【作者】 徐慧；

【导师】 周志翔；

【作者基本信息】 华中农业大学 ， 园林植物与观赏园艺， 2011， 硕士

【摘要】 植被覆盖动态变化是土地资源和环境变化研究的重要内容。本文以遥感技术为基础,基于1999-2009年SPOT VEGETATION的时间序列影像数据,综合应用多种数学方法,分析了长江流域植被覆盖的空间分布和动态变化特征。用平均值法得到的多年平均NDVI影像图显示长江流域植被覆盖空间分布上的平均状况,用slope分析法、Mann-kendall趋势分析法、变化矢量分析法(CVA)和Hurst指数R/S分析法对不同尺度(年、季、月)的NDVI时间序列进行数学运算,分析长江流域植被覆盖的变化特征,以及不同流域分段、水系分区、不同海拔梯度以及不同植被类型NDVI的空间分布和时间变化特征。研究结果显示：(1)长江流域植被覆盖的整体状况较好,1999-2009年的年平均NDVI值<0.1无植被区仅占流域总面积的1.8%,NDVI在0.5-0.65之间的中高植被覆盖区和NDVI>0.65的高植被覆盖区分别占到长江流域总面积的47.4%和31.7%。季节水平上,春秋两季主要以中高植被覆盖为主,夏季以高植被覆盖为主,而冬季以中植被覆盖为主；月水平上7月中高、高植被覆盖区的面积比占到了流域总面积的88%,2月份的植被生长状况最差,年均NDVI>0.5以上的面积比重最小,不到9%。空间分布整体上符合南多别少,东多西少的趋势。年NDVI值较少的区域主要分布在青海长江源头和中下游大型湖泊及周边大中城市。(2)年内变化上,长江流域NDVI年内变化呈单峰型,2月值最低,7月份达到最大值后开始逐渐下降。4月出现最大增幅,10月有最大减幅。在不同流域段水系区内NDVI的年内变化曲线有所差异,下游流域段的各水系NDVI年内变化呈现双峰型,尤其是下游农田的变化曲线,峰值出现在4月和8月,6月有小低谷值；中游、下游和<200m的低平原在8月出现最大值。东部地区的生长季比西北部地区结束要晚。(3)在年际变化中,近11年来长江流域植被覆盖整体呈显著上升趋势,年均增加斜率为0.0059,11年增幅为0.0649,在1999年的基础上增长了14.25%。。其中,夏、秋两季对长江流域植被覆盖状况的改善贡献最大,季均NDVI增幅分别为0.070、0.067,上升了11.92%、13.75%。在不同植被类型中,森林、灌草地及农田NDVI均呈明显增加趋势,其中农田增幅最大为0.079,上升了17.17%。在不同海拔梯度上,500-1000米海拔带上的NDVI增幅最大为0.092,上升17.94%。从水系分区上看,仅太湖水系的植被覆盖呈退化和减少趋势。(4)变化趋势的空间分布上,近11a来,长江流域年均NDVI有92.3%像元呈上升趋势,而降低像元数仅占7.7%。表明长江流域近十来年的“长防林工程”、“天保工程”等植被恢复工程的生态效益已开始显现。表现减少趋势的区域主要分布在长江源头,金沙江流域中上游,岷沱江水系北部,局部江南丘陵地带,和部分大中城市。根据F检验,变化趋势不显著地区面积比例为39.4%,一般显著增加的区域在流域内呈零散分布,面积占18.5%,大部分地区变化呈显著增加,面积占到42.1%。春、夏、秋、冬四季呈显著增加面积分别占流域面积的54.1%、42.1%、35.5%、9.8%。12个月当中,4月份显著增加的面积最大,占到了43.5%的流域面积,12月显著增加的面积最小,只占4.5%。(5)采用变化矢量(CVA)分析法对1999-2009年、1999-2004年和2004-2009年3个时间段进行了变化矢量的模(‖△p‖)计算,发现半数以上的像元6年或11年NDVI‖△p‖的值位于0.2-0.4的低变化区。通过转移矩阵发现2004-2009年间的变化强度不如1999-2004年。1999-2009年的‖△p‖空间分异：下游>中游>上游；乌江水系>上游干流区>太湖水系(前3位),金沙江水系的‖△p‖值最小；500-1000m的低海拔山地‖△p‖较大,海拔>4000m的高海拔区‖△p‖较小；4种主要植被类型上,农田>灌丛>森林>草地。‖△P‖的大小与植被造林工程、农林业生产和城市化进程等人类活动息息相关。(6) Mann-kendall的分析结果与slope分析结果很接近,两者之间对年际NDVI的变化显著与不显著上有较一致的判断,但在区分增加趋势极显著和显著水平上有10%左右的差异。根据Z值的大小分布可以较灵活的选择显著性水平是此方法的一个优势。年均NDVI序列Mann-kendall分析结果的Z值空间分异：在0.1水平上(1.64)下游NDVI增加趋势不显著；水系区内金沙江、岷沱江增加趋势不显著、太湖水系减少趋势不显著；>4000m高海拔地区增加趋势不显著；植被类型中草地的增加趋势不显著,其余流域区段/海拔/植被类型的变化趋势显著。(7) Hurst指数分析结果显示,长江流域年均NDVI不是纯粹的随机游走而是有偏的随机游走序列,具有自相似性和时间依赖性。全流域上的Hurst(?)旨数平均值为0.8344,各流域区段/海拔梯度/植被类型像元上的平均Hurst指数都在0.75以上,表明变化趋势具有可持续性。其中下游Hurst(?)旨数比上游和中游要高；水系区内汉江水系、中游干流区和下游干流区的Hurst(?)旨数较高；海拔梯度上的Hurst指数值差别不大,以500-1000m的低海拔山地最高,>4000的高海拔地区最低；植被类型上农田>灌丛>森林>草地。说明这些地区/植被类型上的NDVI序列数据的自相似性更高,持续性相对较强,可能与人类规律性的生产活动有关。上述结论将为长江流域生态环境的保护和建设,土地资源管理和决策以及区域经济、环境、社会的可持续发展,提供一定的参考依据。更多还原

【Abstract】 Vegetation dynamic is the important research content of land resources and environmental changes. Based on the remote sensing technology and the image data of time series from 1999 to 2009 in SPOT VEGETATION, spatial distribution and the characteristics of dynamics of vegetation among Yangtze river Basin were analyzed through kinds of mathematical methods.The average NDVI image map obtained by using mean value method showed the average condition of the spatial distribution of Yangtze river vegetation by using slope analysis, Mann-kendall trend analysis, CVA, and Hurst index R/S analysis. Math operations on-the NDVI time sequencing of each year,season and month were done,the research on the characteristics of Yangtze river vegetation’s dynamic, and the NDVI spatial distribution characteristic of different vegetation, watershed, water system division was done as well. The results showed:(1) The whole Yangtze River basin vegetation was in good