自组织映射神经网络(SOM)降尺度方法对江淮流域逐日降水量的模拟评估

利用1961~2002年ERA-40逐日再分析资料和江淮流域56个台站逐日观测降水量资料,引入基于自组织映射神经网络(Self-Organizing Maps,简称SOM)的统计降尺度方法,对江淮流域夏季(6~8月)逐日降水量进行统计建模与验证,以考察SOM对中国东部季风降水和极端降水的统计降尺度模拟能力。结果表明,SOM通过建立主要天气型与局地降水的条件转换关系,能够再现与观测一致的日降水量概率分布特征,所有台站基于概率分布函数的Brier评分(Brier Score)均近似为0,显著性评分(Significance Score)全部在0.8以上;模拟的多年平均降水日数、中雨日数、夏季总降水量、日降水强度、极端降水阈值和极端降水贡献率区域平均的偏差都低于11%;并且能够在一定程度上模拟出江淮流域夏季降水的时间变率。进一步将SOM降尺度模型应用到BCCCSM1.1(m)模式当前气候情景下,评估其对耦合模式模拟结果的改善能力。发现降尺度显著改善了模式对极端降水模拟偏弱的缺陷,对不同降水指数的模拟较BCC-CSM1.1(m)模式显著提高,降尺度后所有台站6个降水指数的相对误差百分率基本在20%以内,偏差比降尺度前减小了40%~60%;降尺度后6个降水指数气候场的空间相关系数提高到0.9,相对标准差均接近1.0,并且均方根误差在0.5以下。表明SOM降尺度方法显著提高日降水概率分布,特别是概率分布曲线尾部特征的模拟能力,极大改善了模式对极端降水场的模拟能力,为提高未来预估能力提供了基础。     

锦州地区逐日降水概率分布特征

利用锦州地区的逐日降水量观测资料对逐日降水量的概率分布进行了统计分析,采用最大似然估计法得到Gamma函数分布的形状参数α和尺度参数β,通过Gamma概率分布模拟观测站点逐日降水的概率分布。结果表明：锦州地区逐日降水频率整体趋势先上升后下降,基本呈对称式分布,降水概率有一定的振荡,个别日会出现远超相邻日期的降水频率,7月21日降水频率最高,在不计微量降水的情况下,最低逐日降水概率有多个日期为0。各季降水频率偏低是造成义县地区干旱的原因之一;北镇夏季平均降水频率最低,但其夏季平均降水量却为锦州地区最高,说明北镇可能易出现较大量级降水或易出现极端降水天气。清明期间降水频率在50%以上、高考期间降水频率在80%以上,符合大众日常对特殊日期降水情况的认知;逐日降水频率可以为公众气象服务提供新的思路。凌海、北镇更容易出现极端降水天气;锦州地区日降水出现小雨天气概率最高,暴雨以上降水概率较低,锦州地区各站极少出现大暴雨以上量级降水,对锦州降水量级预报,尤其是暴雨或大暴雨以上降水量级的预报起到一定的指示作用。     

基于频率匹配法的中国降水多模式预报订正研究

基于TIGGE资料中欧洲中期天气预报中心、日本气象厅、美国国家环境预报中心及英国气象局1～7 d日降水量预报以及中国自动站观测资料与CMORPH降水产品融合的逐时降水量网格数据集,利用频率匹配法(Frequency-Matching Method,FMM)对中国降水预报进行客观订正。首先利用卡尔曼滤波方法对降水频率进行调整,并根据不同区域降水强度差异将全国分为7个子区域分别进行频率匹配。结果表明,FMM可以有效减小降水量预报的误差。经过频率匹配法订正后各模式降水预报的平均绝对误差(Mean Absolute Error,MAE)大幅减小,且订正后各量级降水的雨区面积更加接近实际观测值。FMM对小于5 mm和大于15 mm的降水预报技巧改进明显。此外,FMM降低了模式预报的小雨空报率和大雨漏报率,并且明显提高了晴雨预报的准确率。FMM明显消除了大范围小雨空报区域,但是对强降水预报FMM仅能调整降水量大小,强降水落区预报并不能得到明显改善。     

