Estimating Wetland Organic Carbon Storage In The Greater Khingan Mountains Based On Remote Sensing And Measured Data

Wetlands have the strongest function and the highest value of ecosystem services,the richest biodiversity,the greatest significance of protection,the strongest carbon sequestration potential and capability and the highest purification ability of ecosystem.Wetlands are usually called as kidneys of the Earth and nature's supermarkets.Wetlands are large carbon pools and are highly productive ecosystems that have a critical impact on the global carbon cycle.Wetland carbon sources and sinks were payed widely attention to,and a slight change in wetland organic carbon can have great impacts on global terrestrial carbon cycling.Boreal wetland ecosystems are usually located in middle and high latitudes region with a large terrestrial carbon reservoir.The borealwetland SOC is thus sensitive to climate change,and a slight change may influence the carbon budgets.As a typical boreal-wetland region,the Greater Khingan Mountains(GKM)have the second largest extent of permafrost in China and store large carbon pools.Evaluating accurately wetland organic carbon storage has important significance for the conservation,utilization,management and ecosystem restoration of wetland ecosystem,provides scientific basis for the coordinated development between ecosystem environment and economy,which also provides scientific reference for the studies of wetland carbon cycling and has critical influence on the estimation of global carbon storage.The wetland-type systems were confirmed through field surveys in the GKM.Landsat 8 OLI and GF-1 images,object-based classification and manual interpretation –synergies methods were used for extracting wetlands and mapping wetlands.The overall accuracy of wetland classification was high and can satisfy this study.Based on the field soil samples,the geographic information system(GIS)technology and ordinary kriging interpolation were utilized to model the spatial distribution of wetland soil organic carbon density and storage.This study analyzed the horizontal and vertical distributions of soil organic carbon concentration and their driving-force factors in the Arc GIS platform.Then,the wetland soil organic carbon(SOC)density and storage were estimated.Based on Landsat 8 OLI images,31 variables were selected to model wetland biomass.Based on the field samples of wetland biomass,the multiple linear regression method was used to build models to invert wetland biomass in the GKM.Combined wetland biomass and carbon coefficients,wetland vegetation organic carbon density was calculated.The wetland carbon storage equals soil organic carbon plus vegetation carbon.Through this study,the main conclusions are followed:(1)The wetland-type systems included grass wetlands,shrub wetlands and forest wetlands,which were confirmed through field surveys in the GKM.Fusion of Landsat 8 OLI and GF-1 images could optimize spatial and spectral distribution in the e Cognition 8.64 platform,which is useful to improve classification accuracy.In term of phenology,different wetland vegetation types have some differences of spectrum and texture during them and also with other landuse types.The images in that time were selected to extract wetlands in the GKM.In this study,multi-temporal satellite images were selected in late June(or early July)and late August(or early September).(2)In the e Cognition 8.64 platform,Landsat 8 OLI and GF-1 images were segmented to objects using multi-resolution segmentation for object-based classification.Then,texture,location,shape,context,mean value,brightness,standard deviation,NDVI,NDWI,RVI,DVI,PVI,SAVI and TSAVI were considered as rule variables to build decision trees.Grass wetlands,shrub wetlands and forest wetlands and non-wetland types in the GKM were extracted combining object-based classification and manual interpretation –synergies methods.The results of wetland classification show that grass wetlands,shrub wetlands and forest wetlands were 33,140.31 km2,3,763.81 km2 and 877.26 km2,respectively.(3)Based on wetland biomass from field samples,combined 31 variables including the reflectance of 7 bands from Landsat 8 OLI,NDVI,NDWI,RVI,DVI,PVI,SAVI,TSAVI,MSAVI,RDVI,NIR/G,EVI,TVI,RI,NDI,BI,GVI,WI,PC1,PC2,PC3,MNF1,MNF2,MNF3 and MNF4,using the multiple linear regression mathod built the biomass inversion models and model wetland vegetation biomass distribution.The results show the more variables were utilized,the higher model accracy was in grass wetlands and forest wetlands.The biomass model for shrub wetlands had certain predictive power,but the predictive accuracy was lower than grass wetlands and forest wetlands.R2,RMSE and r RMSEwere used to evaluate model accuracy.The most of biomass in wetlands distributed 0-4 kg/m2,the higher carbon storage distributed north of the GKM.(4)The differences of soil organic carbon concentration(SOCc)in the 0-30,30-60 and 60-100 cm intervals were significant(p<0.05).The SOCc decreased significantly with depths.The coefficients of variation increased with depths.The SOCcs during different wetland types in the 0-30 cm were significantly different,and the SOCcs in shrub wetlands and forest wetlands were significantly higher than that in grass wetlands(p<0.05).The differences in the same interval and different wetland types significantly decreased with depths(p>0.05).(5)Based on statistical analysis,wetland types played an important roles in influencing the wetland SOCc.Soil types had few impacts on the SOCc.Grass wetland SOCcs were significantly negative correlation with elevation(p<0.05)and positive correlation with precipitation(p<0.01).Shrub wetland SOCcs were significantly positive correlation with precipitation(p<0.05).Forest wetland SOCcs were significantly negative correlation with temperature(p<0.01)and positive correlation with precipitation(p<0.01).Wetland SOCcs had high correlation with climate factors and low correlation with topographic factors.The influence of considering factors decreased with depths.(6)The SOC density in wetlands was modelled using the ordinary kriging interpolation method,then the SOC storage was calculated.The SOC densities in the 0-30,30-60 and 60-100 cm intervals were 28.43 kg/m2,17.45 kg/m2 and 15.60 kg/m2,respectively.The SOC storages in grass wetlands,shrub wetlands and forest wetlands were 1,723.39,276.89 and 57.54 Tg C,respectively.The SOC densities in the 0-30 cm interval were significantly different(p<0.05),the order was shrub wetlands > forest wetlands > grass wetlands.The SOC stored in the 0-30 cm interval was more than 50% of all SOC storage,and decreased with depths.(7)The total wetland storage in the GKM was 2.21 Pg C.The mean wetland organic dendity was 58.50 kg C·m-2.The lowest wetland type was grass wetlands,but they had highest SOC storage,because they had the most area.