| Soil organic carbon is closely linked to global climate change,and the use of ecological models integrated with soil databases is the basis for predicting changes in organic carbon under scenarios of future climate conditions and changes in farmland management practices.However,previous studies were mostly limited to medium and small scale soil databases,and lacked large,medium,and small scale series of studies,which affected the improvement of model prediction accuracy and key parameters.In this study,a typical subtropical region of China,Fujian Province,was selected as the study area,and a soil map"polygon"was used as the basic simulation unit,using"1:50,000（county level）,1:200,000（prefecture level）,1:500,000（provincial level）and 1:1,000,000（national level）",and the DNDC model（Denitrification and Decomposition）,based on meteorological data from 1980 to 2016 and soil property data in 2016.Twenty-one scenarios of common and prescribed future climate and farm management measures in Intergovernmental Panel on Climate Change（IPCC）were set to predict the dynamic changes of soil organic carbon from 2017 to 2053.The results can be used to quantify the simulation errors at other scales and reveal the scale effects by using the high precision soil database（1:50,000）at the regional scale on the one hand,and to clarify the main control factors at different scales and elucidate the variation mechanism by using a combination of structural equation modeling and linear regression on the other hand,which can provide theoretical references for the quantitative evaluation of uncertainty and the establishment of multi-scale high precision models using different mapping databases in agricultural soil organic carbon simulation.The main findings of the study are as follows:1.From the validation results of the large sample in 2016,the simulated values of soil organic carbon in Fujian Province at four scales of 1:50,000（n=6880）,1:200,000（n=6452）,1:500,000（n=6061）and1:1,000,000（n=6125）were all within the range of the measured values（100%）with significant correlation（p<0.01）,and the correlation coefficients r were 0.34～0.44,and the mean absolute error（MAE）and root mean square error（RMSE）were small,ranging from 4.27～4.69g·kg-1and 5.39～5.90 g·kg-1respectively,indicating that the DNDC model can be better applied to the simulation of soil organic carbon in paddy soil in Fujian Province.2.The results of the scenario analysis of the regional high-precision soil database（1:50,000）from 2017 to 2053 showed that the overall soil of paddy fields in Fujian Province showed a"carbon sink"effect under all climate factors and farm management measures（21 scenarios in total）,and the carbon sequestration rate ranged from 15 to 838 kg C·hm-2·a-1.The effect of farm management measures on soil carbon sequestration rate was more obvious than that of climatic factors.From different landscape types,the highest annual average carbon sequestration rate was found in the plains,ranging from 70 to 859 kg C·hm-2·a-1under the 21scenarios.From different soil subtypes,salinized paddy soils had the highest average annual carbon sequestration rate,ranging from 74 to 903kg C·hm-2·a-1under different scenario analyses.3.The results of scenario analysis for different spatial resolution soil databases（1:200,000 to 1:1,000,000）from 2017 to 2053 showed that the spatial scale caused significant differences in the simulation estimates of soil organic carbon under 21 scenario analyses of future climate and farm management practices.Taking the simulation results at the 1:50,000 scale as the benchmark,the relative deviations of the annual average carbon sequestration rates from 1:200,000 to 1:1,000,000 in paddy soils of Fujian Province ranged from 0.44%to 38.39%,and the farm management measures were much more influenced by the scale than the climate factors.From different landscape types,the average annual carbon sequestration rate in hilly mountainous areas was most influenced by scale,and the relative deviations from 1:200,000 to 1:1,000,000 scales ranged from 2.31%to 322.98%.From different soil subtypes,the average annual carbon sequestration rate of submergenic paddy soils was most affected by scale,with the relative deviations ranging from 2.82%to4206.73%from 1:200,000 to 1:1,000,000 scales,and the relative deviations of hydromorphic paddy soils were least affected by scale,ranging from 0.08%to 433.30%from 1:200,000 to 1:1,000,000 scales.4.In terms of the magnitude of the scenario analysis influenced by the scale effect,the annual average carbon sequestration rate under the scenario of no N fertilization and 6℃ temperature increase was the most influenced by the scale,and the relative deviations between 1:200,000and 1:1,000,000 scales ranged from 11.58%to 38.39%and 13.57%to20.87%,respectively,with the highest at 1:1,000,000 scales reaching38.39%.In contrast,the average annual carbon sequestration rate was least affected by scale under the scenarios of composite management and90%straw return rate,and the relative deviations ranged from 1.77%to2.99%and 0.44%to 1.92%at 1:200,000 to 1:1,000,000 scales,respectively,with the highest of 2.99%at 1:1,000,000 scale.Therefore,the high-resolution soil database is the optimal choice for elucidating the scale effects in geomorphologically complex regions.5.The results of structural equation modeling showed that initial soil organic carbon,clay particles and fertilizer were the main influence pathways in the four scales of soil database from 1:50,000 to 1:1,000,000,and the standardized path coefficients ranged from 0.35 to 0.62.Soil p H,slope and climate factors had less effect on soil carbon sequestration rate,and the total effect at the four scales ranged from 0.02 to 0.16.The rougher the soil map is,the soil organic carbon is gradually weakened by the initial soil properties and gradually enhanced by topographic,climatic and external factors. |
PDF Full Text Request