The impacts of optical properties on radiative forcing due to dust aerosol

There are large uncertainties in the quantitative assessment of radiative effects due to atmospheric dust aerosol. The optical properties contribute much to those uncertainties. The authors perform several sensitivity experiments to estimate the impacts of optical characteristics on regional radiative forcing in this paper. The experiments involve in refractive indices, single scattering aibedo, asymmetry factor and optical depth. An updated dataset of refractive indices representing East Asian dust and the one recommended by the World Meteorology Organization （WMO） are contrastively analyzed and used. A radiative transfer code for solar and thermal infrared radiation with detailed aerosol parameterization is employed. The strongest emphasis is on the refractive indices since other optical parameters strongly depend on it, and the authors found a strong sensitivity of radiative forcing on refractive indices. Studies show stronger scattering, weaker absorption and forward scattering of the East Asian dust particles at solar wavelengths, which leads to higher negative forcing, lower positive forcing and bigger net forcing at the top of the atmosphere （TOA） than that of the WMO dust model. It is also found that the TOA forcings resulting from these two dust models have opposite signs in certain regions, which implies the importance of accurate measurements of optical properties in the quantitative estimation of radiative forcing.     

利用深度神经网络和先兆信号的江苏夏季降水客观预测方法

利用1961—2019年江苏省67个站降水量和气候指数数据集等资料,选取大气环流、海温和积雪等先兆信号的不同组合作为预测因子方案,通过对比不同机器学习方法对江苏省夏季降水开展预测试验。结果表明,深度神经网络（Deep Neural Network,DNN）较传统统计方法和其他机器学习方法有一定优势,深度神经网络结合动态权重集合因子方案对江苏省夏季降水的预测技巧最高,其独立样本检验结果稳定,2015—2019年的平均PS评分为76.0,距平符号一致率为0.62,距平相关系数达0.35,尤其对江苏省中南部的预测技巧更高,具有业务应用价值。不同预测因子方案对比分析表明,大气环流因子在江苏省夏季降水预测中做主要贡献,而海温因子和积雪等其他因子也有正贡献,说明使用综合性预测因子以及集合方案有助于提升季节预测准确率。      

Single scattering from frozen hydrometeors at microwave frequencies

Scattering of electromagnetic waves from homogeneous or coated spheres can be computed in a mathematically exact way using the Mie theory. Therefore, for many approaches in remote sensing, frozen hydrometeors are parameterized as ice spheres. However, many frozen hydrometeors have non-spherical overall shapes and lack a spherically symmetric internal structure. They exist in a huge variety of shapes and exhibit different mixtures of ice, water and air. Therefore it is desirable to accurately compute scattering from non-spherical particles in order to clearly understand the effect the shape of a hydrometeor has on its scattering pattern.In this study, single scattering parameters like scattering cross section, absorption cross section, and asymmetry factor were calculated for frozen hydrometeors using the Discrete Dipole Approximation (DDA). The particles were modeled as hexagonal plates, columns, needles and dendrites by applying known dimensional relationships. The calculations were carried out over a wide range of centimeter and millimeter-wavelengths (1 GHz to 300 GHz), since millimeter-wave radiometers are highly sensitive to scattering by frozen hydrometeors in the atmosphere.The study results show that for size parameters < 1 (a ratio between wavelength and particle size) the scattering cross section of randomly orientated ice crystals is close to that of an equal volume ice sphere. Absorption cross section and asymmetry factor of non-spherical particles however are up to twice as high as that of equal volume ice spheres. Further the influence of the assumed model for the refractive index of ice at microwave wavelengths on the scattering parameters is investigated.     

