Abstract: By using modified carbon-water cycle model EPPML (ecosystem productivity process model for landscape), the carbon absorption and respiration in Qianyanzhou artificial masson pine forest ecosystem in 2003 and 2004 were simulated, and the sensitivity of the model parameters was analyzed. The results showed that EPPML could effectively simulate the carbon cycle process of this ecosystem. The simulated annual values and the seasonal variations of gross primary productivity (GPP), net ecosystem productivity (NEP), and ecosystem respiration (R_e) not only fitted well with the measured data, but also reflected the major impacts of extreme weather on carbon flows. The artificial masson pine forest ecosystem in Qianyanzhou was a strong carbon sink in both 2003 and 2004. Due to the coupling of high temperature and severe drought in the growth season in 2003, the carbon absorption in 2003 was lower than that in 2004. The annual NEP in 2003 and 2004 was 481.8 and 516.6 g C·m^-2·a^-1, respectively. The key climatic factors giving important impacts on the seasonal variations of carbon cycle were solar radiation during early growth season, drought during peak growth season, and precipitation during post-peak growth season. Autotrophic respiration (R_a) and net primary productivity (NPP) had the similar seasonal variations. Soil heterotrophic respiration (R_h) was mainly affected by soil temperature at yearly scale, and by soil water content at monthly scale. During wet growth season, the higher the soil water content, the lower the R_h was; during dry growth season, the higher the precipitation during the earlier two months, the higher the R_h was. The maximum RuBP carboxylation rate at 25 ℃ (V_m25), specific leaf area (SLA), maximum leaf nitrogen content (LN_m), average leaf nitrogen content (LN), and conversion coefficient of biomass to carbon (C/B) had the greatest influence on annual NEP. Different carbon cycle process could have different responses to sensitive parameters. For example, the increase of V_m25and LN could effectively promote carbon absorption and respiration, the decrease of LN/LN_m could decrease the carbon absorption and respiration, and, the increase of SLA and C/B could promote carbon absorption but inhibit soil respiration. However, the most sensitive parameters derived from annual carbon fluxes were not completely the same as those derived from growth season or non-growth season carbon fluxes.

Key words: carbon cycle, process-based model, abnormal weather, net ecosystem productivity, soil heterotrophic respiration, green roof, roof runoff, management mechanism, influence factor.