Study on Quantitative Precipitation Estimation by Polarimetric Radar Using Deep Learning

Accurate radar quantitative precipitation estimation(QPE)plays an essential role in disaster prevention and mitigation.In this paper,two deep learning-based QPE networks including a single-parameter network and a multi-parameter network are designed.Meanwhile,a self-defined loss function(SLF)is proposed during modeling.The dataset includes Shijiazhuang S-band dual polarimetric radar(CINRAD/SAD)data and rain gauge data within the radar’s 100-km detection range during the flood season of 2021 in North China.Considering that the specific propagation phase shift(KDP)has a roughly linear relationship with the precipitation intensity,KDP is set to 0.5°km^(-1 )as a threshold value to divide all the rain data(AR)into a heavy rain(HR)and light rain(LR)dataset.Subsequently,12 deep learning-based QPE models are trained according to the input radar parameters,the precipitation datasets,and whether an SLF was adopted,respectively.The results suggest that the effects of QPE after distinguishing rainfall intensity are better than those without distinguishing,and the effects of using SLF are better than those that used MSE as a loss function.A Z-R relationship and a ZH-KDP-R synthesis method are compared with deep learning-based QPE.The mean relative errors(MRE)of AR models using SLF are improved by 61.90%,51.21%,and 56.34%compared with the Z-R relational method,and by 38.63%,42.55%,and 47.49%compared with the synthesis method.Finally,the models are further evaluated in three precipitation processes,which manifest that the deep learning-based models have significant advantages over the traditional empirical formula methods.

Quantitative Precipitation Estimation Using X-band Phased Array Weather Radar

This study utilized data from an X-band phased array weather radar and ground-based rain gauge observations to conduct a quantitative precipitation estimation(QPE)analysis of a heavy rainfall event in Xiong an New Area from 20:00 on August 21 to 07:00 on August 22,2022.The analysis applied the Z-R relationship method for radar-based precipitation estimation and evaluated the QPE algorithm s performance using scatter density plots and binary classification scores.The results indicated that the QPE algorithm accurately estimates light to moderate rainfall but significantly underestimates heavy rainfall.The study identified disparities in the predictive accuracy of the QPE algorithm across various precipitation intensity ranges,offering essential insights for the further refinement of QPE techniques.

Operational Evaluation of the Quantitative Precipitation Estimation by a CINRAD-SA Dual Polarization Radar System

In this paper,a quantitative precipitation estimation based on the hydrometeor classification(HCA-QPE)algorithm was proposed for the first operational S band dual-polarization radar upgraded from the CINRAD/SA radar of China.The HCA-QPE algorithm,localized Colorado State University-Hydrometeor Identification of Rainfall(CSUHIDRO)algorithm,the Joint Polarization Experiment(JPOLE)algorithm,and the dynamic Z-R relationships based on variational correction QPE(DRVC-QPE)algorithm were evaluated with the rainfall events from March 1 to October 30,2017 in Guangdong Province.The results indicated that even though the HCA-QPE algorithm did not use the observed rainfall data for correction,its estimation accuracy was better than that of the DRVC-QPE algorithm when the rainfall rate was greater than 5 mm h-1;and the stronger the rainfall intensity,the greater the QPE improvement.Besides,the HCA-QPE algorithm worked better than the localized CSU-HIDRO and JPOLE algorithms.This study preliminarily evaluated the improved accuracy of QPE by a dual-polarization radar system modified from CINRAD-SA radar.

Radar Quantitative Precipitation Estimation Based on the Gated Recurrent Unit Neural Network and Echo-Top Data

