【作者】 李明；

【导师】 吴正方；

【作者基本信息】 东北师范大学 ， 自然地理学， 2011， 博士

【摘要】 目前,越来越多的证据表明全球气候正逐步变暖,而植物物候现象是全球变化的积分仪和自然环境变化的综合指示器。有关物候对气候变化的响应研究正成为全球变化的热点问题。森林是全球生态系统的重要组成部分,森林物候是反映短期或长期气候变化对森林生长阶段影响的综合性生物指标。遥感技术的发展,为在景观层面进行物候观测提供了技术手段,有效地实现了物候监测从点向面的空间尺度转换。本文主要运用GIS和遥感技术,使用10天合成的SPOT/NDVI多时相遥感数据、气象数据以及各种相关的图表和统计资料等,构建了长白山地森林生长季遥感监测模型(Double Logistic模型),并利用该模型计算了1999~2008年长白山地森林植被的各物候参数;分析了1999~2008年长白山森林生长季始期、生长季末期、生长季长度、DN年最大值、DN年振幅和DN生长季积分值的时空分布格局及年际变化趋势;探讨了长白山地的气候变化特征;并在此基础上进一步深入分析了森林生长季变化与气候因素的关系,揭示了长白山森林生长季对区域降水和气温变化的响应方式和反馈机制。论文的主要结论如下:1 Double logistic模型的构建基于遥感和地理信息系统技术,本文结合logistic模型和一个全局函数的优点,构建了double logistic模型。该模型不仅可以同时提取多年的森林物候期,而且可以消除时间序列的边界效应,它能更灵活地拟合复杂的曲线。同时,由于森林开始生长和结束生长是两个不同的生物物理过程,对于拟合曲线的两侧采用动态的比例更符合森林的生长过程;再者,该模型是在没有设置预定的阈值或者经验参数前提下而逐个像元运算的,所以它能够更加稳定地表征不同像元之间物理意义,具有更好的通用性。野外观测数据和先前学者的研究结果都证明了利用double logistic模型动态监测长白山地植被物候期是一种可行的方法。2长白山地森林植被物候参数时空格局(1)在1999~2008年间,长白山地大部分地区森林生长季始期发生在第100~140天,其中在第100~110天和在第110~120天开始生长的地区所占面积比较大。提前的区域仅占研究区面积的32.46%,平均提前速率约0.71d/a;延迟的区域占总面积的67.54%,平均延迟速率约为0.43d/a。(2)大部分地区森林生长季末期发生在第270~290天,其中在第280~290天开始生长的地区所占面积比较大。长白山地森林的生长季末期中南部表现为延迟趋势,南端和北部表现为一定的提前趋势。延迟区域占有较大比例,约为65.3%,平均延迟速率为0.53d/a;提前的区域占总面积的34.7%,平均提前速率为0.57d/a。(3)大部分地区森林生长季长度140~180天,其中长度在160~180天的地区所占面积比较大。长白山地森林的生长季长度中东部表现为延长趋势,南端和西北部表现为一定的缩短趋势,整体上呈现东南—西北向的空间分异格局。生长季延长的区域占研究区面积的59.69%,平均延长率为0.67d/a;生长季缩短的区域占总面积的40.31%,平均缩短率约为1d/a。(4)生长季DN积分值、DN年最大值和DN值年振幅整体上都有增加的趋势,且空间格局具有相似性,是因为三者之间是相互关联的。如果生长季长度不变、DN年最大值或者振幅的增加必然会导致生长季DN积分值的增加;如果生长季长度和生长季的DN基准值不变,那么DN年最大值和振幅的含义就是一样的了。3长白山地的气候变化特征综合年平均气温和年降水量的时间变化来看,可以认为长白山地近50年来来气候的年代际变化经历了一个“冷湿—冷干—暖湿—暖干”的过程,并且暖干的过程仍将持续。从地域分布上看,年均增温趋势分布与春季的增温空间格局较相似,增温幅度与春季也相差无几,与其它几个季节的空间分布存在明显的差异。秋季降水量变化趋势空间分布特征与夏季相似,但是降水量增加的区域略有增加。年降水量的空间分布主要受夏季和秋季降水量的影响,故它与夏、秋季二者叠加后的分布趋势一致。4森林植被生长季变化与气候的关系长白山地大部分地区的NDVI与旬降水量、旬均气温的年内相关系数和偏相关系数都较高,呈强显著相关,但温度对森林植被的生长过程的影响大于降水。各种森林植被类型,无论是针叶林还是阔叶林,它们的生长季始期大都与其之前各月月均温呈显著负相关,而生长季末期则与其之前各月月均温呈显著正相关;而它们的生长季始末期与降水的关系就较复杂一些,针叶林的始期与其之前的各月降水量呈正相关,阔叶林则呈大都呈负相关。除蒙古栎外,其它森林类型的生长季末期与其前各月的降水量呈负相关。更多还原

