【作者】 张黎明；

【导师】 潘剑君； 史学正；

【作者基本信息】 南京农业大学 ， 土壤学， 2009， 博士

【摘要】 土壤有机碳演变和全球气候变化息息相关,利用动态模型来模拟和预测土壤有机碳的演变现已成为研究的热点。但是,以前的研究多局限于中、小比例尺,对于大、中、小尺度系列制图尺度数据库的工作少有报道,而且对于使用不同系列制图尺度数据库对土壤有机碳模型模拟精度的影响还不清楚。在模型验证方面,也往往根据田间长期定位观测结果对模型进行验证,但从区域尺度上并没有对模型进行验证,这样的模拟或预测结果存在较大的不确定性。因此,本研究以模拟生物地球化学过程较为成熟的DNDC (DeNitrification and DeComposition)模型为例,选择基本上为水稻土的太湖流域作为研究区,通过模拟该地区230多万hm2水稻土在1：1400万、1：400万、1：100万、1：50万、1：20万和1：5万6种制图比例尺下1982～2000年19年间的土壤有机碳演变,并将不同制图尺度下的模拟结果与2000年该地区1000多个采样点实测值进行比较,尝试从区域角度验证并评价模型适宜性,以便为进一步修正模型和评估不同制图尺度下模拟精度提供理论依据。本文的主要研究结论如下：1)从不同制图尺度下的区域验证来看,太湖地区水稻土2000年土壤有机碳模拟值和实测值在1：20万、1：50万和1：400万3个制图尺度下相对误差都≤±5%,达到了模拟结果很准确的水平,其中在1：20万制图尺度下的相对误差最小,只有0.28%,模拟精度最高；1：5万和1：100万制图尺度下模拟值和实测值的相对误差分别为6.4%和5.1%,≤±10%,也达到了模拟结果可行的标准,而1：1400万的模拟值和实测值的相对误差为20.0%,说明模拟结果不可靠。从目前的大多数研究来看,我国的DNDC模型国家尺度模拟中土壤属性数据大多使用1：1400万土壤图和《中国土种志》资料,这有可能造成有机碳模拟的较大误差。2)太湖地区不同水稻土亚类6个制图尺度下的模型适宜性也有很大差异。漂洗型水稻土在1：20万制图尺度下的模拟精度最高,潴育型水稻土是太湖地区分布面积最大的亚类,1：400万是该亚类比较理想的模拟尺度。渗育型水稻土一般占到总水稻土面积的16%以上,该亚类在1：50万制图尺度下的模拟精度最高。潜育型水稻土的土壤有机碳含量是所有亚类中最高的,相对于其他亚类,DNDC模型对潜育型水稻土的有机碳模拟效果在各个制图尺度下都比较差,相对误差都超过10%,但相对而言,在1：50万制图尺度下的模拟精度最高。脱潜型水稻土分布面积一般占到总水稻土面积的18%以上,而且该亚类的土壤有机碳含量仅次于潜育型水稻土,在1：50万制图尺度下该亚类的模拟精度最高。淹育型水稻土是太湖地区分布面积最小的亚类,该亚类在1：5万制图尺度下模拟效果最好。3)不同制图尺度的数字化土壤图对碳储量模拟估算的影响也不同。从本研究来看,随着制图比例尺的减小,DNDC模型模拟的太湖地区水稻土2000年表层(0～30cm)有机碳总储量基本呈增加趋势,尤其在1：1400万制图尺度下的土壤有机碳总储量明显高于其他尺度。太湖地区土壤有机碳总储量主要受潴育型水稻土、潜育型水稻土、脱潜型水稻土和渗育型水稻土控制,这4个水稻土亚类的总储量占不同制图尺度下总储量的93%以上。但不同制图尺度影响最大的是潜育型水稻土和潴育型水稻土,尤其在1：1400万制图尺度下这2个水稻土亚类的有机碳总量明显高于其它尺度。4)不同制图尺度对太湖地区水稻土有机碳的年变化(dSOC)模拟也有很大影响,土壤属性数据最详细的1：5万制图尺度在1982～2000年19年来水稻土表层(0～30cm)有机碳总体呈升高趋势的面积达147万hm2,占总水稻土面积的63.4%,19年来固定土壤有机碳1.48 Tg；但在1：20万、1：50万、1：100万、1：400万和1：1400万5种制图比例尺下DNDC模型模拟的1982～2000年19年表层(0～30cm)水稻土分别亏损有机碳：0.78 Tg、2.86 Tg、2.33 Tg、0.44 Tg和7.86 Tg,说明由于不同类型土壤的有机碳属性特征不同,土壤制图比例尺的变化,使得区域内土壤总面积和各类型土壤的面积比例发生显著改变,从而导致土壤制图比例尺对有机碳模拟结果产生显著影响。5)不同管理措施和气候因子下太湖地区水稻土有机碳的情景分析表明,加大作物生物量还田、免耕或采取秸秆还田为基础的保护性耕作措施将有效的增加土壤有机碳含量,适度烤田和施用化肥也有利于土壤有机碳的积累。气候因子对土壤有机碳的影响较为复杂,总趋势是土壤有机碳随着温度的升高分解速度在加快,说明未来气候变暖必定会造成大量土壤有机碳的损失,这也与RothC模型模拟结果相一致。6)不同土壤数据单元对生物地球化学模型DNDC的土壤有机碳模拟精度有很大影响。目前国内应用最广泛的1：1400万“县级”单元法估算的太湖地区水稻土有机碳年变化(dSOC)与土壤属性最为详尽的1：5万“图斑”单元法模拟值在整个地区总量和“县级”单元水平上相差都很大,大多数“县级”单元之间的dSOC相对偏差高于300%；而1：5万“县级”单元法的模拟值与1：5万“图斑”单元法估算值之间dSOC差异相对较小,并且1：5万“图斑”单元法模拟的“县级”单元dSOC和整个地区dSOC总量基本都在1：5万“县级”单元法最大与最小值范围之间,这一方面验证了DNDC模型以“县”作为最小模拟单元,并用模拟值范围来表达区域dSOC方法的合理性,另一方面也说明了详细的土壤数据单元是保证地球生物化学过程模型模拟精度的重要因子。因此,在今后的国家和区域尺度有机碳模拟中使用更详细的土壤资料是非常必要的。更多还原

【Abstract】 The dynamic of soil organic carbon (SOC) has a close relationship with global climate change. Simulating and forecasting the dynamic of SOC by model has now become a popular method. Previous studies were primarily limited to medium and small mapping scale, while studies on the series mapping scale databases of large, medium and small scale were rarely reported. Furthermore, the influence of usage of different series mapping scale databases on the simulation precision of SOC model was not clear. In terms of modeling verification, it was usually carried out based on the results of long-term observation at field rather than regional scale, thus the simulation and forecast results had a large uncertainty.In this paper, taking the DNDC (DeNitrification and DeComposition) model which is comparatively mature at modeling biogeochemical process for example, selecting the