基于三种机器学习方法的降水相态高分辨率格点预报模型的构建及对比分析

冬季降水相态及其转变时间的精细化客观预报对提高气象预报和服务质量具有重要的现实意义。利用京津冀地区国家级自动气象站观测资料及网格化快速更新精细集成产品,统计分析了京津冀地区复杂地形下各类降水相态温度和湿球温度平均气候概率的分布差异及不同降水相态时网格化快速更新精细集成产品中可能影响降水相态判断的特征信息。然后将地面观测天气现象资料、复杂地形下降水相态气候特征及高分辨率模式输出产品作为特征向量,分别基于梯度提升（XGBoost）、支持向量机（SVM）、深度神经网络（DNN）3种机器学习方法建立了降水相态的高分辨率客观分类模型,并对同样条件下3种机器学习方法对雨、雨夹雪和雪3种京津冀主要降水相态的预报效果进行了对比检验,进一步提升了雨夹雪复杂降水相态的客观分类预报技巧。      

冬奥延庆赛区风的预报技巧比较分析

为了提高2 min平均的10 m风预报精度,开展了多种建模和检验方法比较.根据欧洲数值中心集合预报系统产品及北京海陀山的5个测站资料,使用一元回归、岭回归、神经网络、粒子群-神经网络等方法建模,进行2021年2月逐日的未来3天6 h间隔预报误差订正,并从多个角度分析预报精度差异.结果为:(1)系统误差、预报准确率检验表...     

深度学习方法在干旱预测中的应用

使用标准化降水蒸散指数作为农业干旱的监测指标,以站点气象要素和大尺度环流要素为驱动变量,建立干旱预测模型,分析评价传统的整合移动平均自回归(ARIMA)时间序列模型以及不同深度神经网络模型(DNN)的预测效果.结果 表明:DNN模型的总体预测能力优于ARIMA模型;同基于长短期记忆网络(LSTM)提出的传统LSTM预测...     

An Innovative Bias-Correction Approach to CMA-GD Hourly QuantitativePrecipitation Forecasts

This paper proposes a simple and powerful optimal integration (OPI) method for improving hourly quantitative precipitation forecasts (QPFs, 0-24 h) of a single-model by integrating the benefits of different bias- corrected methods using the high-resolution CMA-GD model from the Guangzhou Institute of Tropical and Marine Meteorology of China Meteorological Administration (CMA). Three techniques are used to generate multi-method calibrated members for OPI: deep neural network (DNN), frequency-matching (FM), and optimal threat score (OTS). The results are as follows: (1) The QPF using DNN follows the basic physical patterns of CMA-GD. Despite providing superior improvements for clear-rainy and weak precipitation, DNN cannot improve the predictions for severe precipitation, while OTS can significantly strengthen these predictions. As a result, DNN and OTS are the optimal members to be incorporated into OPI. (2) Our new approach achieves state-of-the-art performances on a single model for all magnitudes of precipitation. Compared with the CMA-GD, OPI improves the TS by 2.5%, 5.4%, 7.8%, 8.3%, and 6.1% for QPFs from clear-rainy to rainstorms in the verification dataset. Moreover, OPI shows good stability in the test dataset. (3) It is also noted that the rainstorm pattern of OPI relies heavily on the original model and that OPI cannot correct for deviations in the location of severe precipitation. Therefore, improvements in predicting severe precipitation using this method should be further realized by improving the numerical model’s forecasting capability.     

ConvLSTM Based Temperature Forecast Modification Model for North China

The correction of model forecast is an important step in evaluating weather forecast results. In recent years,post-processing models based on deep learning have become prominent. In this paper, a deep learning model named ED-ConvLSTM based on encoder-decoder structure and ConvLSTM is developed, which appears to be able to effectively correct numerical weather forecasts. Compared with traditional post-processing methods and convolutional neural networks, ED-ConvLSTM has strong collaborative extra...     