中国东南地区复杂地形下降水概率预报的订正研究

使用TIGGE (the THORPEX interactive grand global ensemble)资料集下欧洲中期天气预报中心(the European Centre for Medium-Range Weather Forecasts,ECMWF)逐日起报的预报时效为24～168 h的日降水量集合预报资料,集合预报共包括51个成员,利用左删失的非齐次Logistic回归方法(left-Censored Non-homogeneous Logistic Regression,CNLR)和标准化的模式后处理方法(Standardized Anomaly Model Output Statistics,SAMOS)对具有复杂地形的中国东南部地区降水预报进行统计后处理。结果表明:采用CNLR方法能够有效改进原始集合预报的平均绝对误差(Mean Absolute Error,MAE)和连续分级概率评分(Continuous Ranked Probability Score,CRPS),提升了降水的定量预报和概率预报的预报技巧。而使用SAMOS方法对数据进行预处理,考虑地形等因素的影响,能在CNLR方法的基础上进一步订正由于地形影响造成的预报误差,并得到更加准确的全概率的降水概率预报。     

贵州地区CMORPH卫星降水产品的误差订正

CMORPH卫星反演降水产品具有全天候、全球覆盖的特点,其时空分布相对均匀、独立,但是CMORPH本质上是通过间接手段反演得到,其降水精度无法与地面观测降水精度相比,并且存在一定的系统误差。结合地面自动站降水资料采用概率密度匹配法对贵州地区CMORPH卫星反演降水产品进行系统误差订正,该方法将每个格点的卫星降水累积概率分布曲线和地面降水概率密度分布曲线匹配,获取降水误差订正值;其中误差订正效果受降水累积概率分布拟合曲线的影响,而考虑到降水累积概率分布是非正态分布,因此选用Gamma分布拟合降水累积概率分布曲线。通过对2018年5月三次降水过程的订正结果分析得到如下结论:(1) 逐时的CMORPH卫星反演降水产品存在明显的非独立系统误差,误差范围随降水量级的变化而变化,存在低值高估的特点;(2) 在小时尺度下地面降水的累积概率密度呈指数衰减分布,而CMORPH的降水累积概率密度分布更加复杂,其在中雨、大雨区间内的降水概率较高;(3) 通过概率密度匹配法订正后的CMORPH与订正前相比降水空间结构更加贴近地面降水,强降水中心的量级和范围明显减小,平均绝对误差和均方根误差均减小,其中偏差订正值在0.114~0.468 mm/h,均方根误差订正在0.24~1.49 mm/h之间。经概率密度匹配法订正后的CMORPH卫星反演降水产品精度明显提升,更加接近于实际降水。      

近32年泰国降水的主要变化趋势研究

基于泰国气象局提供的近32年(1981～2012)站点逐日降水观测资料,利用线性趋势和集合经验模态分解(Ensemble empirical mode decomposition,EEMD)等分析方法,本文重点研究了泰国及其五个地理分区内各等级降水量与降水日数出现正异常(第95百分位及以上)的站点比例变化,并深入分析了...     

CLIGEN天气发生器模拟沈阳地区降水的适用性评价

利用1981—2015年沈阳地区7个气象站的日观测数据,通过CLIGEN(Climate Generater)天气发生器模拟沈阳地区日降水序列数据,并统计模拟日降水量、月降水量、年降水量及年最大日降水量,利用平均值、标准差、偏度及峰度对CLIGEN天气发生器模拟的沈阳地区降水进行适用性评价。结果表明:CLIGEN天气发生器对沈阳地区日降水量、月降水量和年降水量平均值的模拟效果较好,模拟降水量的平均相对误差绝对值(Mean Absolute Relative Error,MARE)分别为2.1%、1.3%和3.3%,年最大日降水量的模拟精度稍差。对于降水最大值方面,CLIGEN天气发生器对沈阳地区日最大降水量和年最大降水量的模拟效果较差,模拟的日最大降水量和年最大降水量相对误差的最大值分别为-27.2%、18.3%。GLIGEN天气发生器能较好地模拟沈阳地区的月降水量和年降水量,t检验、F检验和K-S检验均表明,模拟的日最大降水量与年最大日降水量仅康平站达到极显著水平。从总体模拟效果来看,CLIGEN天气发生器能较好地模拟沈阳地区平均降水的统计特征。     

陕西新一代天气雷达与自动站联合估测降水质量评估

利用2016年和2017年共5次降水过程数据,对天气雷达-自动站联合估测降水和自动站降水、雷达OHP产品进行对比分析,并给出联合估测降水拟合的Z-R关系。结果表明:自动站1h降水量≤1mm时,联合估测降水的平均相对误差大,但均方根误差在可接受范围内,且联合估测降水量非常接近自动站降水量;自动站1h降水量1mm时,联合估测降水的效果较好,其中20mm时联合估测降水的精度最好,平均相对误差低于8%;自动站1h降水量1mm时,联合估测降水量优于雷达OHP(1h累积降水量)产品;同一次降水过程,不同时次的Z-R关系不同,无明显变化规律。     

