Font Size: a A A

Estimating Forest Aboveground Biomass In Dongguan City Using Mixed-effects Model

Posted on:2019-08-11Degree:MasterType:Thesis
Country:ChinaCandidate:H N FengFull Text:PDF
GTID:2393330563985721Subject:Agriculture
Abstract/Summary:PDF Full Text Request
Forests are important terrestrial ecosystems that play an important role in the exchange of carbon cycle between the terrestrial biosphere and the atmosphere.Forests are also the largest carbon storage on land and the most economic carbon sinks.Forests absorb carbon dioxide from the atmosphere through photosynthesis and store it as aboveground biomass.Forest biomass accounts for 70% of the terrestrial ecosystem biomass,especially forest aboveground biomass,which accounts for 70% of the total forest biomass.Biomass is a key biophysical parameter for assessing and simulating terrestrial carbon storage and dynamic changes.Accurate estimation of regional aboveground biomass has important implications for understanding regional ecosystems and their responses to global warming,and provides a reference for the assessment of regional forest management treaties and policies.Since the 1950 s,the research on forest biomass has developed rapidly,and there have been many methods for estimating forest biomass.Especially with the rapid development of remote sensing technology in recent years.Combined with geo-statistical methods,remote sensing technology has become the current mainstream method for estimating forest biomass.However,the spatial resolution of remote sensing data,the characteristics of forest biomass in sample plots,the size of the study area,the researcher's knowledge and skills,etc.will all influence the choice of remote sensing estimation methods.In this study,a linear mixedeffects model was used for the quantitative estimation of forest aboveground biomass,we built a linear mixed effect model with random effects as forest types and assessed the ability of linear mixed effects model to estimate aboveground biomass,and obtained the following conclusions:1.According to the mixed effect model obtained,the spectral reflectance of different forest types has a random effect,and the differences in forest types have an impact on the vegetation index,thus affecting the aboveground biomass estimation.Therefore,when estimating forest aboveground biomass,adding forest classification information can improve model estimation accuracy.2.With the comparison between multiple regression model and the linear mixed effects model,we found out that the linear mixed effects model has fewer parameters for estimating forest biomass,and it has better fitting and higher estimation accuracy,the linear mixed effect model has a coefficient of determination of 0.689.(Multiple regression model determination coefficient is 0.513),and the root mean square error is 20% lower than the multiple regression model.At the same time,this research also shows that the linear mixed effects model can better cope with sparse data and can obtain good prediction results.3.The aboveground biomass of the forest areas in Dongguan(except the Great Barrier National Forest Park and Guanyinshan National Forest Park)was low,because the planting area of Litchi in Dongguan was large,and the biomass of litchi was generally low.However,the aboveground biomass of eucalyptus and Acacia accounts for half of the total biomass,and their total area is less than that of the Litchi.In summary,this study used the linear mixed effects model to estimate the forest aboveground biomass in Dongguan,and it turned out that the linear mixed-effects model can tackle with the imbalance of vegetation types in Dongguan.With the linear mixed effects model having forestry as a random effect,we not only considered the forest types but also took into account the forest types with fewer samples while estimating the aboveground biomass in Dongguan,and got a good prediction accuracy.
Keywords/Search Tags:Forest, Aboveground Biomass, Remote Sensing, Landsat, Tree Species Classification, Linear Mixed-effects model
PDF Full Text Request
Related items