【作者】 马金玉；

【导师】 罗勇； 李世奎； 申彦波；

【作者基本信息】 南京信息工程大学 ， 气候系统与全球变化， 2011， 博士

【摘要】 太阳辐射是自然环境中各种物理过程的主要能量来源,是驱动天气、气候形成和演变的基本动力,是地面生态系统的能量主要来源。太阳辐射在经过大气层时,由于受到云、水汽、大气中的CO2和臭氧等气体以及气溶胶粒子等的吸收、反射和散射作用,使到达地面的太阳辐射减弱。我国的太阳能资源十分丰富,各地太阳总辐射年曝辐量为3350-8370MJ/m2,大部分地区的太阳能资源优于欧洲和日本等太阳能利用发达国家,太阳能资源具有很大的开发潜力。本论文主要根据利用具有长年代(1961-2009年)的地面太阳总辐射、直接辐射和散射辐射以及有关气象要素的观测资料,采用线性倾向估计、小波分析和Mann-Kendall统计检验等方法,分析了中国地面太阳辐射长期变化特征,以及基于WRF模式的短期预报方法试验研究。研究结果表明：(1)中国近50年地面太阳总辐射长期变化趋势：在58个总辐射实际观测站中,通过α=0.05显著性检验的有34个站,其年太阳总辐射总体呈下降趋势,但随区域有所差异。其年代际距平变化趋势为：上世纪60年代和70年代以上升为主,但在70年代逐渐转为下降,80年代以后以下降明显为主,90年代以后个别站点略有上升。其累积距平的变化趋势有四类：一是先上升后下降型；二是先上升后下降,然后又略上升型；是先上升后下降,然后又上升型；四是变化不明显型。其年内变化,以冬季下降最为明显,春、夏、秋季为下降略明显。年太阳总辐射的年际周期为6～9年,年代际周期为10～13年、29～33年,突变时段大多发生在70年代。导致以上变化的原因很复杂,从有关要素与太阳总辐射的统计关系看,以年日照百分率、年平均风速、扬沙日数和年平均低云量的关系较大,还有人类活动引起的气候变化导致的影响,这些要素显著性的站点分布,具有一定的区域性和局地性。其中,年日照百分率和年平均风速与年太阳总辐射呈正相关,具有普遍意义。(2)中国近50年地面太阳直接辐射和散射辐射变化趋势特征：①采用归一化气候倾向系数方法,有更好的比较性,能较好地揭示辐射气候趋势的时空分布特征；②全国地面年太阳直接辐射和散射辐射的观测站点中,大部分站点呈下降趋势,但地区、年代际和季节有升降的差异。各站点的地面太阳直接辐射变化趋势只有变化明显和下降明显两类,地面年太阳散射辐射变化趋势为下降明显和变化不明显；③基于观测站点地理分布的特点,按气候带划分和站点占有的百分比进行分类统计分析,更能阐明辐射长期变化具有一定的地带性和局地差异性;④近50年全国14个站地面年太阳直接辐射的年代际周期为15～18年和33～34年。地面年太阳散射辐射的年际周期为8～10年,年代际周期为30～36年。站点的突变时间不完全一致,直接辐射的突变主要发生在60年代中期和70年代中期,散射辐射主要发生在80年代末到90年代初；⑤1961～2009年全国12个省级大城市地面年太阳直接辐射在80年代以前,大部分站点为明显上升,从80年代起至今以明显下降为主,但有波动。而地面年太阳散射辐射在2000年以前以下降明显为主,2000年以后以上升明显为主。⑥大部分省级大城市站点的直接辐射和散射辐射长期变化的下降趋势较其他城市更明显,这可能与城市化发展快慢引起的环境要素变化有关。⑦采用年太阳总辐射和年平均总云量、低云量实际观测资料,建立估算地面年太阳直接辐射和散射辐射的多元回归重建方程,具有较好的效果,可为增补该要素资料序列提供一种有效的方法(3)中国城市化对太阳辐射的影响：采用我国13个主要城市站点多年太阳辐射观测资料以及相关气象要素、城市人口等作为基本资料,采用气候倾向估计及有关要素的相关统计分析表明：①城市人口与总辐射的倾向系数呈显著性相关,随着城市由大到小,总辐射下降的倾向系数由大到小；②长年代的人口数据与地面年太阳辐射之间有一定的关系,两者之间的变化波动关系可用多项式模拟曲线表达,且绝大部分城市通过α=0.05显著性检验；③随着城市人口的增加与太阳辐射变化具有一定的相关关系,通过α=0.05显著性检验站点的地面太阳辐射增(减)趋势因城市环境条件不同而异。呈增加趋势的有拉萨、广州等,呈减少趋势的有格尔木、北京等。地面年太阳直接辐射呈减少趋势的有郑州、乌鲁木齐等,呈增加趋势的有拉萨、兰州等。地面年太阳散射辐射呈减少趋势的有格尔木、兰州等,呈增加趋势的有郑州、上海等。散直比呈减少趋势的有格尔木和兰州,呈增加趋势的有郑州、成都等。直射比呈减少趋势的有郑州、沈阳等,呈增加趋势的有格尔木、拉萨和兰州。散射比呈减少趋势的有格尔木和兰州,呈增加趋势的有郑州、沈阳等。(4)太阳能预报方法及其应用和问题：在参考了国内外大量有关太阳能预报文献的基础上,对太阳能预报方法的原理、机理以及具体的预报方法和应用进行了归纳评述。太阳能预报包括预测太阳辐射量和光伏发电功率,这对光伏发电系统并网运行有重要意义,是当前太阳能开发利用的一个关键问题。太阳辐射的预报方法主要有传统统计、神经网络、卫星遥感和数值模拟等方法。文中基于光伏发电应用的需求,分析了不同预报方法的优点和不足,并探讨了若干有待进一步改善的问题,展望了国内太阳能预报技术方法的发展和应用前景。根据目前实际应用于太阳能预报方法的个例看,主要是卫星资料、模式预报结果结合气象观测统计和外推方法,以及神经网络预测,而数值天气模式仍是当前预报的热点和难点。因此,今后太阳能预报技术的研究重点和方向主要是综合利用天气预报数据、卫星遥感数据以及地面云量观测信息,形成多层次、多信息融合的综合预报系统,可取得更好的太阳能预报效果。(5)基于WRF模式的短期逐时地面太阳总辐射预报试验：以新疆吐鲁番气象站为试点,预报试验期为2010年4月、7月、10月和2011年1月,每天对未来60小时进行逐时预报,并将预报结果分为24小时和48小时两段进行统计分析。试验结果表明：除1月份预报值系统偏大外,其他三个月预报结果与实际观测值趋势基本一致,并能刻画出天气转型时总辐射的波动变化。以4月份为例,两个时段的白天逐时总辐射预报值与观测值相关关系较好,均通过a=0.01的显著性检验。白天逐时总辐射预报值的相对误差与总辐射预报值的相关关系也较好,通过α=0.01的显著性检验。从预报值与实测值的平均相对误差看,有三分之二天数的日平均相对误差低于25%。从白天预报相对误差的逐时平均来看,中午时段相对误差最小,小于5%,9:00～14:00之间的相对误差低于20%,但早晚相对误差较大可达50%以上。更多还原

【Abstract】 Solar radiation is the primary energy source of various physical processes in the natural environment, and is basic force of driving weather and climate format and evolute. Solar radiation is main source of energy of the terrestrial ecosystems. When solar radiation goes through the atmosphere, due to absorption, reflection and scattering of clouds, water vapor and atmospheric gases such as CO2 ozone and aerosol particles, it reduces. The solar energy resources of China are abundant. The total solar radiation exposure yearly around China is 3350-8370MJ/m2. Solar energy resources in most areas of China are better than ones of other countries such as Europe and Japan where solar energy is highly developed. Solar energy has great potential for development. This thesis is based on long-term (1961～2009) total solar radiation, direct radiation and diffuse radiation, and observations of meteorological elements. It uses linear trend, wavelet Analysis and Mann-Kendall(M-K) statistical test methods to analyze China’s long-term variation of surface solar radiation. It is also based on short-term forecasting model based on WRF Experimental Studies. The results show that:(1)Regional long-term trend of global solar radiation in China:The global radiation and another meteorological elements data (1961～2009) from 58 meteorological stations is applied to analyze the global radiation characteristics and