The Gated Recurrent Unit(GRU) neural network has great potential in estimating and predicting a variable. In addition to radar reflectivity(Z), radar echo-top height(ET) is also a good indicator of rainfall rate(R). In this study, we propose a new method, GRU_Z-ET, by introducing Z and ET as two independent variables into the GRU neural network to conduct the quantitative single-polarization radar precipitation estimation. The performance of GRU_Z-ET is compared with that of the other three methods in three heavy rainfall cases in China during 2018, namely, the traditional Z-R relationship(Z=300R1.4), the optimal Z-R relationship(Z=79R1.68) and the GRU neural network with only Z as the independent input variable(GRU_Z). The results indicate that the GRU_Z-ET performs the best, while the traditional Z-R relationship performs the worst. The performances of the rest two methods are similar.To further evaluate the performance of the GRU_Z-ET, 200 rainfall events with 21882 total samples during May–July of 2018 are used for statistical analysis. Results demonstrate that the spatial correlation coefficients, threat scores and probability of detection between the observed and estimated precipitation are the largest for the GRU_Z-ET and the smallest for the traditional Z-R relationship, and the root mean square error is just the opposite. In addition, these statistics of GRU_Z are similar to those of optimal Z-R relationship. Thus, it can be concluded that the performance of the GRU_ZET is the best in the four methods for the quantitative precipitation estimation.

Application of X-band Polarimetric Phased-array Radars in Quantitative Precipitation Estimation

The performance of different quantitative precipitation estimation(QPE) relationships is examined using the polarimetric variables from the X-band polarimetric phased-array radars in Guangzhou,China.Three QPE approaches,namely,R(ZH),R(ZH,ZDR) and R(KDP),are developed for horizontal reflectivity,differential reflectivity and specific phase shift rate,respectively.The estimation parameters are determined by fitting the relationships to the observed radar variables using the T-matrix method.The QPE relationships were examined using the data of four heavy precipitation events in southern China.The examination shows that the R(ZH) approach performs better for the precipitation rate less than 5 mm h-1, and R(KDP) is better for the rate higher than 5 mm h-1, while R(ZH,ZDR) has the worst performance.An adaptive approach is developed by taking the advantages of both R(ZH) and R(KDP) approaches to improve the QPE accuracy.

A STUDY ON VARIABLE QUANTITATIVE PRECIPITATION ESTIMATION USING DOPPLER RADAR DATA

With the pros and cons of the traditional optimization and probability pairing methods thoroughly considered, an improved optimal pairing window probability technique is developed using a dynamic relationship between the base reflectivity Z observed by radar and real time precipitation I by rain gauge. Then, the Doppler radar observations of base reflectivity for typhoons Haitang and Matsa in Wenzhou are employed to establish various Z-I relationships, which are subsequently used to estimate hourly precipitation of the two typhoons. Such estimations are calibrated by variational techniques. The results show that there exist significant differences in the Z-I relationships for the typhoons, leading to different typhoon precipitation efficiencies. The typhoon precipitation estimated by applying radar base reflectivity is capable of exhibiting clearly the spiral rain belts and mesoscale cells, and well matches the observed rainfall. Error statistical analyses indicate that the estimated typhoon precipitation is better with variational calibration than the one without. The variational calibration technique is able to maintain the characteristics of the distribution of radar-estimated typhoon precipitation, and to significantly reduce the error of the estimated precipitation in comparison with the observed rainfall.

Application of the Attenuation Correction Technology in C-band Radar Precipitation Estimation

[ Objective] The research aimed to study application of the attenuation correction technology in C-band radar precipitation estimation. [ Method~ Based on CINRAD-CB radar data in Shaanxi, we conducted the attenuation correction experiment by using iteration method and Kufa method respectively. Moreover, we conducted application expedment of the Kufa attenuation correction method in the quantitative precipitation esti- mation. [ Result~ Attenuation correction technology could compensate for attenuation problem of the echo at the distant range. Calculation result of the iteration method finally tended to that of the Kufa method. Moreover, iteration method spent more time. Therefore, Kufa attenuation correction technology was more suitable for business operation. When strong echo was near radar, generated attenuation was more obvious, and application value of the attenuation correction was bigger. Attenuation correction technology was used for quantitative precipitation estimation, which was favor- able for improving accuracy of the precipitation estimation. But we should conduct detailed planning on calculation scheme of the precipitation esti- mation because that different calculation schemes had great influences on accuracy of the quantitative precipitation estimation. [ Cendusien] This research provided a basis for improving accuracy of the quantitative precipitation estimation in Shaanxi. Key words Attenuation correction

A Strategy for Merging Objective Estimates of Global Daily Precipitation from Gauge Observations, Satellite Estimates, and Numerical Predictions