condition. And the average NDVI value per year from 1999 to 2009 was smaller than 0.1. Non-vegetatioon area was up to 1.8% in the whole basin. High vegetation whose NDVI value was between 0.5-0.65 accounts for 47.4% of the whole watershed area, and the middle-high vegetation whose NDVI value was higher than 0.65 accounted for 31.7% of the whole watershed area. On the seasonal level, high vegetation was the main vegetation in spring and autumn, high vegetation was the main vegetation in summer as well, and the middle vegetation was the main vegetation in winter. On the month level, the middle, middle-high vegetation accounted for 88% of the whole watershed area. In February, the condition of vegetation was the wrost, and the area whose NDBVI value each year was higher than 0.5 accounted the least proportion, which was less than 9%. The trend of whole spatial distribution was that vegetation in south was more than that in north, that in east was more than that in west. The area with little NDVI value per year mainl located in Qinghai to the Yangtze river headwaters, large lakes in the middle and lower reaches of the Yangtze Rive rand nearby large and medium-sized cities.(2) The NDVI change curve appeared unimodal within a year, February had the minimum value, the curve dropped after July got the maximum value. April gained the greatest growth based on last month March, while October dropped most sharply. The curve’s appearance diversed from different streams/ water systems/ elevation/ vegetation forms, it was bimodal in water systems of down streams, particularly for farmland in down reaches, the double-humped curve peaked in April and August, and had a lower value in between in June; while the whole Yangtze River Basin peaked in July, in middlestream and downstream area the curve peaked in August. It was implied that growing season was delayed in East part compared to West.(3) For interannual change, the vegetation cover enhanced obviously in Yangtze River Basin as a whole in the last 11 years, the slope for annual NDVI change trend line was 0.0059, increased 0.0649 for 11 years, amplified 14.25% from 1999. Summer and Autumn contributed much more for improved vegetation cover ratio, seasonal NDVI increased 0.070 and 0.067, account for 11.92% and 13.75% for 1999. There were general increasing trend for four main vegetation forms with farmland increased most remarkable, amount to 0.079, up to 17.17% gaining rate; at the elevation of 500-1000m, NDVI gaining rate was 17.94%, increased 0.092, topped in the 6 elevation classes; among 12 water systems, only Tai lake basin showed decline trend.(4) Space distribution of change trend (slope image) demonstrated that annual NDVI increased in most areas of Yangtze River Basin, around 92.3% pixels showed increasing trend, compared with 7.7% pixels that decreased which indicated ecological benefit from vegetation restoration such as " Shelterbelt Construction project of the Yangtze River "and "Natural Forest Protection Programme"。Where NDVI value declined scattered in the source region, up-middle streams of Jinsha river, north of Min-Tuo river basin, part of Chiang-nan hilly region and certain urban areas. Through F-test, classified change trend image was obtained, which showed percentage of nosignificant chang area was 39.4%, scattered significant increased area 18.5%, while 42.1% area in the whole basin where NDVI was increased extremely significant. For four seasons, NDVI extremely remarkable increased area accounted for 51.1%,42.1%,35.5% and 9.8% separately. Among 12 monthes, April NDVI enhanced most vigorously, about 43.5% area increased extremely significant in this month, while December had the least significant increased area which accounted only for 4.5%.