短中期降水温度天气过程区域分布的研究

运用旋转主分量方法对分布于全国621个站点的候降水量和2214个站的逐日最高最低气温进行分析。得到4个季节的降水气温主特征模态及其相对应的时间变率。分析结果表明，该方法所分解得到的特征模态较好地反映了全国不同区域降水温度演变的差别，依据各模态的相关系数将全国划分为不同的降水温度天气区。最后。获得表征全国不同天气特点的252个代表站，这些站的资料可作为提高要素客观预报、扩展服务领域的基本资料。     

概率密度匹配法对中国区域卫星降水资料的改进

为考察概率密度匹配法 (PDF方法) 对中国区域卫星反演降水产品系统误差订正的适用性,基于逐日和逐时我国地面观测降水量资料,引入PDF方法,分别对逐日0.25°×0.25°水平分辨率和逐时0.1°×0.1°水平分辨率的CMORPH (Climate Prediction Center Morphing Technique) 卫星降水产品的系统误差进行订正。在分析CMORPH卫星降水产品误差特征的基础上,根据两种资料不同的时空分辨率和误差特点,调整概率密度匹配时选取样本的时间和空间范围,设计相应的订正方案。评估结果表明: PDF方法订正后, 两种分辨率卫星降水资料在中国区域系统误差均显著减小,达到了理想的订正效果。在我国站点稀疏的西部地区,订正后的CMORPH卫星降水产品仍保持卫星观测的降水空间分布,降水量也明显接近于地面观测降水量。可见,PDF方法是中国区域卫星反演降水产品系统误差订正的一种有效方法。     

Station-based verification of qualitative and quantitative MM5 precipitation forecasts over Northwest Himalaya (NWH)

Weather forecasts by any forecast system are verified using either distributions-oriented or measures-oriented forecast verification measures. Both the forecast verification schemes represent different aspects of the forecast quality, and advantages of them can be utilized to get better insight and to identify particular strengths (deficiencies) in the forecast performance of any forecast system. Keeping this in view, multi-faced verification (binary and continuous) of quantitative precipitation forecasts for consecutive 3 days by a Regional Meso-scale Weather Simulation Model (MM5 Model) has been carried out to get complete insight into its performance. The MM5 model forecasts at 10-km resolution for 792 days of six winters (winter 2003/2004 to winter 2008/2009) are compared with the observational data of six stations in the complex topography of Northwest Himalaya (NWH) in India. The model forecasts are verified using binary categorical forecast verification measures such as Probability of Detection, False Alarm Rate, Miss Rate, Correct Non-occurrence, Critical Success Index and Percent correct, and continuous forecast verification measures such as Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). BIAS is computed to know over-forecast/under-forecast tendency of a precipitation day (PT day) by the MM5 model. MAE (RMSE) of the MM5 model is computed separately for all days, PT days and no precipitation days (NPT days). MAE (RMSE) of PT days is found to be relatively larger as compared to NPT days and all days. These findings indicate that MAE (RMSE) computed separately for all days, PT days and NPT days provides better insight into the performance of the MM5 model. Results also suggest that the MM5 model shows reasonably good performance for binary forecasts (PT days/NPT days) for day 1 (0–24 h), day 2 (24–48 h) and day 3 (48–72 h). However, large errors are seen in predicting the observed precipitation amount of PT days over NWH.     

基于CNN和RF算法的ECMWF降水分级订正预报方法

利用卷积神经网络(CNN)和随机森林回归模型,提出了一种新的欧洲中期天气预报中心(ECMWF)降水订正预报方法。该方法首先根据ECMWF模式对站点雨量预报值所属的等级进行划分,再计算出不同等级相对应的高相关因子矩阵。进一步利用CNN模型对高相关矩阵进行综合特征提取的学习和训练。最后对CNN模型最终输出的特征因子中,选取若干个与预报站点相关性高的特征,并与ECMWF降水量场插值到预报站点的因子一起,作为随机森林回归模型的输入因子进行预报建模。通过对10个预报试验站点未来24h降水量的分级和不分级订正预报试验,结果表明:(1)ECMWF降水量分级订正预报方法的平均绝对偏差和均方根误差分别比利用ECMWF插值到站点的预报方法减小了20%和15%;(2)24h暴雨及以上的降水分级订正预报方法的平均TS评分为0.32,也显著高于EC插值的0.19;(3)与利用同样的预报模型对全样本(不分级)的传统数值预报模式产品订正预报方法相比,本文提出的分级订正预报方法在总体预报精度和暴雨及以上的强降水预报TS评分上均有更高的预报技巧。     

