MODIS NDVI饱和性对高寒草甸草地生物量遥感估测的影响——以青藏高原东缘为例  

MODIS NDVI saturation assessment of alpine meadow grassland biomass estimation using remote sensing:a case study in the eastern edge of the Qinghai⁃Tibet Plateau

作　　者：胡仁杰 陈璇黎 陈金 张世荣 匡依利 于红妍 吉汉忠 赵晓军 宜树华[1,2] 孟宝平 李猛 HU Renjie;CHEN Xuanli;CHEN Jin;ZHANG Shirong;KUANG Yili;YU Hongyan;JI Hanzhong;ZHAO Xiaojun;YI Shuhua;MENG Baoping;LI Meng(Institute of Fragile Eco-Environment,Nantong University,Nantong 226007,China;School of Geographic Sciences,Nantong University,Nantong 226007,China;Qinghai Service and Guarance Center of Qilian Mountain National Park,Xining 810000,China;Qinghai Province Grassland Station in Haibei Prefecture,Haibei 812200,China)

机构地区：[1]南通大学脆弱生态环境研究所,南通226007 [2]南通大学地理科学学院,南通226007 [3]祁连山国家公园青海片区保障服务中心,西宁810000 [4]青海省海北州草原站,海北812200

出　　处：《生态学报》2024年第14期6357-6372,共16页Acta Ecologica Sinica

基　　金：国家自然科学基金(42071056);2022中央林草生态保护恢复资金,北林草[2022]134号资助;南通市科技计划(JC12022061)。

摘　　要：自20世纪70年代开始,归一化植被指数(NDVI)在天然草地地上生物量估测研究中得到了广泛应用。然而,NDVI对高密度植被生物量遥感估测存在饱和现象,使草地生物量遥感估测有较大的不确定性。以青藏高原东缘高寒草甸为例,基于比值植被指数(RVI)探讨了NDVI的饱和性,并评估了NDVI饱和性对高寒草甸地上生物量时空动态变化分析的影响。结果表明:(1)虽然基于中分辨率成像光谱仪(MODIS)NDVI构建的草地地上生物量估测模型精度较基于RVI构建的估测模型高,但模型对高寒草地地上生物量(生物量大于2314.627 kg/hm^(2))灵敏度较RVI估测模型低,即NDVI阈值大于0.73时,估测模型呈现饱和现象(低估了草地地上生物量);(2)结合RVI和NDVI的相关关系,对饱和部分NDVI遥感植被指数进行校正,校正后最优地上生物量遥感估测模型为线性模型(y=5908.5x-2198.9,R=0.6190,RMSE=902.41 kg/hm^(2)),较调整前RMSE降低了11.72 kg/hm^(2);(3)就NDVI饱和性空间分布而言,从全年6月—9月初(全年第161—257天)饱和性呈现先自东南向西北延伸,后自西北向东南消退的变化趋势,平均低估值介于158.45—293.92 kg/hm^(2)之间,最大低估值出现在8月初(全年第225天),超过600 kg/hm^(2);(4)此外,NDVI饱和性对草地地上生物量年际动态变化趋势分析具有较大的影响,去除饱和性影响后草地地上生物量基本不变的区域减小了21.44%,而年际变化小于-10 kg/hm^(2)和大于30 kg/hm^(2)的区域分别增加了8.48%和16.19%。研究探讨了NDVI饱和性对草地地上生物量遥感估测的影响,以期为精确评估高寒草地地上生物量提供理论依据,同时也为高寒草地资源可持续发展提供科学依据。Since 1970s,the normalized difference vegetation index(NDVI)has been widely used in aboveground biomass estimation in natural grasslands.However,there was a saturation phenomenon in NDVI⁃based estimation model in high⁃density vegetation biomass,which led to significant uncertainty in grassland biomass estimation.The alpine meadow grassland on the eastern edge of the Qinghai⁃Tibet Plateau was used to explore the saturation of NDVI based on the ratio vegetation index(RVI),and the impact of NDVI saturation on the spatiotemporal dynamic variation analysis of aboveground biomass was evaluated.The results showed that:1)Although the accuracy of the aboveground biomass estimation model based on NDVI was higher than that model based on RVI,the sensitivity of the model in high aboveground biomass(biomass greater than 2314.63 kg/hm^(2))was lower than that model based on RVI.In other words,when the NDVI was greater than 0.73,the estimation model presented a saturation phenomenon(underestimating aboveground biomass);2)Combined with the RVI,the saturated NDVI was adjusted.Then the optimal biomass remote sensing estimation model was constructed(y=5908.5x-2198.9,R=0.6190,RMSE=902.41 kg/hm^(2)),the RMSE decreased 11.72 kg/hm^(2) than before;3)From June to early September of the year(161—257 d),the saturation of NDVI⁃based aboveground biomass estimation model showed extending from southeast to northwest firstly,and then disappeared from northwest to southeast.The average underestimation aboveground biomass ranged from 158.45 kg/hm^(2) to 293.92 kg/hm^(2),and the maximum value occurred in early August(the 225th day of the year),exceeding 600 kg/hm^(2);4)In addition,the saturation of NDVI⁃based model had a significant impact on the analysis of annual dynamic variation in aboveground grassland biomass.After removing the saturation effect,the area where grassland aboveground biomass remained unchanged decrease by 21.44%,while the area with less than-10 kg hm^(-2) a^(-1) and greater than 30 kg hm^(-2)a^(-1) increased by 8.4

关 键 词：高寒草甸地上生物量 NDVI饱和性阈值 遥感估测 青藏高原东缘 

分 类 号：S812[农业科学—草业科学]