Taihu Lake region whose soil type is almost paddy soil as the study area.The dynamic of SOC on more than 2.3 M.hm-2 paddy soil between 1982 and 2000 under six kinds of mapping scales i.e.1:14,000,000,1:4,000,000,1:1,000,000,1:500,000,1:200,000 and 1:50,000 was simulated and the results under different mapping scales were compared with the measured values of more than 1000 sampling points at year 2000.The model suitability was verified and assessed from the entire region, in order to provide a theoretical basis for further amendment of the model and assessment of simulation accuracy under different mapping scales. The main research conclusions were as follows.1) From the regional verification under different mapping scales, the relative errors between the simulated and the measured values of paddy soil SOC in Taihu Lake region in the year 2000 under mapping scale of 1:200,000,1:500,000 and 1:4,000,000 were all less than±5%, which indicated that the simulation results reached accurate level. The mapping scale of 1:200,000 which had the minimum relative error of 0.28% had the highest simulation accuracy. The relative errors between the simulated and the measured values under mapping scale of 1:50,000 and 1:1,000,000 were 6.4% and 5.1% respectively, which indicated that the simulation results were feasible. The relative error between the simulated and the measured values under mapping scale of 1:14,000,000 was 20.0%, which indicated that the simulation result was unreliable. Viewed from most studies at the present time, soil attribute data used in the simulation of DNDC model at national scale were predominantly from 1:14,000,000 soil map and the Second National Soil Survey, which would bring about comparatively big error of SOC simulation.2) The model suitability of paddy soil subgroup in Taihu Lake region under six mapping scales had great differences. The bleaching paddy soil had the highest simulation accuracy under 1:200,000 mapping scale. The hydromorphic paddy soil had the largest distribution area in Taihu Lake region and 1:4,000,000 was its ideal simulation scale. The percolated paddy soil whose area accounted for 16% of the whole paddy soil area had the highest simulation accuracy under 1:500,000 mapping scale. The gleyed paddy soil had maximum SOC content and its SOC simulation results, whose relatively errors were all above 10%, were the worst under each mapping scale compared with other subgroup. Relatively speaking, its simulation accuracy under the mapping scale of 1:500,000 was the best. The degleyed paddy soil whose area accounted for 18% of the whole paddy soil area and whose SOC content was outranked only by the gelyed paddy soil had the highest simulation accuracy under 1:500,000 mapping scale. The submergic paddy soil had the least distribution area in Taihu Lake region and its simulation result was the best under the 1:50000 mapping scale.3) Digital soil map of different mapping scale had different influence on the simulated estimation of SOC storage. In this study, with the decrease of mapping scale, the total SOC storage of surface paddy soil (0-30cm) in Taihu Lake region in the year 2000 simulated by the DNDC Model shown a trend of increase basically, especially the total SOC storage of 1:4,000,000 mapping scale