This paper describes a strategy for merging daily precipitation information from gauge observations, satellite estimates （SEs）, and numerical predictions at the global scale. The strategy is designed to remove systemic bias and random error from each individual daily precipitation source to produce a better gridded global daily precipitation product through three steps. First, a cumulative distribution function matching procedure is performed to remove systemic bias over gauge-located land areas. Then, the overall biases in SEs and model predictions （MPs） over ocean areas are corrected using a rescaled strategy based on monthly precipitation. Third, an optimal interpolation （OI）-based merging scheme （referred as the HL-OI scheme） is used to combine unbiased gahge observations, SEs, and MPs to reduce random error from each source and to produce a gauge--satellite-model merged daily precipitation analysis, called BMEP-d （Beijing Climate Center Merged Estimation of Precipitation with daily resolution）, with complete global coverage. The BMEP-d data from a four-year period （2011- 14） demonstrate the ability of the merging strategy to provide global daily precipitation of substantially improved quality. Benefiting from the advantages of the HL-OI scheme for quantitative error estimates, the better source data can obtain more weights during the merging processes. The BMEP-d data exhibit higher consistency with satellite and gauge source data at middle and low latitudes, and with model source data at high latitudes. Overall, independent validations against GPCP-1DD （GPCP one-degree daily） show that the consistencies between B MEP-d and GPCP-1DD are higher than those of each source dataset in terms of spatial pattern, temporal variability, probability distribution, and statistical precipitation events.

QpefBD:A Benchmark Dataset Applied to Machine Learning for Minute-Scale Quantitative Precipitation Estimation and Forecasting

Nowcasts of strong convective precipitation and radar-based quantitative precipitation estimations have always been hot yet challenging issues in meteorological sciences.Data-driven machine learning,especially deep learning,provides a new technical approach for the quantitative estimation and forecasting of precipitation.A high-quality,large-sample,and labeled training dataset is critical for the successful application of machine-learning technology to a specific field.The present study develops a benchmark dataset that can be applied to machine learning for minutescale quantitative precipitation estimation and forecasting(QpefBD),containing 231,978 samples of 3185 heavy precipitation events that occurred in 6 provinces of central and eastern China from April to October 2016-2018.Each individual sample consists of 8 products of weather radars at 6-min intervals within the time window of the corresponding event and products of 27 physical quantities at hourly intervals that describe the atmospheric dynamic and thermodynamic conditions.Two data labels,i.e.,ground precipitation intensity and areal coverage of heavy precipitation at 6-min intervals,are also included.The present study describes the basic components of the dataset and data processing and provides metrics for the evaluation of model performance on precipitation estimation and forecasting.Based on these evaluation metrics,some simple and commonly used methods are applied to evaluate precipitation estimates and forecasts.The results can serve as the benchmark reference for the performance evaluation of machine learning models using this dataset.This paper also gives some suggestions and scenarios of the QpefBD application.We believe that the application of this benchmark dataset will promote interdisciplinary collaboration between meteorological sciences and artificial intelligence sciences,providing a new way for the identification and forecast of heavy precipitation.

Short-Term Dynamic Radar Quantitative Precipitation Estimation Based on Wavelet Transform and Support Vector Machine

Currently,Doppler weather radar in China is generally used for quantitative precipitation estimation(QPE)based on the Z–R relationship.However,the estimation error for mixed precipitation is very large.In order to improve the accuracy of radar QPE,we propose a dynamic radar QPE algorithm with a 6-min interval that uses the reflectivity data of Doppler radar Z9002 in the Shanghai Qingpu District and the precipitation data at automatic weather stations(AWSs)in East China.Considering the time dependence and abrupt changes of precipitation,the data during the previous 30-min period were selected as the training data.To reduce the complexity of radar QPE,we transformed the weather data into the wavelet domain by means of the stationary wavelet transform(SWT)in order to extract high and low-frequency reflectivity and precipitation information.Using the wavelet coefficients,we constructed a support vector machine(SVM)at all scales to estimate the wavelet coefficient of precipitation.Ultimately,via inverse wavelet transformation,we obtained the estimated rainfall.By comparing the results of the proposed method(SWTSVM)with those of Z=300×R1.4,linear regression(LR),and SVM,we determined that the root mean square error(RMSE)of the SWT-SVM method was 0.54 mm per 6 min and the average Threat Score(TS)could exceed 40%with the exception of the downpour category,thus remaining at a high level.Generally speaking,the SWT-SVM method can effectively improve the accuracy of radar QPE and provide an auxiliary reference for actual meteorological operational forecasting.