基于自组织映射的长江中下游夏季天气分型及其降水特征

利用欧洲中心1979—2015年夏季6—8月ERA-Interim逐日再分析资料和国家气候中心CN05.1格点化降水观测数据集,引入基于自组织映射SOM(Self-Organizing Maps)方法进行长江中下游地区夏季海平面气压空间距平场的客观分型,得到该地区25种地面天气型及其系统演变特征,发现天气型的稳定、转移与天气系统强弱有关。高低压系统越强,天气型停滞频率越高,天气型越稳定;反之,天气型越不稳定。基于SOM天气型转移概率,发现三条与局地降水联系的系统演变路径,其中1号路径暖空气势力强盛,副高北上,推动锋面北抬,产生江北降水型,多发生在7月;路径2反映冷空气势力强盛推动锋面南下的天气过程,产生沿江降水型,该天气型在6、7月均易发生;路径3表现为台风移动变化对长江下游江南地区降水的影响,为江南降水型,且集中在8月。通过引入SOM方法对逐日天气尺度环流进行分型,从气候态上捕捉与降水有关的天气系统的移动变化特征,体现SOM方法在模拟天气系统演变的优势。     

Identification of typical synoptic patterns causing heavy rainfall in the rainy season in Japan by a Self-Organizing Map

In order to systematically and visually understand well-known but qualitative and complex relationships between synoptic fields and heavy rainfall events in Kyushu Islands, southwestern Japan, during the BAIU season, these synoptic fields were classified using the Self-Organizing Map (SOM), which can convert complex non-linear features into simple two-dimensional relationships. It was assumed that the synoptic field patterns could be simply expressed by the spatial distribution of (1) wind components at the 850 hPa level and (2) precipitable water (PW) defined by the water vapor amount contained in a vertical column of the atmosphere. By the SOM algorithm and the clustering techniques of the U-matrix and the K-means, the synoptic fields could be divided into eight kinds of patterns (clusters). One of the clusters has the notable spatial features represented by a large PW content accompanied by strong wind components known as low-level jet (LLJ). The features of this cluster indicate a typical synoptic field pattern that frequently causes heavy rainfall in Kyushu during the rainy season.In addition, an independent data set was used for validating the performance of the trained SOM. The results indicated that the SOM could successfully extract heavy rainfall events related to typical synoptic field patterns of the BAIU season. Interestingly, one specific SOM unit was closely related to the occurrence of disastrous heavy rainfall events observed during both training and validation periods. From these results, the trained SOM showed good performance for identifying synoptic fields causing heavy rainfall also in the validation period. We conclude that the SOM technique may be an effective tool for classifying complicated non-linear synoptic fields and identifying heavy rainfall events to some degree.     

Comparison of artificial intelligence techniques for prediction of soil temperatures in Turkey

Soil temperature is a meteorological data directly affecting the formation and development of plants of all kinds. Soil temperatures are usually estimated with various models including the artificial neural networks (ANNs), adaptive neuro-fuzzy inference system (ANFIS), and multiple linear regression (MLR) models. Soil temperatures along with other climate data are recorded by the Turkish State Meteorological Service (MGM) at specific locations all over Turkey. Soil temperatures are commonly measured at 5-, 10-, 20-, 50-, and 100-cm depths below the soil surface. In this study, the soil temperature data in monthly units measured at 261 stations in Turkey having records of at least 20 years were used to develop relevant models. Different input combinations were tested in the ANN and ANFIS models to estimate soil temperatures, and the best combination of significant explanatory variables turns out to be monthly minimum and maximum air temperatures, calendar month number, depth of soil, and monthly precipitation. Next, three standard error terms (mean absolute error (MAE, °C), root mean squared error (RMSE, °C), and determination coefficient (R ²)) were employed to check the reliability of the test data results obtained through the ANN, ANFIS, and MLR models. ANFIS (RMSE 1.99; MAE 1.09; R ² 0.98) is found to outperform both ANN and MLR (RMSE 5.80, 8.89; MAE 1.89, 2.36; R ² 0.93, 0.91) in estimating soil temperature in Turkey.

     

Comparison of artificial neural network and multivariate linear regression methods for estimation of daily soil temperature in an arid region

Soil temperature (T _S) strongly influences a wide range of biotic and abiotic processes. As an alternative to direct measurement, indirect determination of T _S from meteorological parameters has been the focus of attention of environmental researchers. The main purpose of this study was to estimate daily T _S at six depths (5, 10, 20, 30, 50 and 100?cm) by using a multilayer perceptron (MLP) artificial neural network (ANN) model and a multivariate linear regression (MLR) method in an arid region of Iran. Mean daily meteorological parameters including air temperature (T _a), solar radiation (R _S), relative humidity (RH) and precipitation (P) were used as input data to the ANN and MLR models. The model results of the MLR model were compared to those of ANN. The accuracy of the predictions was evaluated by the correlation coefficient (r), the root mean-square error (RMSE) and the mean absolute error (MAE) between the measured and predicted T _S values. The results showed that the ANN method forecasts were superior to the corresponding values obtained by the MLR model. The regression analysis indicated that T _a, RH, R _S and P were reasonably correlated with T _S at various depths, but the most effective parameters influencing T _S at different depths were T _a and RH.     

