical processes are investigated. The results of the study show that, the rainfall enhancement and rainfall intensity in the point seeding case are more than those in the horizontal seeding. Moreover, the study reveals that, there is a vertical cloud extension enhancement of 0.5 km for clouds with top height of 10.5 km. The most important sources of the rain water production are found to be the accretion of cloud water by rain (P _RACW) and by snow (P _SACW), and for the graupel production is dry growth of the graupel (P _GDRY). The results of this study are confirmed by the results of other investigators and are found to be comparable with the recorded data at rain gauge stations.     

青藏高原地区云水时空变化特征及其与降水的联系

利用1984-2009年ISCCP的云量、云光学厚度(COT)、云水路径(CWP)资料,分析了青藏高原云水的分布特征、变化趋势,及其与夏秋季降水、冬春季降雪的联系.结果表明:青藏高原地区大气中的云水有着显著的季节变化与水平分布差异;青藏高原春夏季总云量、高云云量高于秋冬季,CWP、COT与总云量的分布特征具有较好的一致性.高原云量高值区位于喀喇昆仑山与高原东南部;可可西里地区由于羌塘高压的下沉作用为云量低值区.青藏高原总云量在1984-2009年间呈现减少趋势;而CWP在高原总体以增加为主,但在各区域上的变化不一致,高原东部CWP增加而西部出现较弱的减小,这与来自孟加拉湾的水汽输送增加有关.青藏高原中东部地区夏秋季降水受云量减少影响较小而与CWP的增加相一致呈增长趋势;该地区冬春季降雪略有减少,与总云量的年际变化具有正相关.     

Real-time nowcast of a cloudburst and a thunderstorm event with assimilation of Doppler weather radar data

Extreme weather events such as cloudburst and thunderstorms are great threat to life and property. It is a great challenge for the forecasters to nowcast such hazardous extreme weather events. Mesoscale model (ARPS) with real-time assimilation of DWR data has been operationally implemented in India Meteorological Department (IMD) for real-time nowcast of weather over Indian region. Three-dimensional variational (ARPS3DVAR) technique and cloud analysis procedure are utilized for real-time data assimilation in the model. The assimilation is performed as a sequence of intermittent cycles and complete process (starting from reception, processing and assimilation of DWR data, running of ARPS model and Web site updation) takes less than 20 minutes. Thus, real-time nowcast for next 3 h from ARPS model is available within 20 minutes of corresponding hour. Cloudburst event of September 15, 2011, and thunderstorm event of October 22, 2010, are considered to demonstrate the capability of ARPS model to nowcast the extreme weather events in real time over Indian region. Results show that in both the cases, ARPS3DVAR and cloud analysis technique are able to extract hydrometeors from radar data which are transported to upper levels by the strong upward motion resulting in the distribution of hydrometeors at various isobaric levels. Dynamic and thermodynamic structures of cloudburst and thunderstorm are also well simulated. Thus, significant improvement in the initial condition is noticed. In the case of cloudburst event, the model is able to capture the sudden collisions of two or more clouds during 09–10 UTC. Rainfall predicted by the model during cloudburst event is over 100 mm which is very close to the observed rainfall (117 mm). The model is able to predict the cloudburst with slight errors in time and space. Real-time nowcast of thunderstorm shows that movement, horizontal extension, and north–south orientation of thunderstorm are well captured during first hour and deteriorate thereafter. The amount of rainfall predicted by the model during thunderstorm closely matches with observation with slight errors in the location of rainfall area. The temporal and spatial information predicted by ARPS model about the sudden collision/merger and broken up of convective cells, intensification, weakening, and maintaining intensity of convective cells has added value to a human forecast.     

CondMerg: an open source implementation in R language of conditional merging for weather radars and rain gauges observations

Monitoring of extreme events requires accurate measurement of rainfall intensities and merging weather radar data with ground information is a very common technique used to obtain the required precision. In order to do this, several methods exist but very few open source implementations are available. CondMerg is the first open source software developed in R language implementing the conditional merging method and some other experimental variants based on it. It is a cross-platform software, easily adaptable to different needs, optimized for batch processing of multiple events but also usable in near real time applications. For its execution it requires two inputs: a CSV file with rain gauges measurements and a geo-referred TIF file with weather radar quantitative precipitation estimations; main outputs are TIF files with merged observations although the code also returns information about cross-validation, with scatter plots and indexes. All TIF files are ready to be managed by common GIS software for easy visualization and analysis. Use of the program is very simple: execution can be interactive or non-interactive and, in both cases, it just requires to set some parameters at the beginning of the program and run it. The code has been tested on different extreme rain events occurred in the Piedmont region (northwestern Italy) showing improved accuracy of reconstructed rainfall fields.     