was apparently higher than any other scale. Overall, the total SOC storage of the Taihu Lake region was mainly controlled by the hydromorphic paddy soil, the gleyed paddy soil, the degleyed paddy soil and the percolated paddy soil, for the SOC storage of these four paddy soil subtypes accounted for more than 93% of the total SOC storage under each mapping scale. The gleyed paddy soil and the hydromorphic paddy soil had the greatest influence on different mapping scale and their SOC storage under the 1:14,000,000 mapping scale was apparently higher than any other scale.4) Different mapping scale had significant impact on dSOC in Taihu Lake region. The area with increased SOC of surface paddy soil (0-30cm) from 1982 to 2000 of the 1:50,000 mapping scale with the most detailed soil attribute data was 1.47 M.hm-2, up to 63.4% of the whole paddy soil area, and 1.48 Tg C were sequestered in the soil over the past 19 years. However,0.78 Tg C、2.86 Tg C、2.33 Tg C、0.44 Tg C and 7.86 Tg C were lost respectively in the surface paddy soil (0-30cm) from 1982 to 2000 at the mapping scale of 1:200,000, 1:500,000,1:1,000,000,1:4,000,000 and 1:14,000,000. This indicated that mapping scale had a notable impact on SOC simulation result because different soil type had different SOC attribute and the ratio between total soil area and area of each soil type changed significantly with variation of soil mapping scale.5) Scenario analysis of paddy soil SOC in Taihu Lake region with different management practice and different climate factor indicated that more crop biomass return in the filed, no-tillage and basic conservation tillage practice such as straw return could increase SOC content effectively and reasonable midseason drainage and fertilizer application would benefit the accumulation of SOC. Climate factors had a quite complex influence on SOC, but the general trend was that the decomposition velocity of SOC speeded up along with the rise in temperature. The result showed that global warming in the future would be bound to bring about loss of SOC. This was in line with the result simulated by the RothC model.6) Different soil data unit had remarkable impact on SOC simulation accuracy of the DNDC model. The dSOC estimated by county-based database under 1:14,000,000 scale differed widely from that estimated by 1:50,000 soil map-based databases at the level of both entire region and county. The relative deviation of dSOC between most county-based database was more than 300%. The difference between dSOC estimated by county-based database and estimated by 1:50,000 soil map-based database was relatively minor. The dSOC and its total change in the past 19 years estimated by 1:50,000 soil map-based databases varied between the maximum and the minimum value simulated by county-based database under 1:50,000 scale. The result indicated that it was reasonable for the DNDC model taking a county of the smallest simulation unit and expressing regional dSOC by the scope of simulation value on one hand and detailed soil data unit was the guarantee of simulation accuracy of the biogeochemical process model on the other hand. Therefore, it would be necessary to use more detailed soil data for SOC simulation of national and regional scale in the future studies.更多还原