基于植被指数的新疆和静县草原地上生物量模拟
发布时间：2019-06-12 00:50
【摘要】：草原地上生物量的研究是草原生态学与生态经济学研究中的一个重要方面,实时准确的获取草地植被覆盖动态变化以及草原地上生物量的分布状况,是合理利用及保护草地资源的前提。以新疆和静县草原为研究区,草原地上生物量为研究对象,Landsat影像为主要遥感数据,采用野外实测采样点地上生物量数据为建模数据,利用像元二分模型,分级反演四个不同时期草地植被覆盖度,结合重心迁移模型,研究其变化特征;基于植被指数NDVI、RVI与草原地上生物量实测值间的相关分析,分别建立一元线性模型、指数模型、二次多项式模型,通过SPSS统计分析,进行模型精度检验,确定适用研究区草原地上生物量反演的最优模型,利用最优模型进行草原地上生物量的遥感反演。1.基于像元二分模型的草地植被覆盖度估测值与野外实测值之间进行线性相关分析,得到相关系数为0.7267,表明运用此方法对新疆和静县草地植被覆盖度研究是可行的。通过面积加权计算出研究区2000、2005、2010、2015年平均植被覆盖度分别为4.2851、4.3042、4.4252、3.7524,平均植被覆盖度呈现先增加再增加后减少的趋势。2000~2015年平均植被覆盖度整体减少0.5327,其中,2015年较2010年植被覆盖度减少最多,减少15.20%。2.2000~2015年,五个不同等级植被覆盖度由低到高转化率依次为54.82%,60.08%,5.52%,266.82%,772.37%,三级植被覆盖转化率最小,转化率为5.52%,五级植被覆盖转化率最大,转化率为772.37%。2000~2015年,一级植被覆盖面积及占比减少最多,减少了798164.10hm2,2015年覆盖度面积比2010年减少57.23%;二级植被覆盖面积占比呈现持续上升趋势且增加最多,增加473682.07hm2,2015年覆盖度面积是2000年的159.75%。2000~2015年,研究区植被覆盖表现为退化,退化面积为1376348.49hm2,占研究区面积49.42%,较植被覆盖度增加面积多出1261164.41hm2,是植被覆盖度增加面积的11.95倍。2000~2015年,一级植被覆盖重心向西北方向迁移49.78km,三级植被覆盖重心向东北方向迁移了4.24km,二级、四级和五级植被覆盖重心均向西南方向迁移,迁移距离分别为38.07km、53.00km、83.35km。3.基于NDVI和RVI两种植被指数,与实测采样点地上生物量数据进行统计分析,建立了NDVI-线性模型、NDVI-指数模型、NDVI-二次多项式模型、RVI-线性模型、RVI-指数模型、RVI-二次多项式模型共6种回归模型,其中RVI-二次多项式模型与草原地上生物量的相关性最高,相关系数达到0.911,预测决定系数为0.830,预估精度为85.31%;其次是RVI-线性模型,相关系数为0.908,预测决定系数为0.829,预估精度为78.52%;NDVI-二次多项式模型,相关系数为0.907,预测决定系数为0.822,预估精度为81.22%;NDVI-线性模型,相关系数为0.903,预测决定系数为0.814,预估精度为77.01%;NDVI-指数模型,相关系数为0.877,预测决定系数为0.768,预估精度为72.86%;RVI-指数模型,相关系数为0.854,预测决定系数为0.728,预估精度为70.11%;均通过P0.001检验。RVI-二次多项式模型是所建模型中相关系数、决定系数以及预测模型精度都高于其它5种模型,是研究区草原地上生物量遥感反演的最优模型,模型为:y=62.121x2+1146.7x-377.66,R2=0.830,n=67,预估精度为85.31%,可以较好的反映研究区草原地上生物量特征。
[Abstract]:The study of the aboveground biomass of the grassland is an important aspect in the study of the grassland ecology and the ecological economics, and the dynamic change of the vegetation cover and the distribution of the aboveground biomass of the grassland in real time are the prerequisite for the rational use and protection of the grassland resources. Taking the grassland of Xinjiang and Jingxian as the research area, the aboveground biomass of the grassland is the research object, the Landsat image is the main remote sensing data, the ground biomass data in the field actually measured sampling point is the modeling data, Based on the correlation analysis between the vegetation index NDVI, RVI and the measured value of the aboveground biomass of the grassland, a univariate linear model, an exponential model and a quadratic polynomial model are set up, and the model accuracy test is carried out by using the statistical analysis of SPSS. The optimal model of the aboveground biomass inversion of the grassland in the applicable study area is determined, and the remote sensing inversion of the aboveground biomass of the grassland is carried out by using the optimal model. The correlation coefficient of 0.7267 is obtained based on the linear correlation between the estimated value of the vegetation coverage of the grassland and the field measured value, and it is proved that the method can be used to study the vegetation coverage of the grassland in XinJiang and Jingxian. The average vegetation coverage of the study area was 4.2851, 4.3042, 4.4252, and 3.7524, respectively. The average vegetation coverage of 2000-2015 was decreased by 0.5327, among which, the vegetation coverage in the year of 2015 was the most, the decrease of 15.20%. The vegetation coverage of the five different grades is 54.82%, 60.08%, 5.52%, 266.82%, 772.37%, and the third-level vegetation cover conversion is the least, the conversion rate is 5.52%, the five-stage vegetation cover conversion rate is the largest and the conversion rate is 772.37%. The coverage area of secondary vegetation decreased by 57.23% in 2010, the coverage area of secondary vegetation increased by 57.23% in 2010, the coverage area of secondary vegetation increased by up to 473682.07 hm2, and the coverage area in 2015 was 