Compilation of a global N₂O emission inventory for tropical rainforest soils using a detailed biogeochemical model

摘要: NITROUS OXIDE (N 2 O) is a potent trace gas contributing to approximately 6% the observed anthropogenic global warming. Soils have been identified be major source of atmospheric N O and tropical rainforest soils are thought account for largest part. Furthermore, various studies shown that magnitude emissions from soil highly variable on spatial temporal scales. Detailed, process-based models coupled Geographic Information Systems (GIS) considered promising tools calculation emission inventories. This methodology explicitly accounts governing microbial processes as well environmental controls. Moreover, mechanistic biogeochemical operating in daily time-steps (e.g. ForestDNDC-tropica) capture intra- inter-annual variations emissions. However, detailed datasets required model calibration testing, but currently few numbers. In this study an automated measurement system was used derive O, methane (CH 4 ) carbon dioxide (CO soil-atmosphere exchange important parameters Kenya Southwest China. Distinct differences were C at investigated sites forest types. common features such pulse after dry season or pronounced moisture dependency both sites. The derived unique these regions so far no information about strength available and, first time, CH CO recorded sub-daily resolution. utilized conjunction with high-resolution Australian rainforests re-calibration ForestDNDC-tropica using multi-site, parallel Bayesian approach. Extensive validation sensitivity underlined good agreement improved fluxes. Based newly developed GIS database worldwide, new then inventory. Daily years 1991 - 2000 calculated. results show striking strength. calculations estimate revised previously 1.2 3.6 Tg yr -1 (based wide range areas considered) 1.3 . As accuracy output dependant data quality driving models, uncertainty assessment performed quantify data-induced presented Using Latin hypercube sampling approach, calculated 0.9 2.4 Another key finding strong seasonal variability originating scale. So far, not captured by statistical upscaling approaches.