A CRPS-Based Spatial Technique for the Verification of Ensemble Precipitation Forecasts

Traditional precipitation skill scores are affected by the well-known"double penalty"problem caused by the slight spatial or temporal mismatches between forecasts and observations.The fuzzy(neighborhood)method has been proposed for deterministic simulations and shown some ability to solve this problem.The increasing resolution of ensemble forecasts of precipitation means that they now have similar problems as deterministic forecasts.We developed an ensemble precipitation verification skill score,i.e.,the Spatial Continuous Ranked Probability Score(SCRPS),and used it to extend spatial verification from deterministic into ensemble forecasts.The SCRPS is a spatial technique based on the Continuous Ranked Probability Score(CRPS)and the fuzzy method.A fast binomial random variation generator was used to obtain random indexes based on the climatological mean observed frequency,which were then used in the reference score to calculate the skill score of the SCRPS.The verification results obtained using daily forecast products from the ECMWF ensemble forecasts and quantitative precipitation estimation products from the OPERA datasets during June-August 2018 shows that the spatial score is not affected by the number of ensemble forecast members and that a consistent assessment can be obtained.The score can reflect the performance of ensemble forecasts in modeling precipitation and thus can be widely used.

A MODEL FOR QUANTITATIVELY ESTIMATING SHORTRANGE PRECIPITATION BASED ON GMS DIGITALIZED CLOUD MAPS-PART Ⅰ:ANALYSIS OF QUANTITATIVE CLOUD-PRECIPITATION RELATIONS AND MODEL DESIGN

Some typical samples are used to explore the quantitative correlation with their features between a convective cloud and its rainfall field,with which to develop two morphological functions for the correlation and by singling out their most suitable groups of parameters we propose a model for quantitatively estimating precipitation in the context o{ the in-advance recognition of meso-α convective system properties and its precipitating center.From the model fitting precision and forecasting accuracy we find that it is feasible to utilize geostationary meteorological satellite (GMS) digitalized imagery for estimating short-term rainfall in a quantitative manner.Also,evidence suggests that the model is supposed to be restricted in its applicability due to the fact that the employed samples are from rather typical rainfall events that are large-scale,slow-moving and have well-defined genesis and dissipative stages.

A MODEL FOR QUANTITATIVELY ESTIMATING SHORTRANGE PRECIPITATION BASED ON GMS DIGITALIZED CLOUD MAPS-PART Ⅱ:ASSESSMENT OF RESULTS FROM THE MODEL FOR SATELLITE-IMAGE QUANTITATIVELY ESTIMATING PRECIPITATION

This is Part Ⅱ of this series.It introduces the technique for recognizing MαCS phased properties and its precipitation center or centers by means of dynamic digitalized cloud maps and presents the assessment of the effectiveness of the model proposed in Part Ⅰ as to its fitting and forecasting accuracy.

一种可用于登陆台风定量降水估计(QPE)方法的初步建立

借鉴Adler-Negri[1]、Goldenberg等[2]及李俊等[3]的工作,通过对三者工作的有机结合及完善,针对登陆台风GMS-5 IR1TBB特征及逐时观测雨量强度及水平分布特点,初步建立一种可用于登陆台风的定量降水估计(QPE)方法,并结合0104号登陆台风“尤特”个例,从各站点逐时雨量、过程雨量以及区域面雨量角度,分析检验了初步建立的云估计降水方法的定量估计能力。结果表明:(1)所建QPE方法可以反映出登陆台风逐时降水的水平分布不均匀性,可以分离出对流降水和层云降水,但对大于15.0 mm/h的降水强度估计能力有限。(2)51.7%的站点过程雨量相对误差小于20%,过程雨量相对误差小于40%的站点数占总站点数的75.9%,表明所建QPE方法对过程雨量的估计能力还是相当强的,这也间接反映了其对逐时雨量较强的估计能力。(3)所建QPE方法对逐时面雨量也具有一定的估计能力,可以为抗旱、防洪决策服务提供一定的参考。

