Multi-factor modeling of above-ground biomass in alpine grassland: A case study in the Three-River Headwaters Region, China
Multi-factor modeling of above-ground biomass in alpine grassland: A case study in the Three-River Headwaters Region, China
复制标题
高寒草地地上生物量多因素模拟——以中国三河源区为例
DOI：
10.1016/j.rse.2016.08.014
复制
发表时间：
2016-12-01
影响因子：
13.5
通讯作者：
Xie, Hongjie
中科院分区：
文献类型：
--
作者：
Liang, Tiangang;Yang, Shuxia;Xie, Hongjie
本文以青海省南部牧区（即三江源地区）为研究对象，利用MODIS植被指数，结合15个站点的长期气候和草地监测数据，对估算高寒草地植被地上生物量（AGB）的各种方法进行了评价。结果表明：(1)12 a来，草地AGB和NDVI存在较大的空间差异，草地生长高峰期AGB均值在329 ~ 3653 kg DW/ha之间，对应的NDVI均值为0.25 ~ 0.72；(2)草地AGB受地理位置、地形、气候、土壤、草类等多种因素的影响。单因素AGB模型仅占牧草生长高峰期AGB变化的15-49%，其中基于ndvi的AGB模型在我们测试的所有线性遥感模型中表现最好（46%）；(3)虽然多因子模型（基于经纬度和草被、草高）估算AGB的效果最好（70%），但由于目前草高建模的困难，无法进行操作。备选和可操作的多因素模型f(x,y,c)（纬度，经度，草被）可以实现对AGB的合理估计（63%），草被由MODIS反射率建模，未来将结合无人机技术进一步改进。使用该模型f(x,y,c)，与基于NDVI的最佳单因素模型（RMSE为887 kg DW/ha）相比，AGB估计的均方根误差（RMSE）降低了20%（即151 kg DW/ha）。(C) 2016 Elsevier Inc.版权所有。
In this study, we evaluate various methods for estimating the above-ground biomass (AGB) of alpine grassland vegetation using Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation indices, in combination with long-term climate and grassland monitoring data collected at 15 site-specific stations, in the pastoral area of southern Qinghai Province (i.e., the Three-River Headwaters Region) of China. The results show that (1) over the past 12 years, there were considerable spatial variations in the grassland AGB and NDVI, with the average AGB in the peak period of grassland growth in the range of 329-3653 kg DW/ha, corresponding to an average NDVI of 0.25-0.72; (2) Grassland AGB is affected by various factors, such as geographic location, topography, climate, soil, and grass types. Single-factor AGB models only account for 15-49% of the variations in the grassland AGB during the peak period of grass growth, with NDVI-based AGB model to be the best (46%) among all linear remote sensing models we tested; and (3) although the multi-factor model (based on latitude, longitude, and grass cover and height) performs the best (70%) in estimating the AGB, it is not possible for operation due to the current difficulty of grass height modeling. The alternative and operational multi-factor model f(x,y,c) (latitude, longitude, and grass cover) can achieve reasonable estimation of AGB (63%), with the grass cover modeled from the MODIS reflectance, which would be further improved in conjunction with unmanned aerial vehicle technology in the future. Using this model f(x,y,c), the root-mean-square error (RMSE) of AGB estimation is reduced by 20% (i.e., 151 kg DW/ha) as compared with the best single-factor model based on the NDVI (RMSE of 887 kg DW/ha). (C) 2016 Elsevier Inc. All rights reserved.