基于雷达组合反射率拼图和深度学习的中尺度对流系统识别、追踪与分类方法

中尺度对流系统（Mesoscale Convective System,MCS）是很多对流性天气的主要致灾体,可导致严重的气象和水文灾害,如雷暴大风、冰雹、龙卷风和山洪。对MCS进行准确的识别和追踪,并根据追踪轨迹及获得的MCS特征实现MCS的分类,对灾害天气的分析和预报有重要意义。基于京津冀地区2010—2019年的雷达组合反射率拼图资料,分别使用支持向量机（SVM）、随机森林（RF）、极度梯度提升决策树（XGBoost）和深度神经网络（DNN）4种机器学习方法,研发了京津冀地区MCS的自动识别算法。使用时、空重叠追踪法对识别的MCS进行追踪匹配,得到包含强度、时间和空间信息的MCS追踪数据资料。在区分线状对流系统和非线状对流系统的基础上,进一步从经典的尾随层云（Trailing Stratiform,TS）、前导层云（Leading Stratiform,LS）和平行层云（Parallel Stratiform,PS）三类准线性MCS的概念模型和结构特征出发,根据追踪轨迹计算MCS的运动方向和MCS近似长轴两侧层状云和强对流云的面积占比,建立准线性MCS的分类算法。MCS的识别属于二分分类问题,以命中率（POD）、虚警率（FAR）、临界成功指数（CSI）和准确率（ACC）为评价指标,综合对比各项指标发现DNN模型较SVM、RF和XGBoost模型对MCS的识别效果更好。使用时、空重叠追踪法对DNN模型识别的MCS进行追踪,结合对两个追踪实例的分析,发现本研究所用的算法取得了很好的追踪结果,也进一步说明了深度学习方法识别MCS的准确性和优势。根据追踪轨迹计算某时刻MCS的运动方向,结合识别的层状云和强对流云的分布位置,准确实现了TS、LS和PS型准线性MCS的分类,为准线性MCS的生命史预测及其致灾天气特别是短时强降水的强度、位置和持续时间的客观预报提供了一种技术思路。      

基于BP神经网络模型的雷电潜势预报

为了使用神经网络较好地解决在雷电潜势预报中常见的非线性问题,本文通过计算南京地区2008年6～8月46个对流参数与雷电发生的相关系数,选取了与雷电发生关系较好的刀、SI、CIN等7个对流参数作为BP神经网络的输入因子。利用2008年的资料所建立的BP神经网络模型,预报了南京地区2009年6～8月的雷暴活动潜势,结合实际雷暴发生情况,得到此模型的POD为80．9％,FAR为9．5％,CSI为74．5％,PDFD为2．9％,FOM为19．1％。表明该BP模型预报准确率较高,性能稳定,有较好的推广价值。     

Identification of Aerosol Types and Their Optical Properties in the North China Plain Based on Long-Term AERONET Data

Characterization of aerosols is required to reduce uncertainties in satellite retrievals of global aerosols and for modeling the effects of these aerosols on climate.Aerosols in the North China Plain(NCP) are complex,which provides a good opportunity to study key aerosol optical properties for various aerosol types.A cluster analysis of key optical properties obtained from Aerosol Robotic Network(AERONET) data in Beijing and Xianghe during 2001-11 was performed to identify dominant aerosol types and their associated optical properties.Five dominant aerosol types were identified.The results show that the urban/industrial aerosol of moderate absorption was dominant in the region and that this type varied little with season.Urban/industrial aerosol of weak absorption was the next most common type and mainly occurs in summer,followed by that strong aerosols occurring mainly in winter.All were predominantly fine mode particles.Mineral dust(MD) and polluted dust(PD) occurred mainly in spring,followed by winter,and their absorption decreased with wavelength.In addition,aerosol dynamics and optical parameters such as refractive index and asymmetry factor were examined.Results show that the size of coarse mode particles decreased with AOD indicating the domination of external mixing between aerosols.     

Spectral absorption coefficients and imaginary parts of refractive indices of Saharan dust during SAMUM-1

During the SAMUM-1 experiment, absorption coefficients and imaginary parts of refractive indices of mineral dust particles were investigated in southern Morocco. Main absorbing constituents of airborne samples were identified to be iron oxide and soot. Spectral absorption coefficients were measured using a spectral optical absorption photometer (SOAP) in the wavelength range from 300 to 800 nm with a resolution of 50 nm. A new method that accounts for a loading-dependent correction of fibre filter based absorption photometers, was developed. The imaginary part of the refractive index was determined using Mie calculations from 350 to 800 nm. The spectral absorption coefficient allowed a separation between dust and soot absorption. A correlation analysis showed that the dust absorption coefficient is correlated ( R ² up to 0.55) with the particle number concentration for particle diameters larger than 0.5 μm, whereas the coefficient of determination R ² for smaller particles is below 0.1. Refractive indices were derived for both the total aerosol and a dust aerosol that was corrected for soot absorption. Average imaginary parts of refractive indices of the entire aerosol are 7.4 × 10⁻³, 3.4 × 10⁻³ and 2.0 × 10⁻³ at wavelengths of 450, 550 and 650 nm. After a correction for the soot absorption, imaginary parts of refractive indices are 5.1 × 10⁻³, 1.6 × 10⁻³ and 4.5 × 10⁻⁴.     