南京市强降水天气长期动态及变异性规律

基于南京市1951～2016年汛期(6～9月)各月降水资料,分析研究区强降水天气的长期动态及变异性规律。长期动态结果表明,近66年来南京市强降水天气发生频率显著增加,平均每10年增加0.31次;不同规模强降水中,暴雨、大暴雨天气发生频率均呈增加趋势,其中暴雨天气呈显著增加。多年(7a、10a、15a)月际排序值滑动标准差均通过置信度99%显著性检验,强降水天气变异性呈显著下降趋势;强降水天气汛期盛行月份趋于集中(7月),月间格局趋于稳定,与多年的月际排序值滑动标准差得出强降水天气变异性呈下降趋势的结论相吻合。     

用GMS卫星资料反演复杂地形下的降水率

强降水主要由生命史短的中小尺度天气系统造成,对此类天气系统的预报,目前只有依靠卫星和雷达的实时监测并结合中系统的概念模式外推来完成。由于中国西北地区地形极为复杂,造成雷达盲区,影响其估算降水率。因此采用GMS-4卫星的红外和可见光展宽云图资料,经处理并转换后,再加入相应网格点上的数字化地形高度资料作为因子之一,用多级逐步判别模式估算逐时雨强等级,最后形成一套可在微机上对雨强场进行图像显示及处理的软件系统,满足了现时预报的需要。结果表明,小雨以上的降雨区域不论面积、形状均与实况基本一致。     

利用FY-2C双光谱图像反演白天像素级逐时雨强

基于中国静止气象卫星FY-2C的多光谱数据,研究了由热红外亮温和可见光反照率拟合反演白天地面像素级逐时雨强分布的方法.利用降雨概率判识矩阵区分了降雨云和非降雨云,技术得分为0.5073,比简单利用阈值划分方法显著提高了雨区判识精度.对逐时降水率的分析,允许误差为±20%内,正确率为52.19%.对降水等级的分析,经1200多个实测有雨样本的检验,其对小雨、中雨、大雨和暴雨的实测命中率达62.79%.若将降水率小于0.5mm/h的微雨这种临界状态作无雨处理,则有雨样本的降水强度判识准确率达到88.17%.     

Hydrometeorological thresholds for landslide initiation and forest operation shutdowns on the north coast of British Columbia

Debris flows and debris avalanches are the most widespread and hazardous types of landslides on the British Columbia north coast. Triggered by heavy rain, they pose risks to forestry workers in sparsely developed regions. The scarcity of long-term quality rain gauges and the lack of weather radar information create significant challenge in predicting the timing of landslides, which could be used to warn and, when necessary, evacuate forestry personnel. Traditional methods to relate rainfall antecedents and rainfall intensity to known landslide dates have proven to be unsatisfactory in this study due to extreme spatial variability of rainfall, enhanced by the orographic effect and the scarcity of rain gauges in a very large area. This has led to an integration of meteorological variables in a landslide advisory system that classifies three types of approaching storms by the 850-mbar wind speed and direction, the occurrence of subtropical moisture flow, and the existence of a warm layer characterized by high thickness values of the 500- to 1,000-mbar pressure levels. The storm classification was combined with a 4-week antecedent rainfall and the 24-h rainfall measured near or in the watershed where logging operations are taking place. This system, once implemented, is thought to reduce loss of life, injury, and economic losses associated with forestry works in the study area.     

