Font Size: a A A

Research On Global Forest Soil Respiration Dynamics Based On Artificial Neural Network Model

Posted on:2021-01-21Degree:MasterType:Thesis
Country:ChinaCandidate:Q Y TianFull Text:PDF
GTID:2370330611490131Subject:Ecology
Abstract/Summary:PDF Full Text Request
Forest is the main body of terrestrial ecosystem and the main regulator of atmospheric CO2 concentration.Land-use change has become an important driving force for changes in percentages of forest cover.It is very important to accurately estimate the forest cover,forest increase and loss and the global forest carbon emissions for the formulation of forest ecosystem management strategies and the estimation of land ecosystem carbon flux.Based on the updated annual soil respiration database of global forest,global climate data and high-resolution global forest change?GFC;30m resolution?,the objectives of this study were:?1?to calculate the variation in global forest areas;?2?to build an artificial neural network model based on k-fold cross validation;?3?to estimate and predict variation in forest soil respiration caused by global forest areas.This study focused on the dynamic changes of the coverage area,forest increase and annual loss of three forest types?boreal forest,temperate forest and tropical forest?from 2000 to 2017,and discussed the temporal and spatialdistribution of soil respiration caused by the dynamic changes of global forest.The main results are as follows:?1?Based on the same model inputs,three statistical models?exponential model,linear model,multiple linear regression model?were compared with the artificial neural network model.The results show that the prediction accuracy of the artificial neural network model was higher than those of the other three models.This model was also known as a soil respiration prediction model with the highest accuracy in global forest.?2?Based on GFC data,the forest loss area from 2001-2017 showed an upward trend with the change of time?P <0.05?,which was shown as: tropical forest?6,448 km2?> boreal forest?2,371 km2?> temperate forest?984 km2?.?3?Based on the global forest area in 2000,the average of global forest soil respiration in 2000-2017 was 48.56 ± 0.52 Pg C yr-1,and the mean values of soil respiration of boreal forest,temperate forest and tropical forest were 6.25±0.27 Pg C yr-1 ?8.9± 0.23 Pg C yr-1 ?33.41± 0.36 Pg C yr-1,respectively.The estimated boreal forest soil respiration increased 0.14 Pg C yr-1 for each 10 mm increase in the mean annual air precipitation?P = 0.002?;the estimated tropical forest soil respiration increased 1.56 Pg C yr-1 for each 1? increase in the mean annual air temperature?P = 0.032?.From 2001 to 2013,the total increment of potential global forest soil respiration caused by forest increase was 0.778 Pg C,and the annual average soil respiration was 0.060 ± 0.001 Pg C yr-1.Among which the total increments of potential soil respiration of boreal forest,temperate forest,and tropical forest were 0.112 Pg C,0.294 Pg C,and 0.372 Pg C,respectively.With each increment of 10000 km2 yr-1,the increment of potential soil respiration increased by 0.006 Pg C yr-1,0.010 Pg C yr-1 and 0.010 Pg C yr-1,respectively?P < 0.001?.From 2001 to 2017,the total reduction of potential global forest soil respiration caused by the loss of forest area was3.749 Pg C,and the annual growth rate of soil respiration was 0.011 Pg C yr-1?P<0.0001?,and the potential soil respiration reduction caused by the area loss of boreal forest,temperate forest and tropical forest were 0.451 Pg C,0.734 Pg C,and 2.565 Pg C,respectively.With the change of time,the annual growth rates of potential soil respiration in boreal forest,temperate forest and tropical forest were 0.001 Pg C yr-1,0.0011 Pg C yr-1and 0.009 Pg C yr-1?P < 0.01?.For every10,000 km2 yr-1 reduction in forest loss area,the potential reduction in soil respiration in boreal forest,temperate forest and tropical forest was reduced by0.004 Pg C yr-1,0.009 Pg C yr-1 and 0.010 Pg C yr-1,respectively?P <0.0001?.?4?Based on the global forest area in 2000,the spatial distribution maps of global forest soil respiration from 2000 to 2017 showed that the average soil respiration rates of boreal forest,temperate forest and tropical forest were 525 g cm-2 yr-1,934 g cm-2 yr-1 and 1231 g cm-2 yr-1,respectively.Globally,the regions with higher and lower soil respiration were respectively distributed in tropical forest and boreal forest,and the spatial distribution of soil respiration was controlled by temperature and precipitation.From 2001 to 2013,the potential increase of soil respiration caused by the increase of the area of boreal forest,temperate forest and tropical forest was mainly distributed in boreal coniferous forests?Russia,Finland,Sweden,Canada?,temperate coniferous forests and temperate broadleaf mixed forests?Northwestern United States,Southwestern Canada,Southeastern United States,Heilongjiang province,China,Ukraine,Italy,France and other European countries?and tropical and subtropical moist broadleaf forests?Mexico,Cuba,Brazil,Southeast China,Democratic Republic of Congo,Indonesia,etc.?.From 2001 to 2017,the potential soil respiration reduction caused by the area loss of boreal forest,temperate forest and tropical forest were mainly distributed in boreal coniferous forests?Russia,Alaska,Canada?,temperate coniferous forests?Western United States,British Columbia,Heilongjiang province,China?,temperate broadleaf mixed forests?southeast coastal area of the United States?,tropical and subtropical moist broadleaf forests?Brazil,Indonesia,Malaysia,Africa?,and tropical and subtropical dry broadleaf forests?Cambodia?,Brazil,Indonesia and Malaysia have intensive soil respiration reduction,which was related to local deforestation.
Keywords/Search Tags:k-fold cross validation, artificial neural network model, Forest change, soil respiration dynamics
PDF Full Text Request
Related items