Open AccessEstimating the net ecosystem exchange for the Dayekou Guantan forest by integrating MODIS and Flux data
Author(s) -
Xiaoliang Shi,
Na Zhang,
Mengyue Wu,
Hao Ding,
Chong Cheng,
Shu Yu
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/461/1/012081
Subject(s) - environmental science , enhanced vegetation index , eddy covariance , vegetation (pathology) , remote sensing , primary production , flux (metallurgy) , leaf area index , scale (ratio) , ecosystem , carbon flux , atmospheric sciences , normalized difference vegetation index , geography , vegetation index , ecology , cartography , medicine , materials science , pathology , geology , metallurgy , biology
Estimating net ecosystem carbon exchange (NEE) on regional and global scales is important for the carbon cycle and the greenhouse effect. The eddy covariance technique provides long-term continuous monitoring of site-specific NEE at the tower footprint scale. Remote sensing technology can continuously and systematically monitor multiple information of terrestrial ecosystem at regional scale, and it becomes an important tool to extend NEE to large scale combining with flux data. In present study, according to TG model, we established several new NEE models (using linear regress method based on single variable or multiple variables) that are suitable for Dayekou Guantan forest station with Moderate-resolution Imaging Sepctroradiometer (MODIS) products. Variables include enhanced vegetation index (EVI), land surface water index (LSWI) and land surface temperature (LST, including daytime LST and nighttime LST´). Compared with models based on single variable, models based on EVI, LSWI and LST have the best effect. Results showed that this method had good precision (2008) (R 2 and RMSE reached 0.8014 and 0.7364, respectively) and generally captured the expected seasonal patterns of NEE. We validated the model using independent flux (2009), which demonstrated this method performed well for estimating NEE (R 2 and RMSE reached 0.8618 and 0.5538, respectively). In addition, during the whole process, variables had obvious seasonal dynamical characteristics and closely related to NEE seasonal dynamics. However, uncertainty still existed in this method. In future research, more influencing factors should be selected to simulate NEE more accurately and extend the scope of research to a larger extent.