基于QPE和QPF的遗传神经网络洪水预报试验

以湖北省清江上游水布垭控制流域为例,利用分组Z-I关系并结合地面雨量站资料对雷达估算降水进行校准,计算出流域实况平均面雨量;再利用遗传算法和神经网络相结合的方法建立订正AREM预报降水的模型;最后,将订正前后的AREM预报降水输入新安江水文模型进行洪水预报试验。结果表明:订正后AREM预报降水能明显提高过程的累计降水量预报精度,平均相对误差减小幅度在60%以上,对逐小时过程降水预报精度也有一定提高,但与实况相比仍有一定差距;订正前后AREM预报降水的洪水预报试验的确定性系数的场次平均从-32.6%提高到64.38%,洪峰相对误差从39%减小到25.04%,确定性系数的提高效果优于洪峰相对误差,整体上洪水预报精度有所提高。

基于FY-4A QPE的中亚五国降水时空分布特征

FY-4A定量降水估计产品(Quantitative PrecipitationEstimation,QPE)为深入研究中亚五国降水的时空分布特征提供了数据源。本文首先采用全球降水观测(Global PrecipitationMeasurement,GPM)多星集成降水终级产品IMERG-F(Integrated Multi-satellite RetrievalsforGPM Final run)评估FY-4A QPE,然后利用FY-4A QPE分析中亚五国的降水特点及时空分布特征,结果表明:(1)FY-4A QPE能够精细地反映中亚五国降水的空间分布差异,降水估计结果比较合理且与IMERG-F的时序变化具有较好的一致性。(2)中亚五国年平均降水量的空间分布差异大,且与海拔高度有关,高海拔地区的年平均降水量超过500 mm,但面积占比不足10%;低海拔地区的年平均降水量不足350 mm,但面积占比却超过90%。(3)中亚五国降水的空间分布有明显的季节性,夏季降水范围最广,平均降水量超过50 mm;秋季平均降水量最小,绝大部分地区平均降水量不足40 mm。吉尔吉斯斯坦和塔吉克斯坦四季降水相对充足,部分区域季节平均降水量超过480 mm;哈萨克斯坦中西部、乌兹别克斯坦中西部和土库曼斯坦北部季节平均降水量不足40 mm。(4)根据月平均降水量超过40 mm区域的聚集度,中亚五国月平均降水的空间分布可以大致分为点状离散分布型、干旱型、半干半湿型和三明治型四种分布形态。(5)中亚五国夏季降水多发区的逐小时平均降水量具有“准三小时”周期性日变化特征,午后至前半夜是降水多发时段之一,降水类型以小雨为主,其次是少量的中雨。

水利测雨雷达的应用优势与挑战

监测近地面2 km以下“云中雨”的测雨雷达,是水利行业以流域为单元建设雨水情监测预报“第一道防线”的重要内容,也是实现洪水防御关口前移、防线外推的重要举措。总结了水利测雨雷达的发展历程和技术特点,评估了河北大清河流域和湖南捞刀河、浏阳河流域试点应用中测雨雷达对强降水监测的精度,阐述了水利测雨雷达在雨水情监测预警方面的技术优势,并对进一步发展水利测雨雷达监测系统提出了建议。

X波段相控阵雷达QPE产品在强降水中的应用

对X波段相控阵雷达定量降水估测产品在东莞两次不同类型强降水过程中的适用性进行评估。结果表明:X波段相控阵雷达QPE产品能够刻画出降水的空间分布特征,但对雨量的估测整体偏高。对于不同降雨率,雷达的估测偏差有所不同,其中对较弱的降水估测存在湿偏差,对较强的降水估测存在干偏差;且台风降水过程中较弱降水的湿偏差比季风降水过程小,而较强降水的干偏差比季风降水过程大。不同类型降水过程由湿偏差转为干偏差的降雨率节点不同,主要是雷达QPE对不同类型过程中1~2 mm的5 min雨量估测偏差的差异造成的。两次过程雷达QPE对降雨率为10 mm/h以上的降水估测精度均更高。相比于R(ZH)关系,雷达QPE产品更能反映出降水的增强和减弱,特别是对较强降水的估测更好,在强降水预警方面应有更好的参考作用。