典型非球形冰晶粒子的凝华增长数值模拟试验

在凝华增长过程中,冰晶的形状随着温度和湿度的改变而改变,准确模拟冰晶粒子的演变对于提高云模式的模拟能力起着非常重要的作用。在现有的云模式中,冰晶形状通常假设为球形,而在实际大气中,冰晶形状十分复杂。本研究中,我们根据冰晶凝华增长理论模型建立了一个单个冰晶粒子增长模型,模拟了温度分别为-1℃~-30℃时,单个典型非球形冰晶粒子的凝华增长过程。与风洞观测数据相对比,该模型能够抓住单个冰晶粒子的轴长,质量以及纵横比随温度和湿度的变化过程。我们进一步将该理论增长模型应用到群粒子的凝华增长过程的模拟。我们釆用欧拉二维正定平流输送法(MPDATA)模拟了典型非球形冰晶群粒子的凝华增长,并对比分析了在不同纵横比分辨率下的模拟效果以及温度变化对冰晶形状的影响,结果表明运用该数值方法可以合理地模拟出群粒子在凝华增长过程中纵横比的演变。与目前采用的拉格朗日-欧拉混合平流算法比较,该算法能够耦合到欧拉动力框架下的分档云模式中去,这对我们研究冰晶粒子形状对云微物理过程和动力过程的影响,以及它们对冰粒子凝华增长的反馈作用具有非常重要的科学意义。     

热带印度洋-太平洋地区海表温度异常的线性转置预报模型

采用ＣＯＡＤＳ的月平均海平面温度跨平（ＳＳＴＡ）资料,建立了预报热带印度洋－太平洋ＳＳＴＡ的线性转置模型（ＬＩＭ）。经检验,对于非独立和独立样本,预报的均方根误差分别在１２个月和１０个月预报时效内小于ＳＳＴＡ的均方差,相对误差在５个月预报时效内都小于５０％。在Ｅ１Ｎｉｎｏ和Ｌａ Ｎｉｎａ时段效果优于其他时段,其中Ｌａ Ｎｉｎａ时段又更好些。对同一地区的ＳＳＴＡ,ＬＩＭ预报优于自回归模型预报、奇异谱     

东亚沙尘源区与下游地区气溶胶光学特性比较

利用AERONET观测资料从气候学的角度比较分析了2001-2011年东亚地区沙尘天气发生时沙尘源区和下游区大气气溶胶光学特性。结果表明：沙尘期间沙尘源区气溶胶光学厚度明显大于下游区,而Angstr?m波长指数却小于下游区,当沙尘暴出现时会降至零甚至负值。气溶胶粒子尺度体积谱分布除敦煌为单峰外,其余各站均呈双峰分布,香河和北京的细粒子浓度明显大于西北地区,这可能是由细的沙尘粒子和污染气溶胶共同造成。在440-1020 nm范围内,中国地区气溶胶单次散射反照率平均值为0.93,韩国和日本站分别为0.93和0.94。沙尘源区与下游区相比,复折射指数实部偏大,虚部偏小。总体来说,沙尘天气下东亚地区在4个波段内平均不对称因子为0.70。     

Annual precipitation forecast for west, southwest, and south provinces of Iran using artificial neural networks

