基于ADC和MODIS遥感数据的高寒草地地上生物量监测研究——以黄河源区为例

发布时间：2019-01-04 18:59

【摘要】：利用2015-2016年8月采集的黄河源区草地生物量数据和MODIS卫星遥感资料,结合农业多光谱相机(agricultural digital camera,ADC)获取的植被指数数据,比较分析3种_(ADC)植被指数(NDVI_(ADC)、SAVI_(ADC)和GNDVI_(ADC))与野外实测草地地上生物量(above-ground biomass,AGB)数据的相关性,筛选出适合构建草地AGB反演模型的_(ADC)植被指数;结合MODIS NDVI(记作NDVIMOD)构建草地地上生物量反演模型,采用留一法交叉验证方法评价各模型精度,确立适宜模拟研究区草地AGB的最优模型;并利用NDVI_(ADC)校正NDVI_(MOD),获得高分辨率、高精度的草地AGB遥感监测改进模型。结果表明,1)基于_(ADC)获取的3种植被指数中,NDVI_(ADC)与高寒草地地上生物量关系最为密切,其次为SAVI_(ADC),拟合效果最差的是GNDVI_(ADC);2)基于NDVI_(ADC)建立的草地AGB监测模型的精度(RMSEP介于383.55~393.18kg DW/hm2;r范围为0.65~0.66)远高于NDVI_(MOD)的模型精度(RMSEP介于421.08~427.00kg DW/hm~2;r范围为0.55~0.58),NDVI_(ADC)反演得到的草地AGB更接近于黄河源区草地实际生物量,且相较于NDVI_(ADC),NDVI_(MOD)的样本值整体偏高;3)在NDVI_(ADC)构建的4类模型中,线性和乘幂模型模拟研究区草地AGB的能力较好,但线性模型精度更高(y=3248.93×NDVI_(ADC)-305.59,RMSEP=383.55kg DW/hm~2,r=0.66),该模型为黄河源区草地生物量的估测提供了一个新型且易操作的方法;4)NDVI_(ADC)与NDVIMOD相关性较高,利用NDVI_(ADC)校正NDVI_(MOD)可以改进草地AGB遥感反演模型,优化模型为y=2455.54×NDVI_(MOD)-301.69。该模型可在大尺度范围内估测黄河源区的草地生物量,且模型精度接近于地表测量法的监测精度。
[Abstract]:Based on the grassland biomass data and MODIS satellite remote sensing data collected from the source region of the Yellow River from 2015 to 2016, the vegetation index data obtained from the agricultural multispectral camera (agricultural digital camera,ADC) were used. The correlation between three _ (ADC) vegetation indices (NDVI_ (ADC), SAVI_ (ADC) and GNDVI_ (ADC) and field measured grassland aboveground biomass (above-ground biomass,AGB) data was compared and analyzed. The _ (ADC) vegetation index suitable for constructing grassland AGB inversion model was selected. Based on MODIS NDVI (NDVIMOD), the model of grassland aboveground biomass inversion was constructed, and the precision of each model was evaluated by the method of cross-validation, and the optimal model for simulating grassland AGB in the study area was established. The improved model of grassland AGB remote sensing monitoring with high resolution and high precision was obtained by using NDVI_ (ADC) correction NDVI_ (MOD),. The results showed that: 1) among the three cropping cover indices obtained based on _ (ADC), NDVI_ (ADC) was most closely related to the aboveground biomass of alpine grassland, followed by GNDVI_ (ADC);, which was the worst fitting result of SAVI_ (ADC),. 2) the precision of grassland AGB monitoring model based on NDVI_ (ADC) (RMSEP is 0.65 ~ 0.66 in 383.55~393.18kg DW/hm2;r range) is much higher than that of NDVI_ (MOD) (RMSEP is between 421.08~427.00kg DW/hm~2;). The range of r is 0.55 ~ 0.58), the AGB obtained from NDVI_ (ADC) inversion is closer to the actual biomass of grassland in the source region of the Yellow River, and the sample value is higher than that of NDVI_ (ADC), NDVI_ (MOD). 3) among the four kinds of models constructed by NDVI_ (ADC), the linear and power models have better ability to simulate the grassland AGB in the study area, but the accuracy of the linear model is higher (YY 3248.93 脳 NDVI_ (ADC) -305.59 DW/hm~2,r=0.66). The model provides a new and easy to operate method for estimating grassland biomass in the source region of the Yellow River. 4) there is a high correlation between NDVI_ (ADC) and NDVIMOD. Using NDVI_ (ADC) correction NDVI_ (MOD) can improve the remote sensing inversion model of grassland AGB. The optimized model is yong 2455.54 脳 NDVI_ (MOD) -301.69. The model can estimate the grassland biomass in the source area of the Yellow River on a large scale, and the accuracy of the model is close to that of the surface survey method.
【作者单位】：草地农业生态系统国家重点实验室兰州大学草地农业科技学院;
【基金】：国家自然科学基金项目(31672484,31372367,41401472) 中国气象局气候变化专项项目(CCSF201603) 长江学者创新团队发展计划(IRT13019)资助
【分类号】：S812

【相似文献】