基于Landsat 8 OLI数据的砒砂岩区生物量遥感估算
刘雨晴, 闫 峰, 陈俊翰

(中国林业科学研究院 荒漠化研究所, 北京 100091)

生物量; 遥感估算; 砒砂岩区

Applicating Landsat 8 OIL to Estimate Biomass in Pisha Sandstone Area
LIU Yuqing, YAN Feng, CHEN Junhan

(Institute of Desertification Studies, Chinese Academy of Forestry, Beijing 100091, China)

biomass; remote sensing estimation; Pisha sandstone area

备注

为了评价砒砂岩区植被生长状况和地上生物量的空间分布特征,采用Landsat 8 OLI和同期实测生物量数据对砒砂岩区生物量估算方法进行了研究。结果 表明:(1)NDVI,RVI,SAVI,MSAVI与地上生物量(Above ground biomass,AGB)的相关性显著,其中MSAVI和AGB的相关性最高(R2=0.4416),SAVI次之(R2=0.3923),NDVI(R2=0.1375)和RVI(R2=0.1306)相对最低,NDVI和RVI在荒漠生态系统生物量遥感估算中并不是效果最佳的植被指数;(2)高斯低通3×3滤波核滤波的MSAVI与AGB的相关性(R2=0.4757)高于未滤波处理图像,滤波后建立的AGB-MSAVI_GLPF3估算模型平均相对误差MRE为13.41%,模型具有较高的估算精度;(3)遥感估算2019年砒砂岩研究区总AGB为9.2×105 t,其中AGB低值区的面积占比为13.03%,AGB中值区面积占比为47.56%,AGB高值区面积占比为39.41%。荒漠生态系统AGB与MSAVI的相关性显著,基于高斯低通滤波建立起的AGB-MSAVI_GLPF3模型可以较好地实现砒砂岩区AGB遥感估算。

In order to evaluate the vegetation growth status and the spatial distribution characteristics of above ground biomass(AGB)in the Pisha sandstone area, Landsat 8 OLI image and in situ AGB data in the same period were used to study the AGB estimation method in the Pisha sandstone area of the Ordos Plateau. The results show that:(1)NDVI, RVI, SAVI and MSAVI had significant correlations with AGB; the correlation coefficient between MSAVI and AGB was the highest(R2=0.4416), correlation coefficient between SAVI and AGB was also higher(R2=0.3923), while the correlations between NDVI, RVI and AGB were relatively low, and the coefficients of determination were 0.137 5 and 0.130 6, respectively; among the four commonly used vegetation indexes, NDVI and RVI were not the best ones for biomass estimation in desert ecosystems;(2)the correlation between AGB and MSAVI filtered by Gaussian low pass filtering with kernel size 3×3 was higher than the image without filtering; the average relative error of AGB-MSAVI_GLPF3 estimation model established with filtering process was 13.41%, and the model had a higher estimation accuracy;(3)the total AGB of the study area in Pisha sandstone was 9.2×105 t in 2019, including 13.03% of low value AGB area, 47.56% of middle value AGB area and 39.41% of high value AGB area. The correlation between AGB and MSAVI was significant in desert ecosystem and the AGB-MSAVI_GLPF3 model established based on Gaussian Low Pass Filter could estimate AGB accurately in the Pisha sandstone area.