159.75% in 2000. In 2000-2015, the vegetation cover of the study area was degraded and the degradation area was 1376348.49 hm2, accounting for 49.42% of the area of the study area. The increase of vegetation coverage is 1261164.41hm2, which is 11.95 times that of the increase of vegetation coverage. In the period 2000-2015, the center of gravity of the first-level vegetation covers the north-west direction for 49.78 km, the center of gravity of the three-level vegetation covers the northeast, 4.24km, the second, the fourth and the five-level vegetation cover the center of gravity to the south of the west, The migration distance is 38.07 km, 53.00 km and 83.35 km respectively. The NDVI-linear model, the NDVI-index model, the NDVI-quadratic polynomial model, the RVI-linear model, the RVI-exponential model, the RVI-quadratic polynomial model and the six regression models are established based on the NDVI and RVI vegetation indices. The correlation coefficient of the RVI-quadratic polynomial model and the aboveground biomass of the grassland is the highest, the correlation coefficient is 0.911, the prediction coefficient is 0.830, the prediction accuracy is 85.31%, the second is the RVI-linear model, the correlation coefficient is 0.908, the prediction determination coefficient is 0.829, and the prediction accuracy is 78.52%; The NDVI-quadratic polynomial model has a correlation coefficient of 0.907, a prediction determination coefficient of 0.822, an estimated accuracy of 81.22%, an NDVI-linear model, a correlation coefficient of 0.903, a prediction determination coefficient of 0.814, an estimated accuracy of 77.01%, an NDVI-index model, a correlation coefficient of 0.877 and a prediction determination coefficient of 0.768, The prediction accuracy is 72.86%, the RVI-index model, the correlation coefficient is 0.854, the prediction determination coefficient is 0.728, the estimated accuracy is 70.11%, and all passes the P0.001 test. The RVI-quadratic polynomial model is the best model for the remote sensing inversion of the aboveground biomass in the study area. The model is: y = 62.121x2 + 1146.7 x-377.66, R2 = 0.830, n = 67, and the estimated accuracy is 85.31%. The aboveground biomass of the grassland in the study area can be better reflected.
【学位授予单位】：新疆师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：Q948.1
本文编号：2497587
[Abstract]:The study of the aboveground biomass of the grassland is an important aspect in the study of the grassland ecology and the ecological economics, and the dynamic change of the vegetation cover and the distribution of the aboveground biomass of the grassland in real time are the prerequisite for the rational use and protection of the grassland resources. Taking the grassland of Xinjiang and Jingxian as the research area, the aboveground biomass of the grassland is the research object, the Landsat image is the main remote sensing data, the ground biomass data in the field actually measured sampling point is the modeling data, Based on the correlation analysis between the vegetation index NDVI, RVI and the measured value of the aboveground biomass of the grassland, a univariate linear model, an exponential model and a quadratic polynomial model are set up, and the model accuracy test is carried out by using the statistical analysis of SPSS. The optimal model of the aboveground biomass inversion of the grassland in the applicable study area is determined, and the remote sensing inversion of the aboveground biomass of the grassland is carried out by using the optimal model. The correlation coefficient of 0.7267 is obtained based on the linear correlation between the estimated value of the vegetation coverage of the grassland and the field measured value, and it is proved that the method can be used to study the vegetation coverage of the grassland in XinJiang and Jingxian. The average vegetation coverage of the study area was 4.2851, 