Impact of cloud microphysics and cumulus parameterization on simulation of heavy rainfall event during 7–9 October 2007 over Bangladesh

In the present study, the Advanced Research WRF (ARW) version 3.2.1 has been used to simulate the heavy rainfall event that occurred between 7 and 9 October 2007 in the southern part of Bangladesh. Weather Research and Forecast (WRF–ARW version) modelling system with six different microphysics (MP) schemes and two different cumulus parameterization (CP) schemes in a nested configuration was chosen for simulating the event. The model domains consist of outer and inner domains having 9 and 3 km horizontal resolution, respectively with 28 vertical sigma levels. The impacts of cloud microphysical processes by means of precipitation, wind and reflectivity, kinematic and thermodynamic characteristics of the event have been studied. Sensitivity experiments have been conducted with the WRF model to test the impact of microphysical and cumulus parameterization schemes in capturing the extreme weather event. NCEP FNL data were used for the initial and boundary condition. The model ran for 72 h using initial data at 0000 UTC of 7 October 2007. The simulated rainfall shows that WSM6–KF combination gives better results for all combinations and after that Lin–KF combination. WSM3–KF has simulated, less area average rainfall out of all MP schemes that were coupled with KF scheme. The sharp peak of relative humidity up to 300 hPa has been simulated along the vertical line where maximum updraft has been found for all MPs coupled with KF and BMJ schemes. The simulated rain water and cloud water mixing ratio were maximum at the position where the vertical velocity and reflectivity has also been maximum. The production of rain water mixing ratio depends on MP schemes as well as CP schemes. Rainfall depends on rain water mixing ratio between 950 and 500 hPa. Rain water mixing ratio above 500 hPa level has no effect on surface rain.     

新疆天山山区降雨的微物理结构特征

使用GBPP-100型雨滴谱仪,于2001年6月12日至7月31日在天山北坡的小渠子气象站和牧业气象试验站,对27次降雨过程进行了雨滴谱观测,共获取了4 719个雨滴谱样本。通过观测资料分析新疆中天山山区积状云、层状云、积状-层状混合云降雨的微物理结构特征。观测分析表明,天山山区降雨雨滴的平均直径0.41~0.55 mm,以积状云最大,混合云次之,层状云最小。最大平均直径0.88~1.12 mm、平均雨强1.18~2.78 mm·h^-1、平均含水量5.23~11.62 g·m^-3,混合云的这三个特征量均为最大。三类云的雨强与数密度呈正相关。积状云、层状云降雨的雨滴谱服从M-P分布,混合云服从Γ分布。由于山区地形的作用,使云中降雨粒子的生长时间受到限制,天山山区降雨小滴浓度高、尺度小,人工降雨潜力大。     

Application of radar wind profiler data in analyzing the process of torrential rains in Urumqi Xinjiang,China

To deeply understand the micro-/mesoscale dynamic characteristics of the torrential rain process in Urumqi and improve the levels of torrential rain monitoring, forecasting and early warning, this paper analyzed the wind profile features and related scientific problems of three typical torrential rain events seen in 2013–2015 in this region. The research results suggested that: (1) Radar wind profiler can record in detail the movement condition of the atmosphere during the process of torrential rains. Carrying out detailed analyses on the wind profile data is conductive to the improvement in monitoring, forecasting and warning to torrential rain event at a single observation station. (2) When Urumqi experiences heavy rain weather, noticeable wind shear is usually observed above the observation station. In the upper air, it is southwest wind, while in the lower air it is northwest wind, which is the typical wind profile pattern for heavy rain events in the Urumqi region. (3) Obvious northwest low-level jet stream is found to go together with precipitation, and the jet is positively correlated with precipitation intensity. The velocity of low-level jet stream greatly impacts the amount and intensity of precipitation. (4) The rainstorm weather phenomena are clearly presented by the time–height chart of radar reflectivity factors. The high reflectivity values correspond positively to the height range of cloud–rain particles as well as the duration and intensity of precipitation, so it can be used as a reference index of precipitation monitoring and early warning. In a word, this research deepens on the recognition to the micro-/mesoscale weather systems during the process of heavy rains in Urumqi. Moreover, it would contribute to the application improvement of wind profiler data in analyzing the heavy rainfalls in this region.     

Spatio-temporal analysis of sub-hourly rainfall over Mumbai,India: Is statistical forecasting futile?