its relationship with another elements by the liner-trend estimation, wavelet analysis and M-K test. The results show that: Between of these 58 weather stations, the passages of significance test of 34 stations are at 95% confidence level. Meanwhile, among these 34 stations, the global radiation trend is ascent, but that is different from different area. The trend of the decadal deviation percentage:the main trend in 1960s is ascent, but gradually it dropped in the 70s, and it decreased significantly after 80s.After 90s, a little site are increase slightly. The trends of the cumulative variance have four types, such as up-down, up-down-a little elevation, up-down-no significant increase. For inner year change, there is the most obvious decline in winter, while a little decreases in spring, summer and autumn. The major cycles of global radiation are 6～9a、10～13a,29～33a. The sudden change periods are almost in the 70s. The reasons which lead to more changes are complex. According to the solar radiation with the relevant elements by statistical methods, the respectively relationship between the annual sunshine percentages, annual average wind speed, blowing sand and the average number of days low cloud cover with the global radiation are closer. In addition, the effect of climate change by human-induced is also important for such region distribution. Especially, the percentage of sunshine and mean annual wind speed was positively correlated with the annual total solar radiation. But the annual average temperature was negatively correlated with solar radiation, which may cause the temperature increase in the greenhouse. Some elements locally, such as blowing dust days and number of days should be negatively correlated in practical, but it is positively correlated by statistic. The impact mechanism is not clear, and needs further study.(2)The trend of direct and diffuse radiation in China over recent 50 years:In this paper, we used the direct and diffuse solar radiation data during 1961-2009 from 14 stations and during 1961-1991 from 49 stations in China. The present study based on the method of linear regression (taken time as an independent variable), wavelet analysis and Mann-Kendall statistical test. The characteristics of the direct and diffusion radiation are analyzed from different angles including that of decadal variation, long-term seasonal change, inter-decadal anomalies, cycle and abrupt change. The results showed:①The spatial and temporal distribution of the climatic trend for the direct and diffuse radiation was revealed by using the normalized climatic tendency coefficient. At the same time, these results are comparable.②The direct solar and diffuse radiation in China showed decrease trend, but there are many difference in the regions, inter-annual and seasonal trend of them.③Based on the characteristics of the geographical distribution on observation stations, according to the climate zone, it is classified with the percentage of sites by statistical method. A certain area and Local differences were exhibited clearly in the long-term changes of the radiation.④Using wavelet and MK test method, the statistical results show that the radiation in different regions with different period and abrupt change.⑤The long-term trends of the direct and diffuse radiation in most part of the provincial cities were decrease which was more obvious than in other cities. According to some study, such climatic phenomenon may be caused by environmental factor changes of urbanization relating to the development of different cities.(3)The influence of urbanization on solar radiation in China:The urban population, solar radiation and related meteorological observation data in 13 major cities was analyzed by the liner-trend estimation method. The results show that:In recent decades, due to urbanization and increasing population, the urban environment change caused by human activities had affected the long-term variations of surface solar radiation.