4.3042, 4.4252, and 3.7524, respectively. The average vegetation coverage of 2000-2015 was decreased by 0.5327, among which, the vegetation coverage in the year of 2015 was the most, the decrease of 15.20%. The vegetation coverage of the five different grades is 54.82%, 60.08%, 5.52%, 266.82%, 772.37%, and the third-level vegetation cover conversion is the least, the conversion rate is 5.52%, the five-stage vegetation cover conversion rate is the largest and the conversion rate is 772.37%. The coverage area of secondary vegetation decreased by 57.23% in 2010, the coverage area of secondary vegetation increased by 57.23% in 2010, the coverage area of secondary vegetation increased by up to 473682.07 hm2, and the coverage area in 2015 was 159.75% in 2000. In 2000-2015, the vegetation cover of the study area was degraded and the degradation area was 1376348.49 hm2, accounting for 49.42% of the area of the study area. The increase of vegetation coverage is 1261164.41hm2, which is 11.95 times that of the increase of vegetation coverage. In the period 2000-2015, the center of gravity of the first-level vegetation covers the north-west direction for 49.78 km, the center of gravity of the three-level vegetation covers the northeast, 4.24km, the second, the fourth and the five-level vegetation cover the center of gravity to the south of the west, The migration distance is 38.07 km, 53.00 km and 83.35 km respectively. The NDVI-linear model, the NDVI-index model, the NDVI-quadratic polynomial model, the RVI-linear model, the RVI-exponential model, the RVI-quadratic polynomial model and the six regression models are established based on the NDVI and RVI vegetation indices. The correlation coefficient of the RVI-quadratic polynomial model and the aboveground biomass of the grassland is the highest, the correlation coefficient is 0.911, the prediction coefficient is 0.830, the prediction accuracy is 85.31%, the second is the RVI-linear model, the correlation coefficient is 0.908, the prediction determination coefficient is 0.829, and the prediction accuracy is 78.52%; The NDVI-quadratic polynomial model has a correlation coefficient of 0.907, a prediction determination coefficient of 0.822, an estimated accuracy of 81.22%, an NDVI-linear model, a correlation coefficient of 0.903, a prediction determination coefficient of 0.814, an estimated accuracy of 77.01%, an NDVI-index model, a correlation coefficient of 0.877 and a prediction determination coefficient of 0.768, The prediction accuracy is 72.86%, the RVI-index model, the correlation coefficient is 0.854, the prediction determination coefficient is 0.728, the estimated accuracy is 70.11%, and all passes the P0.001 test. The RVI-quadratic polynomial model is the best model for the remote sensing inversion of the aboveground biomass in the study area. The model is: y = 62.121x2 + 1146.7 x-377.66, R2 = 0.830, n = 67, and the estimated accuracy is 85.31%. The aboveground biomass of the grassland in the study area can be better reflected.
【学位授予单位】：新疆师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：Q948.1
【参考文献】
相关期刊论文 前10条
1 亚库普·约麦尔;阿里木江·卡斯木;;喀什市城市植被及其覆盖度研究[J];测绘科学;2016年03期
2 卢响军;周益民;侯秀玲;李杰;刘洋;张殷俊;;基于土地利用/覆盖的干旱绿洲区植被覆盖度变化——以新疆生产建设兵团第八师为例[J];中国生态农业学报;2015年02期
3 王先伟;刘梅;柳林;;MODIS光谱指数在中国西南干旱监测中的应用[J];遥感学报;2014年02期
4 除多;德吉央宗;普布次仁;姬秋梅;唐红;;藏北草地地上生物量及遥感监测模型研究[J];自然资源学报;2013年11期
5 万红梅;李霞;董道瑞;;基于多源遥感数据的荒漠植被覆盖度估测[J];应用生态学报;2012年12期
6 高添;徐斌;杨秀春;金云翔;马海龙;李金亚;于海达;;青藏高原草地生态系统生物量碳库研究进展[J];地理科学进展;2012年12期
7 穆少杰;李建龙;陈奕兆;刚成诚;周伟;居为民;;2001-2010年内蒙古植被覆盖度时空变化特征[J];地理学报;2012年09期
8 甘春英;王兮之;李保生;梁钊雄;李志文;温小浩;;连江流域近18年来植被覆盖度变化分析[J];地理科学;2011年08期
9 张飞;塔西甫拉提·特依拜;丁建丽;何祺胜;田源;买买提·沙吾提;王宏;桂东伟;;新疆典型盐渍区植被覆盖度遥感动态监测——以渭干河-库车河三角洲绿洲为例[J];林业科学;2011年07期
10 高清竹;段敏杰;李玉娥;万运帆;郭亚奇;刘硕;马欣;边多;;1981-2006年藏西北地区草地植被盖度动态变化分析[J];中国农业气象;2010年04期
相关硕士学位论文 前1条
1 韩旭;永定河流域植被生物量遥感反演研究[D];北京林业大学;2012年
，本文编号：2497587
本文链接：https://www.wllwen.com/shoufeilunwen/benkebiyelunwen/2497587.html
