【作者】 孙红霞；

【导师】 付强；

【作者基本信息】 东北农业大学 ， 农业水土工程， 2010， 硕士

【摘要】 黑龙江省是我国农业大省,同时也是我国重要的商品粮基地。农业的发展受到天气状况的制约,在全球气候变化背景下,干旱出现的频率更加频繁,旱灾对农业的影响是气象灾害中影响最深远、最严重、最具有破坏力的,因此,研究干旱气候的基本规律以及全球气候变暖背景下干旱的趋势是有必要的。通过研究气象干旱的变化特点,并根据地区特点合理规划与高效利用水资源,发展现代节水农业,对我国粮食安全、生态安全和资源安全都具有重要的战略意义。本论文以齐齐哈尔市为研究对象,数据采用了齐齐哈尔市气象局提供的数据资料,论文的具体研究内容如下：(1)采用M-K秩次相关法、累积滤波器法、距平分析方法从年和季节两个时间尺度上分析了气温和降水量的变化趋势以及年代际变化特征。结果表明,近50年来齐齐哈尔市年气温变化大致呈上升趋势,变化倾向率为0.486℃/10a。各个季节从70年代开始一直到现在,都有不同程度的上升趋势,而且冬季升温最明显,其次是春季；年降水量在呈下降趋势,变化趋势不明显,春、冬季的降水量总体呈增多趋势,且冬季降水量增多幅度高于春季,秋季降水量则呈减少趋势。从各趋势变化来看,在未来几年,夏、秋两季的降水量将会继续减少,而春、冬两季的降水量将会继续增多。(2)用标准差法把降水量分成了五个等级：涝、偏涝、正常、偏旱和旱,列出了该市的旱涝等级表和频率分析表,分别得出了旱涝在时间和空间上变化特点。在时间上,旱涝具有频繁性、交替性与连续性、普遍性特点。在所有的年份当中,发生旱涝的年份远远多于正常年份,这说明齐齐哈尔市是一个灾害多发的地区,且干旱发生的频率高于洪涝发生的频率。在空间上,具有区域性的特点。从旱涝频率统计表中分析可知,富裕县在所有的县市中无论是干旱还是洪涝,发生的概率都是最大的,其次是依安县和龙江县,拜泉县、泰来县、克东县、克山县、讷河市和甘南县相对较小,但是发生干旱概率最小值为31.4%,发生洪涝概率最小值也在百分之五十以上,针对灾害的严重性,进而提出了减灾对策。(3)马尔可夫链是状态和时间均离散的马尔可夫过程,简称“马氏链”。它的最基本特征是“马氏性”,也称“无后效性”,即在系统“现在”的状态已知的条件下,其“将来”的状态与“过去”的状态无关。降水量是一随机序列,符合马氏性,可以应用加权马尔可夫模型对齐齐哈尔市2009年降水量状态进行预测,预测结果为偏涝年。加权马尔可夫模型克服了传统马尔可夫的缺点,达到了充分合理利用信息的目的,而且预测出未来某时段降水量是一个变化区间而不是一个具体值,这样一来,预测的范围扩大了,其预测的可靠性也可以相应地提高。(4)应用了基于遗传算法的投影寻踪自回归模型对降水量进行预测,结果表明,用PPAR模型预测降水量是可行而有效的。PPAR模型计算简便、适用性强,克服了传统投影寻踪方法计算复杂、编程实现困难的缺点,有利于投影寻踪方法的推广应用,为处理线性时序预测提供了一条值得探索的新途径。更多还原

【Abstract】 Heilongjiang Province is a major agricultural province in China, it is also an important commodity grain base.The development of agriculture is constrainted by the weather conditions.on the background of global climate change,the frequency of droughts become more frequent,,drought is the most proround impact,serious and destructive in meteorlogical disasters.thus,study of the basic law of arid climate and blobal warming trends in the context of drought is necessary. By the studying the changes of meteorological drough in accordance with regional characteristics of rational planning and efficient use of water resources,the development of modern water-saving agriculture have important strategic significance on China’s food security ecological security and resource security.This paper has studied the city of Qiqihar. The data of Qiqihar City using data provided by the bureau of meteorology,specific research papers reads as follows:(1) The changes in temperature and precipitation trends and interdecadal variations were anylysised by using M-K rank correlation,cumulative filter method,anomaly analysis from the two-year and seasonal time scales. The results show that:Qiqihar City was upward trend in temperature changes in the past 50 years, the rate of changes tend was 0.486℃/10a,each season starting from 70 years to present.There was various degrees of upward trend,,but the most signeficant warming-up was in winter,followed by spring;Annual precipitation in the downward trend,trends was not obvious,spring and winter precipitation was an increase in the overall trend,and the winter increase in precipitation rate is higher than that in spring, autumn rainfall is decreasing trend.Terms of the changes from the trend to the next years,summer and autumn rainfall will continue to decrease,while the spring and winter precipitation will continue to increase.(2)With the standard deviation method,precipitation was diveded into five grades:logging, partial waterlogging,normal,partial drought and drought.The city’s scale and frequency of drought and flood analysis table were listed, droughts and changes in time and space characteristics were obtained.In time, drought and flood has a freqrency alternating with the continuity and universal character.In all these years,the year of drought and flood occurred far more than anormal year,indicating Qiqihar City is a disaster-prone area,and the frequency of drought occurrence is higher than that floods.In space,.statistical tables from the drought and flood frequency analysis with regional characteristics shows that:Fuyu county in all counties in either drought or flood,the probability of occurrence were the largest one, followed by Yian and Longjiang County,Baiquan,Tailai County,Kedong,Keshan,Nehe City,and Gannan County is relatively small,but the probability of drought minimum 31.4%,the probability of flooding is also 50 percent more than the minimum,Against the severity of desasters,then presents a desaster countermeasures.(3)Markov chain is both the state and time-descrete Markov process,referred to as "Markov chain".Precipitation.It is a random sequence,in line with Markov property,can be applied to the weighted Markov model.Precipitation in the state of Qiqihar City in 2009 to predict the results of partial flood year.Weighted Markov model overcomes the shortcoming of the traditional Markov,reaching the purpose of fully and rationally use of information,but also predicts that a future time interval of precipitation is a change,rather than a specific value, this way,the projected scope of the its predicted reliability can be increased accordingly.(4)Application of genetic algorithm based on projection pursuit autoregressive model predicted precipitation,results showed that PPAR models to predict precipitation is feasible and effective. PPAR model calculation is simple,applicable to overcome the shortcomings of the computational complexity and programming difficulties of traditional projection pursuit method. Projection pursuit method is conducive to the promotion and application providing a new approach in order to deal with linear time series prediction worth exploring.更多还原