MLP-based drought forecasting in different climatic regions

Water resources management is a complex task and is further compounded by droughts. This study applies a multilayer perceptron network optimized using Levenberg–Marquardt (MLP) training algorithm with a tangent sigmoid activation function to forecast quantitative values of standardized precipitation index (SPI) of drought at five synoptic stations in Iran. The study stations are located in different climatic regions based on De Martonne aridity index. In this study, running series of total precipitation corresponding to 3, 6, 9, 12, and 24?months were used and the corresponding SPIs were calculated: SPI3, SPI6, SPI9, SPI12, and SPI24. The multilayer perceptrons (MLPs) for SPIs with the 1-month lead time forecasting, were tested and validated. Four different input vectors were considered during network development. In the first model, MLP constructed by importing antecedent SPI with 1-, 2-, 3-, and 4-month time lags and antecedent precipitation with 1- and 2-month time lags (MLP1). Addition of antecedent North Atlantic Oscillation or antecedent Southern Oscillation Index with 1-month time lag or both of them to MLP1 led to MLP2, MLP3, and MLP4, respectively. The MLP models were evaluated using the root mean square error (RMSE) and the coefficient of determination (R ²). The results showed that MLP4 had a higher prediction efficiency than the other MLPs. The more satisfactory results of RMSE and R ² values of MLP4 for 1-month lead time for validation phase were equal to 0.35 and 0.92, respectively. Also, results indicated that MLPs can forecast SPI24 and SPI12 more accurately than the other SPIs.     

夏季长白山天池站降水量年际变化特征及其对应的环流异常

长白山地处吉林省东南部,作为国家级重点生态功能区,其降水变化特征对该地森林生态系统和水资源结构变化有重要影响。本文基于1979～2016年吉林省47个台站逐月降水资料,探究了长白山天池站夏季降水的气候特征及其相关的环流异常,并与吉林省降水进行对比。分析结果表明天池夏季降水量以及年际变率高出吉林省其它站点一倍左右。此外,天池降水年际变化对应的环流异常与吉林省降水一致,即6月东北亚气旋式异常和东亚高空急流的增强,以及7、8月西太平洋副热带高压增强和东亚高空西风急流偏北,均可引起吉林省和天池降水偏多。另一方面,天池降水变异还表现出其独特的环流异常,与吉林省降水对应的环流异常显著不同或者甚至近乎相反,究其原因为长白山地形所致。本文的结果说明山地的气候和大气环流的关系复杂、多变。     

Forecasting monthly precipitation in Central Chile: a self-organizing map approach using filtered sea surface temperature

Western South America is subject to considerable inter-annual variability due to El Ni?o–Southern Oscillation (ENSO) so forecasting inter-annual variations associated with ENSO would provide an opportunity to tailor management decisions more appropriately to the season. On one hand, the self-organizing maps (SOM) method is a suitable technique to explore the association between sea surface temperature and precipitation fields. On the other hand, Wavelet transform is a filtering technique, which allows the identification of relevant frequencies in signals, and also allows localization on time. Taking advantage of both methods, we present a method to forecast monthly precipitation using the SOM trained with filtered SST anomalies. The use of the SOM to forecast precipitation for Chillan showed good agreement between forecasted and measured values, with correlation coefficients (r ²) ranging from 0.72 to 0.91, making the combined use filtered SST fields and SOM a suitable tool to assist water management, for example in agricultural water management. The method can be easily tailored to be applied in other stations or to other variables.     

大气中的降水与反降水──''98梅雨锋暴雨的天气学成因分析与预报探讨

对我国Ｔ１０６Ｌ１９（客观分析）模式大气,（１９９８年６～７月）做了大气中所有天气学降水（垂直运动形式）的计算。研究表明,’９８长江流域上空暴雨存在着明显的梅雨锋天气（尺度）系统降水,同时,长时间湿空气团的维持及其输运,和在梅雨锋上的非等熵湿绝热运动并不断形成对流不稳定降水,是强降水发生的天气学成因。因此,用模式大气中各种天气学形式的降水去（概率）统计预报实际大气降水,实现了预报因子和预报对象之间