①the urban population and the tendency coefficient of the global radiation was significant. The global radiation coefficient was decrease with the city size;②Ther are some relationship between the population data and the ground global radiation among 13 cities. Such relationship could be simulation by polynomial curve. Most of them the city could through the test of significance;③among the 13 cities, the changes of solar radiation and has a certain correlation with the urban population increasing, and the related coefficient of some city could be through the significance test (α= 0.05).The trend of the ground solar radiation due to the urban environmental conditions vary. Among them, some city show increasing trend, such as Lhasa, Guangzhou, and so on, showed an increasing trend in the city of the Golmud and Beijing, and so on. Direct solar radiation on the ground tended to decrease in the city of Zhengzhou, Urumqi, and so on, showed an increasing trend in the city of Lhasa, Lanzhou, and so on. The Ground diffuse solar radiation on a decreasing trend in the city of Golmud, and Lanzhou, and so on, showed an increasing trend in the city of Zhengzhou, Shanghai, and so on. The ratio of diffuse and direct radiation tended to decrease in the city of Golmud and Lanzhou, showed an increasing trend in the city of Zhengzhou, Chengdu, and so on. The ratio of direct and global radiation tended to decrease in the city of Zhengzhou, Shenyang, and so on, showed an increasing trend in the city of Golmud, Lhasa and Lanzhou, and so on. The ratio of diffuse and global radiation tended to decrease in the city of Golmud and Lanzhou,, showed an increasing trend in the city of Zhengzhou, Shenyang, and so on.(4)Review on methods of solar forecasting:reference to a large number of domestic and foreign literature about solar energy prediction, it was summarized and reviewed on the principle and mechanism of the solar energy prediction, the method for solar forecasting and its application. Solar forecast, including forecasts of solar radiation and photovoltaic solar power, is important for photovoltaic power generation system in network operation. Among them, the solar radiation prediction methods are involve the traditional statistics, neural networks, satellite remote sensing and numerical simulation methods; photovoltaic power generation forecast method is mainly statistical. Based on the application of photovoltaic power generation needs, the different advantages and disadvantages of some forecasting methods will be analyzed. At the same time, a lot of problems for further solar forecasting techniques research with the development of domestic will be proposed in this paper. In term of some case about practical solar energy, the main method is that satellite data, model predictions, meteorological observations data combine with statistics extrapolation methods and neural network, but numerical weather model is still the forecast difficult currently. Therefore, in the future, as for better effect of solar energy forecast, the solar forecasting technology includes model results, satellite remote sensing data, ground observations, and so on. (5)The global solar radiation short-term forecasting based on WRF model:The prediction test is on Turpan weather station for a trial of the next 60 hours in April 2010.The forecast results could be divided into two sections of 24 and 48 hours to statistical analysis.The results show that:The forecasting results are basically consistent with the actual observations data. The global radiation in cloudy could be forecasted. In addition, it could describe the global radiation during weather condition transformation. The hourly global radiation forecasts in two periods during the daytime have good correlation with the observed values, and it was adopted by the significant testα= 0.01. The correlation between hourly relative error during the daytime and prediction values are also good, by the significance testα= 0.01. For the average relative error of the prediction results and measured values, two-thirds days of were less than 25%. For the mean hourly relative error in daytime, the relative error at noon was least(less than 5%); the relative error during 9:00-14:00 was less than 20%, but the relative error sooner or later was up to 50% and above.更多还原