Mumbai, the commercial and financial capital of India, experiences incessant annual rain episodes, mainly attributable to erratic rainfall pattern during monsoons and urban heat-island effect due to escalating urbanization, leading to increasing vulnerability to frequent flooding. After the infamous episode of 2005 Mumbai torrential rains when only two rain gauging stations existed, the governing civic body, the Municipal Corporation of Greater Mumbai (MCGM) came forward with an initiative to install 26 automatic weather stations (AWS) in June 2006 (MCGM 2007), which later increased to 60 AWS. A comprehensive statistical analysis to understand the spatio-temporal pattern of rainfall over Mumbai or any other coastal city in India has never been attempted earlier. In the current study, a thorough analysis of available rainfall data for 2006–2014 from these stations was performed; the 2013–2014 sub-hourly data from 26 AWS was found useful for further analyses due to their consistency and continuity. Correlogram cloud indicated no pattern of significant correlation when we considered the closest to the farthest gauging station from the base station; this impression was also supported by the semivariogram plots. Gini index values, a statistical measure of temporal non-uniformity, were found above 0.8 in visible majority showing an increasing trend in most gauging stations; this sufficiently led us to conclude that inconsistency in daily rainfall was gradually increasing with progress in monsoon. Interestingly, night rainfall was lesser compared to daytime rainfall. The pattern-less high spatio-temporal variation observed in Mumbai rainfall data signifies the futility of independently applying advanced statistical techniques, and thus calls for simultaneous inclusion of physics-centred models such as different meso-scale numerical weather prediction systems, particularly the Weather Research and Forecasting (WRF) model.     

Advances in Measurement Techniques and Statistics Features

Rain Drop Size Distribution (DSD) is one of the key parameters to micro physical process and macro dynamical structure of precipitation. It provides useful information for understanding the mechanisms of precipitation formation and development. Conventional measurement techniques include momentum method, flour method, filtering paper, raindrop camera and immersion method. In general, the techniques havelarge measurement error, heavy workload, and low efficiency. Innovation of disdrometer is a remarkable progress in DSD observation. To date, the major techniques are classified into impacting, optical and acoustic disdrometers, which are automated and more convenient and accurate. The impacting disdrometer transforms the momentum of raindrops into electric impulse, which are easy to operate and quality assured but with large errors for extremely large or small raindrops. The optical disdrometer measures rainfall diameter and its velocity in the same time, but cannot distinguish the particles passing through sampling area simultaneously. The acoustic disdrometer determines DSD from the raindrop impacts on water body with a high temporal resolution but easily affected by wind. In addition, the Doppler can provide DSD with polarimetric techniques for large area while it is affected by updrafts, downdrafts and horizontal winds.DSD has meteorological features, which can be described with the Marshall Palmer (M-P), the Gamma, the lognormal or the normalized models. The M P model is suitable for steady rainfall, usually used for weak and moderate rainfall. The gamma model is proposed for DSD at high rain rate. The lognormal model is widely applied for cloud droplet analysis, but not appropriate for DSD with a broad spectrum. The normalized model is free of assumptions about the shape of the DSD. For practical application, statistical comparison is necessary for selection of a most suitable model. Meteorologically, convective rain has a relatively narrow and smooth DSD spectrum usually described by the M P model. Stratiform rain has a broad DSD spectrum described with the Gamma model. Stratocumulus mixed rain has relatively large drop diameter but small mean size usually described by the Gamma model. The continent rainfall is altitude dependent and it differs from the maritime cloud rainfalls in terms of rain rate and drop diameter. Overall, the meteorological features are useful to improve our understanding of precipitation formation but also important to development of precipitation retrieval techniques with a high accuracy.     

Comparison of recorded rainfall with quantitative precipitation forecast in a rainfall-runoff simulation for the Langat River Basin, Malaysia

Observed rainfall is used for runoff modeling in flood forecasting where possible, however in cases where the response time of the watershed is too short for flood warning activities, a deterministic quantitative precipitation forecast (QPF) can be used. This is based on a limited-area meteorological model and can provide a forecasting horizon in the order of six hours or less. This study applies the results of a previously developed QPF based on a 1D cloud model using hourly NOAA-AVHRR (Advanced Very High Resolution Radiometer) and GMS (Geostationary Meteorological Satellite) datasets. Rainfall intensity values in the range of 3–12 mm/hr were extracted from these datasets based on the relation between cloud top temperature (CTT), cloud reflectance (CTR) and cloud height (CTH) using defined thresholds. The QPF, prepared for the rainstorm event of 27 September to 8 October 2000 was tested for rainfall runoff on the Langat River Basin, Malaysia, using a suitable NAM rainfall-runoff model. The response of the basin both to the rainfall-runoff simulation using the QPF estimate and the recorded observed rainfall is compared here, based on their corresponding discharge hydrographs. The comparison of the QPF and recorded rainfall showed R² = 0.9028 for the entire basin. The runoff hydrograph for the recorded rainfall in the Kajang sub-catchment showed R² = 0.9263 between the observed and the simulated, while that of the QPF rainfall was R² = 0.819. This similarity in runoff suggests there is a high level of accuracy shown in the improved QPF, and that significant improvement of flood forecasting can be achieved through ‘Nowcasting’, thus increasing the response time for flood early warnings.     

