Effects Of Mapping Scale On Soil Organic Carbon Dynamic Simulation In The Uplands Of The Northern Jiangsu Province

Uplands account for over 70%of total farmland area in China,it will be a great potential carbon sink in the future through the reasonable agricultural management measures.Process-based models integrated with high-resolution soil databases play an important role in quantifying soil organic carbon(SOC)dynamic change and optimizing fanning management.At present,the studies are primarily limited to medium and small mapping scales,while studies on large scales are rarely reported.Furthermore,it is still unclear how mapping scales affect soil organic carbon dynamic simulation.This study was carried out using DNDC(DeNitrification-DeComposition)model to investigate the SOC dynamic change from 1980 to 2009 of uplands in the 29 counties(or cities)of the northern Jiangsu Province.Systematic analysis was performed of how the six scales,i.e.1:50,000,1:250,000,1:500,000,1:1,000,000,1:4,000,000 and 1:10,000,000.used in the six soil databases established for the four soil sources of soil profile data,namely,"Soils of County","Soils of Prefecture","Soils of Province" and"Soils of China",affected simulation of SOC dynamic change.The smallest simulation unit of the spatial soil databases consisted of the patches.Simultaneously,compared with the most detailed 1:50,000 soil database based on 983 upland soil profiles containing 17024 polygons,the other five databases varied in relative deviation of the simulation of SOCD and SOCS.The results could provide the theoretical supports for selecting proper mapping scales and formulating reasonable region agricultural management measures.The main research conclusions were as follows.1.From the evaluation of the DNDC model,the simulated average SOC content was 13.66 g kg-1 from 1999 to 2007.By comparison,the observed average SOC content was 13.33 g kg-1 during the same study period.The E was 4.34%,which demonstrated that the DNDC model was satisfactory for modeling SOC of uplands.Furthermore,the RMSE and MAE were 0.53 g kg-1 and 0.89 g kg-1,respectively,which indicated encouraging modeled results for the northern Jiangsu Province as well.2.The modeled results based on the 1:50,000 database indicated that an increase of 37.89 Tg C in the top layer(0-50 cm)for an area of 3.93 Mha in the upland soils cultivated with summer maize and winter wheat in the northern Jiangsu Province from 1980 to 2009,with the average annual SOC change of 321.73 kg C ha-1 y-1.The order of the average annual SOC changes of the eight great soil groups was:Fluvo-aquic soil(338.42 kg C ha-1)>Purplish soil(330.88 kg C ha-1)>Cinnamon soil(321.51 kg C ha-1)>Brown soil(309.07 kg C ha-1)>Saline soil(303.50 kg C ha-1)>Lime concretion black soil(300.12 kg C ha-1)>Lithosol(223.70 kg C ha-1)>Limestone soil(187.96 kg C ha-1).In general,most upland soils were a strong sink of atmospheric CO2 during the period of 1980-2009,and these results were consistent with the initial physical and chemical properties of soil,the applications of chemical fertilizer and farmyard manure,the changes of fainfall and temperature in this region.3.The SOC dynamic change in the uplands of the northern Jiangsu Provinve was effected by scales.The average annual soil organic carbon densities(mean SOCD)and storages(mean SOCS)of 1:50,000,1:250,000,1:500,000,1:1,000,000,1:4,000,000 and 1:10,000,000 databases during the period of 1980-2009 were 30.96 Mg C ha-1 and 121.51 Tg C,32.89 Mg C ha-1 and 131.92 Tg C,33.03 Mg C ha-1 and 134.00 Tg C,34.02 Mg C ha-1 and 130.11 Tg C,34.08 Mg C ha-1 and 138.61 Tg C,25.14 Mg C ha-1 and 99.54 Tg C,respectively.The soil database of 1:50,000 contained 17024 polygons representing 983 upland soil profiles was selected as a baseline,the relative deviations of the mean SOCD and mean SOCS in other mapping scale databases with various soil profile data sources were 6.24%and 8.56%,6.70%and 10.28%,9.90%and 7.07%,10.10%and 14.07%,18.80%and 18.09%,respectively.In general,with the reduction of mapping scales,the relative deviations were increasing.T test showed that extremely significant differences existed between the assessment based on the 1:50,000 database of"Soils of County" and the assessments based on the databases of other scales(p<0.001).The results also indicated that the more precise soil profile data sources and mapping scale databases was necessary to better simulate SOCD and SOCS of upland soils at the regional or national scales.4.There was significant difference of the simulated results among the eight great soil groups under six mapping scales.The relative deviations of the mean SOCD and mean SOCS in 1:250,000-1:10,000,000 mapping scale databases were 1.16%-62.15%and 3.39%-52.51%,4.19%-46.81%and 4.35%-60.68%,12.33%-32.38%and 12.45%-74.54%,0.14%-7.84%and 1.65%-604.03%,10.09%-33.27%and 3.76%-84.67%,respectively.Relatively speaking,the relative deviations of the limestone soil were 18.55%-62.15%and 4.35%-604.03%,which were the biggest.The fluvo-aquic soils covers about 2.07 Mha and ranks first in all the upland soil groups,its relative deviations were 11.09%-19.21%and 9.99%-53.67%,which were bigger.In contrast,the relative deviations of the brown soil were smaller,and the least values were 0.14%and 1.65%.The results indicated that great soil groups present larger effected by different mapping scales than the whole region.It also suggested that the homogenization of great soil groups in the mapping scales' reducing will lead to the uncertainty of space expression.5.It also had larger effects of the simulated results among the different administrative areas under six mapping scales.The relative deviations of the mean SOCD and mean SOCS in 1:250,000-1:10,000,000 mapping scale databases were 1.05%-26.86%and 0.31%-38.42%,1.93%-33.09%and 2.00%-86.29%,1.54%-47.18%and 0.63%-48.11%,3.22%-84.60%and 0.98%-142.50%,0.92%-44.37%and 0.71%-75.35%,respectively.Relatively speaking,the relative deviations of Huai'an and Funing were 3.59%-59.16%and 13.48%-142.50%,2.40%-84.60%and 2.18%-126.57%,respectively,which were the bigger.In contrast,the relative deviations of Suqian and Pizhou were 4.69%-18.76%and 0.63%-9.34%,1.93%-20.71%and 0.88%-18.58%,respectively,which were the smaller.In general,the relative deviations of 1:250,000 and 1:4,000,000 mapping scale databases were least and biggest,respectively.The results indicated that the change of soil organic carbon storage is mainly controlled by the soil organic carbon density,and it should be considered to conversing mapping scales in soil organic carbon estimation and policy-making with large or medium scales.6.Initial SOC content was the most sensitive soil factors to the average annual SOCD change of different mapping scales for the entire study region.Pearson's test showed that the mean SOCD had significant positive correlation with initial SOC content at the levels of p<0.01,the correlation coefficient varied from 0.919 to 0.990.Further,the multiple linear stepwise regression analysis showed that initial SOC content can explain 84.44%-97.93%of the mean SOCD,while clay content,pH and bulk density had a lesser and unstable effects to the average annual SOCD change of different mapping scales.