Estimation Of Soil Organic Carbon Changes In Farmland In Fujian Province Based On Landsat Remote Sensing Image And1:50 000 Soil Database

Carbon pool of farmland plays an important role in the changes of global carbon pool,and its small changes may have a huge influence on the CO₂ concentration of atmospheric,leading to the global greenhouse effect.Currently,MODIS images and small and medium scale databases was often applied to the Remote Sensing retrieval of SOC at regional or national scales.In the study of county and provincial scales,the spatial variation of SOC could not be revealed due to data limitations.In this study,the province of Fujian in subtropical China was taken as the study area,we adopted Remote Sensing and Carbon Cycle Process model to estimate SOC.Landsat TM,Landsat 8 OLI remote sensing images,meteorological data,actual measurement SOC data of farmland,as well as 1:50 000 high accuracy soil database was applied to the model,aiming to combine the advantages of high temporal-spatial resolution of satellite images with the reliability of measured SOC.The accuracy of the model was evaluated using ground-based measured data and compared with the geostatistical estimates.Based on the two-stage SOCD and SOCS inversion results,the study identified the carbon source and sink of farmland in Fujian Province from1987 to 2016.At the same time,we analyzed the spatial and temporal differentiation of soil organic carbon pools in different soil types,landuse types,landforms and administrative areas,and identified the reason.The results can provide scientific guidance for the formulation of relevant policies.Meanwhile,it provided a reference for the estimation of SOC in farmland at regional scale,and provided a scientific basis for the sustainable use of subtropical cultivated land resources and food security.The main findings were as follows:1.Based on the remote sensing and carbon cycle process model,the SOC inversion model of farmland in 1987 and 2016 can estimate the soil carbon pool of farmland in Fujian Province.Specific performances were as follows:modeling and model verification have reached a very significant correlation level?P<0.01?;the MAPE,RMSE and TAA of SOCD measured and estimated values in 1987 and 2016were 0.49 and 0.65,0.62 and 0.80,0.84 and 0.78,respectively.2.Remote sensing and carbon cycle process models,geostatistical methods both had high estimation accuracy,which can estimate SOC in framland in the study area,but the former had obvious advantages in the detailed expression of SOC pattern.3.From 1987 to 2016,the SOCD and SOCS of farmland in Fujian Province increased by 0.08 kg·m^-22 and 10.50×10⁵ t,respectively,indicating that the farmlands in Fujian province as a whole acted as a weak“carbon sink”in the past 30 years.The spatial difference was obvious:in general,the carbon source was generally founded in the eastern coastal areas,while the carbon sink in the western inland areas.4.From the perspective of SOCD in farmland,yellow soil and paddy soil were at the forefront of organic carbon density,and in 1987 it were 3.72 kg·m^-22 and 3.55kg·m^-22 respectively,while 3.59 kg·m^-22 and 3.60 kg·m^-22 in 2016.Windy sand and coastal saline soils had been relatively low,2.63 kg·m^-22 and 2.90 kg·m^-22 in 1987,while 3.33 kg·m^-2and 3.24 kg·m^-2in 2016,respectively.The density of organic carbon in paddy fields was higher than dry land,which was 0.39 kg·m^-2and 0.17kg·m^-2in 1987 and 2016,respectively.The organic carbon density in the mountainous hilly areas was consistently higher than the average level,which was 3.52 kg·m^-2and3.64 kg·m^-22 in 1987 and 2016,respectively.5.From the perspective of SOCS in farmland,paddy soil had the highest organic carbon storage,accounting for more than 90%in 1987 and 2016.The carbon storage of calcareous soil,purplish soil,yellow soil and fluvo-aquic soil was small during the study period,and the total carbon storage accounted for less than 1%.Paddy field organic carbon storage was the largest,accounting for more than 80%during the study period,followed by dry land and irrigated land.The Mountain hilly area had the largest organic carbon reserves,followed by the valley basin area,plain platform area was the lowest.