Rainfed agriculture plays an important role in the agricultural production of the southern and western provinces of Iran. In rainfed agriculture, the adequacy of annual precipitation is considered as an important factor for dryland field and supplemental irrigation management. Different methods can be used for predicting the annual precipitation based on climatic and non-climatic inputs. Among which artificial neural networks (ANN) is one of these methods. The purpose of this research was to predict the annual precipitation amount (millimeters) in the west, southwest, and south of Islamic Republic of Iran with the total area of 394,259?km², by applying non-climatic inputs according to the long-time average precipitation in each station (millimeters), 47.5?mm precipitation since the first of autumn (day), t _47.5, and other effective parameters like coordinate and altitude of the stations, by using the artificial neural networks. In order to intelligently estimate the annual amount of precipitation in the study regions (ten provinces), feedforward backpropagation artificial neural network model has been used (method I). To predict the annual precipitation amount more accurately, the region under study was divided into three sub-regions, according to the precipitation mapping, and for each sub-region, the neural networks were developed using t _47.5 and long-time average annual precipitation in each station (method II). It is concluded that neural networks did not significantly increase the prediction accuracy in the study area compared with multiple regression model proposed by other investigators. However, in case of ANN, it is better to use a structure of 2–6–6–10–1 and Levenberg–Marquardt learning algorithm and sigmoid logistic activation function for prediction of annual precipitation.     

Aerosol polarized phase function and single-scattering albedo retrieved from ground-based measurements

Aerosol optical parameters, polarized phase function and single-scattering albdeo, have been retrieved from ground-based sun photometer measurements in Beijing 2003. The measured aerosol optical thickness varies from 0.12 to 0.77 with an average value of 0.39. The measured Ångström coefficient ranges from 0.75 to 1.47 with an average value of 1.21. The retrieved single-scattering albedo at 870 nm is within the 0.76–0.94 range and the average value is 0.85, suggests there are considerable aerosol absorptions in Beijing. The maximum value of retrieved polarized phase function at 870 nm ranges from 0.068 to 0.225 with an average value of 0.16, and it illustrates good correlations with the Ångström coefficient, i.e. the relative size of aerosol particles. Analyses of measurements and theoretical calculations show the polarized phase function is sensitive to aerosol size distribution and complex refractive index, especially the imaginary part of the refractive index which denotes aerosol light absorbing effects. These results suggest that the polarized phase function is an effective and unique aerosol optical parameter and is able to improve the retrieval of aerosol physical properties.     

利用深度学习预报美国东北部日降水分布

现阶段降水预报主要依靠数值天气预报模式。但受物理参数化、计算资源等因素的影响,基于数值模式的降水预报还存在非常大的不确定性。近年来,深度学习在天气预报领域显示出巨大优势和潜力。本文通过构建神经网络预报美国东北部日降水分布,探讨神经网络模型基于低分辨率气象场(ERA-Interim, 0.7°)预报高分辨率降水(CPC,0.25°)的能力,并比较3种主流网络框架(VGG,ResNet, GoogleNet)在该任务中的表现。结果表明,3种网络框架都对美国东北部日降水分布具有一定的预报能力(VGG框架表现最优),但三者的均方根误差(RMSE)均高于ERA-Interim 24-h(ERA24)的降水预报。3种神经网络的集合预报结果优于ERA24预报,且这三者与ERA24预报结果的集合平均能够显著提高ERA24对不同季节、不同强度降水的预报。     

Least squares support vector machine for short-term prediction of meteorological time series

The prediction of meteorological time series plays very important role in several fields. In this paper, an application of least squares support vector machine (LS-SVM) for short-term prediction of meteorological time series (e.g. solar irradiation, air temperature, relative humidity, wind speed, wind direction and pressure) is presented. In order to check the generalization capability of the LS-SVM approach, a K-fold cross-validation and Kolmogorov–Smirnov test have been carried out. A comparison between LS-SVM and different artificial neural network (ANN) architectures (recurrent neural network, multi-layered perceptron, radial basis function and probabilistic neural network) is presented and discussed. The comparison showed that the LS-SVM produced significantly better results than ANN architectures. It also indicates that LS-SVM provides promising results for short-term prediction of meteorological data.     

Research on the Application of the Radiative Transfer Model Based on Deep Neural Network in One-dimensional Variational Algorithm

As a typical physical retrieval algorithm for retrieving atmospheric parameters, one-dimensional variational(1 DVAR) algorithm is widely used in various climate and meteorological communities and enjoys an important position in the field of microwave remote sensing. Among algorithm parameters affecting the performance of the 1 DVAR algorithm, the accuracy of the microwave radiative transfer model for calculating the simulated brightness temperature is the fundamental constraint on the retrieval ...