【Abstract】 Recently, more and more evidences have shown that global climate is becoming warmer and warmer, and vegetation phenology is an integraph of global changes and a comprehensive indicator of landscape and environment changes. The responses of vegetation phenology to global environment changes have become a focus field of global changes. Forest is an important part in global ecological system, and the forest phenology is the comprehensive biological index which reflect the influences of the short- and long-term climate changes on the forest growth stage. Remote sensing technology offer the technical means for monitoring the vegetation phenology on landscape scale, and realize the spatial transition of phenological data from points to coverage.Combing with ten-day SPOT/NDVI data, meteorological data and varities of statistic data using GIS and remote sensing technology, the paper constructs a double logistic model, and calculate forest phenology metrics based on it, then analysize the spatio-temporal patterns and change trends of the start of growing season (SOS), the end of growing season (EOS), the length of growing season (LOS), annual DN (digital number) max, annual DN amplitude and DN integral of growing period in Changbai Mountains during 1999~2008. Subsequently, the paper evaluates the characteristics of climate change in Changbai Mountains, and analysize the relationship between growing season variation and climate factors, then discuss the response and feedback mechanism of vegetation phenology metrics to regional air temperature and precipitation in different vegetation types. From these researches, some basic conclusions are drawn as follows:1 Double logistic model constructionBased on remote sensing and GIS, the paper develop a double logistic model that combines the merits of logistic model and global function. The model can not only extract vegetation phenophases for several years simultaneously, but also avoid the boundary effects of time-series curve. With this feature, the data obtained are more flexible and complicated for fitting curves. It is more proper to fit the both sides of curve using dynamic ratios for the SOS and EOS being different biophysical processes. Moreover, the model treats each pixel individually without setting absolute thresholds or empirical constants, in which way, the model can characterize the physical meaning of different pixels more accurately. Validation through comparing with the field data and previous research results demonstrate that double logistic model used in this paper is appropriate for the study of vegetation phenology in Changbai Mountains.2 Temporal-spatial patterns of forest phenology metrics in Changbai Mountains (1) During the year 1999~2008, the SOS of forest in Changbai Mountains commences on the 100th to 140th day of year, which is consistent with the period of leaf unfolding for forest in late April and early May. Among of the period the day 100~110 plus the day 110~120 take a greater proportion. The advance areas of SOS are small from the whole aspects, only accounting for 32.46% over the study area, and the average advance ratio is about 0.71 d/yr; The delay areas of SOS take up to 67.54%, and the average delay ratio is about 0.43 d/yr.(2) The EOS concentrate on the day 270~290, of which the day 280~290 takes a greater proportion, it is corresponding to the period of forest defoliation in the fall. The delay areas of EOS take a greater proportion, about 65.3%, which chiefly centralize at the middle-southern area, and the average delay ratio is about 0.53 d/yr. Meanwhile, the advance areas of EOS distribute in the southern extrem and northern part, which occupy 34.7% of whole study area, and the average advance ratio is about 0.57 d/yr.(3) The LOS of forest mainly range from 140 to 180 days, of which the period 160~180 takes a greater proportion. The LOS prolong in middle-eastern areas and shorten in northwestern areas, and the spatial pattern present as Southeast-Northwest discrepancy. The prolonged areas and shortened areas occupy 59.69% and 40.31%, and the prolonged ratio and the shortened ratio are 0.67 d/yr and 1 d/yr, respectively.(4) DN integral of growing period, annual DN max and annual DN amplitude all express increasing trend, and their spatial patterns are similar, because the three metrics are correlated. If LOS keeps invariant, the increase of annual DN max or annual amplitude will result in the increase of DN integral of growing period. If both LOS and base DN keep invariant, annual DN max and annual DN amplitude will have the same meaning.3 Characteristic of climate change in Changbai MountainsConsidering the annual mean temperature and annual precipitation variation, the decadal climate variation have gone through one process:“cold and wet– cold and dry– warm and wet– warm and dry”, and the last period will persist in the future. From the perspective of spatial pattern, the distribution and change ampltude of annual mean temperature and mean temperature in spring are similar, but obvious differences exist between other seasons. The precipitation distribution in spring and in autumn is similar, nothing but the rainfall increasing areas in autumn are little larger than that in sping. The spatial pattern of annual precipitation is consistent with the overlay of precipitation distribution in spring and in autumn.4 Relationship beween climate and the variations of forest growing seasonThe correlation coefficient and partial correlation coefficient are high between decad NDVI and decad precipitation/temperature in most regions of Changbai Mountains, and the influnence of temperature on forest growing process are more significant than that of precipitation. There are a significant negative correlation between the SOS and the average monthly temperature before SOS in majority of various vegetation types, and the EOS have a significant positive correlation with the average monthly temperature before EOS and after SOS. The relationship display relatively complex, the SOS of coniferous forest have a positive correlation with average monthly precipitation before SOS, and it is opposite to broadleave forest. The EOS of each forest types express negative correlation with annual precipitation before EOS and after SOS, except Quercus mongolica.更多还原