(5) Change Vector Analysis method was used to analyse the change intensity during 1999-2009,1999-2004 and 2004-2009 those three time slot by calculate the values of‖Δp‖. The result showed that more than half of the pixels values belonged to the low change grade. According to the transition matrix, the change intensity during 2004-2009 was less than that of 1999-2004.The analysis of spacial variablility of‖Δp‖in 1999-2009 illustrated that:downstream >middlestream >upstream sort by the value of‖Δp‖; for 12 water systems, Wujiang river >upstream of the mainstream>Tai lake area (which were the top three areas), and Jinsha river had the least value of‖Δp‖; the value was relatively high in elevation of 500-1000m and rather low in elevation of more than 4000. For the four main vegetation form, farmland >bushwood >forest > grassland sort by the values.There were positive correlation between value of‖Δp‖and-human activities such as vegetation afforestation, rural activity and urbanization.(6) The result of Mann-kendall Analysis was much the same with slope analysis method, especially in dividing the change by significance or not, but there was about 10% deviation in percentage of extremely significance areas. It has the advantage of choosing significant levels adaptively according to the Z values of the whole data. Z value was the calculated result of the NDVI time series by Mann-Kendall Analysis method, its characteristics of spatial division illustrated that:increasing trend in downstream annual average NDVI time series was not significant, so was Jinsha river and Min-Tuo water systems, and the decreasing trend in Tai lake water system demonstrated nosignificance; the increasing of NDVI in elevation of more than 4000m was not remarkable; grassland either; the rest streams/ water systems/ elevations/ vegetation forms demonstrated significant increasing trend in 0.1 significant level with the threshold value 1.64.(7) Hurst Index indicated annual average NDVI time series was not mere random walk but biased, with self similarity and longtime dependency. The average Hurst value was 0.8344 in the whole basin, while the average value in different streams/ elevation/ vegetation forms were all above 0.75, which demonstrated that the change trend of NDVI were continuable. The hurst index was higher in downstream compared to up/middle streams; Hanjiang river> middlestream of the mainstram> downstream of the mainstream (top three) among 12 water systems; difference among elevation was less remarkable, elevation of 500-1000m had the maximum while elevation above 4000m had the least value; for 4 vegetation form, farmland >bushwood >forest >grassland sorted by Hurst index value. Hurst value may relate to regular agricultural activities.Conclusions of this research may provide as certain reference and evidence for Ecological Environment protection and construction, Land Resources management and policy, and may also help with the Sustainable Development of population, resource, environment and economy.更多还原