利用卷积神经网络开展偏振雷达定量降水估测研究

利用偏振升级改造后的广州新一代天气雷达(CINRAD/SAD)水平反射率ZH、差分传播相移率KDP、差分反射率因子ZDR和广东佛山219个地面气象自动站雨量数据,形成不同偏振量组合的8个数据集。基于卷积神经网络(CNN),建立雷达定量降水估测网络架构QPEnet,并将该架构用于雷达定量降水估测(QPE),评估结果表明:数据集通道数N的增加可降低QPEnet的定量降雨估测的均方根误差(RMSE),并提高相关系数(CORR);对于由ZH形成的数据集Z、Z_1~3 km和Z_6 min,随着通道数N的增加,数据集Z、Z_1~3 km和Z_6 min的性能逐步得到提高,数据集Z_1~3 km和Z_6 min的均方根误差(RMSE)分别是4.71和3.78,比数值集Z分别降低了1.3%和18.7%;数据集Z_1~3 km和Z_6 min的CORR分别是0.82和0.88,比数据集Z分别提高了2.5%和10.0%;对于ZH、KDP和ZDR偏振量组成的数据集里面,数据集Z_ZDR_KDP的拟合性能最好,RMSE为3.97,比数据集Z的RMSE降低了14.6%,CORR是0.86,比数据集Z提高了7.5%;分别对0.6~5 mm、5~10 mm、10~20 mm、20~30 mm、30~40 mm、40~50 mm和50 mm以上的7个降水量级的均方根误差(RMSE)、平均偏差比(MBR)、平均误差(AE)和相对误差(RE)等的统计结果表明,数据集Z_6 min降雨精度最高。

基于雨滴谱参数反演的C波段双偏振雷达降水类型分类方法

降水类型分类对分析区域降水微物理特征、多源降水融合误差模型的构建以及雷达定量测量降水估计等都很重要。本文基于2015~2016年南京信息工程大学C波段双偏振雷达数据和南京地区滴谱仪观测资料,提出一种适用于南京地区的雷达降水类型分类方法,并对降水类型分类结果进行对比验证。首先,基于滴谱仪降雨率时序数据、地基雷达反射率因子平面位置显示数据和地基雷达反射率因子时间—高度显示数据,筛选出36次典型层状和对流降水过程。随后,统计3个滴谱仪站点典型层状(对流)降水的雨滴谱(DSD)参数,拟合得到适用于南京地区的降水类型分类线。将基于滴谱数据统计拟合的分类线应用于基于变分法反演的地基雷达DSD参数,进行地基雷达降水类型分类。根据典型层状(对流)过程降水类型分离指数的时间—高度分布,并对比星载双频测雨雷达(DPR)降水分类产品,对分类效果进行验证。最后,将分类结果应用于雷达分类定量降水估计,进一步说明降水分类的应用效果。结果表明,南京地区3个滴谱仪站点的拟合分类线非常一致,3个站点的典型层状(对流)过程均能够很好地分离在分类线两侧;与DPR降水分类产品进行对比分析,发现南京地区分类线的分类效果相对于其他典型降水分类方法,对层状和对流降水的识别率整体最高,分别为84.56%和72.64%;基于降水分类的雷达定量降水估计的测雨精度均优于未分类的测雨公式,且基于差分传播相移率的测雨公式[R(KDP)]在四种分类测雨公式中整体性能最优,基于水平反射率因子的测雨公式[R(ZH)]在层状云降水反演中性能最优,基于差分传播相移率的测雨公式[R(KDP)]在对流云降水反演中性能最优,基于水平反射率因子和差分反射率的测雨公式[R(ZH,ZDR)]对原有总体测雨公式降水精度的提升最为明显。