A new ensemble-based data assimilation algorithm to improve track prediction of tropical cyclones

This paper proposes a new ensemble-based algorithm that assimilates the vertical rain structure retrieved from microwave radiometer and radar measurements in a regional weather forecast model, by employing a Bayesian framework. The goal of the study is to evaluate the capability of the proposed technique to improve track prediction of tropical cyclones that originate in the North Indian Ocean. For this purpose, the tropical cyclone Jal has been analyzed by the community mesoscale weather model, weather research and forecasting (WRF). The ensembles of prognostic variables such as perturbation potential temperature (θ, k), perturbation geopotential (?, m²/s²), meridional (U) and zonal velocities (V) and water vapor mixing ratio (q _v , kg/kg) are generated by the empirical orthogonal function technique. An over pass of the tropical rainfall-measuring mission (TRMM) satellite occurred on 06th NOV 0730 UTC over the system, and the observations from the radiometer and radar on board the satellite(1B11 data products) are inverted using a combined in-home radiometer-radar retrieval technique to estimate the vertical rain structure, namely the cloud liquid water, cloud ice, precipitation water and precipitation ice. Each ensemble is input as a possible set of initial conditions to the WRF model from 00 UTC which was marched in time till 06th NOV 0730 UTC. The above-mentioned hydrometeors from the cloud water and rain water mixing ratios are then estimated for all the ensembles. The Bayesian filter framework technique is then used to determine the conditional probabilities of all the candidates in the ensemble by comparing the retrieved hydrometeors through measured TRMM radiances with the model simulated hydrometeors. Based on the posterior probability density function, the initial conditions at 06 00 UTC are then corrected using a linear weighted average of initial ensembles for the all prognostic variables. With these weighted average initial conditions, the WRF model has been run up to 08th Nov 06 UTC and the predictions are then compared with observations and the control run. An ensemble independence study was conducted on the basis of which, an optimum of 25 ensembles is arrived at. With the optimum ensemble size, the sensitivity of prognostic variables was also analyzed. The model simulated track when compared with that obtained with the corrected set of initial conditions gives better results than the control run. The algorithm can improve track prediction up to 35 % for a 24 h forecast and up to 12 % for a 54 h forecast.     

基于水文模型的雷达监测降雨量误差传递研究

基于多普勒雷达和相应的雨量计资料,利用卡尔曼滤波校准法对分组Z~I关系估算的雷达降雨进行同化,结合新安江模型,提出采用增长模繁殖法对雷达监测降雨量资料进行扰动,定量分析了模型输入误差对径流模拟的影响。经过湖北省白莲河流域的实际应用研究表明,卡尔曼滤波校准法估算降水量的精度比Z~I关系法有了明显的提高,其相应的洪水预报效果也都优于Z~I关系;提出的增长模繁殖法对雷达监测降雨量资料产生的扰动输入误差在经过模型的传递后有增大的趋势。     

毫米波测云雷达的特点及其研究现状与展望

云在大气的能量分配、辐射传输,尤其是水循环系统中有不可忽视的作用。云探测对云物理、人工影响天气、气候变化和航空航天等领域有重要意义,是大气科学研究的热点之一。尽管目前已经发展了很多种遥感设备对云进行观测(如激光雷达、卫星、云幂仪等),但这些设备无法得到高时空分辨率的云水平和垂直结构,而毫米波雷达是云三维精细结构探测的重要工具。由于毫米波雷达具有更接近小粒子尺度的短波长,因此更适合用来探测弱云,同时毫米波雷达也存在衰减严重的缺点。介绍了毫米波雷达的特点以及其探测小粒子的优势;对比分析了其与新一代多普勒天气雷达、晴空风廓线雷达的差异,得出：毫米波雷达具有高时空分辨率,能够更精确地反映云的垂直和水平结构,比普通天气雷达更适合监测云的变化。概括了国内外毫米波测云雷达的发展现状以及在云物理研究方面的情况,并展望了国内毫米波雷达未来研究的方向。