Effect Of Canopy Layers On The Accuracy Of Forest Ecosystem Carbon Flux

One of the research hotpots under global warming is that the utilization of multi-layer model to estimate the carbon flux of forest ecosystem, to determine the best canopy layered number can improve the simulation accuracy. Based on meteorological gradient observation data during 2003～2008, the net ecosystem production(NEP) was simulated using multi-layer model under different canopy layered numbers of mixed broadleaf-conifer forest of Changbai Mountain. By comparing the modeled NEP with those measured by eddy covariance system, the main results obtained are as follows:1. There was good reappearance ability of diurnal NEP variation under different canopy layered numbers by multi-layer model, NEP is positive during the daytime and below zero at night. The model can simulate the "unimodal" diurnal variation of NEP of the temperate mixed broadleaf-conifer forest ecosystems in Changbai Mountain.2. With the increasing of canopy layered numbers, The ability of NEP interpretation was increased quickly and slightly or remained stable secondly by this model. When the layered numbers were 2,5,10,15, the simulated value could be interpreted separately of 57.44%, 86.49%,85.93% and 85.49% of the NEP variation (p<0.01) from August 1st to 5th in 2004. In July and August, the model had a tendency to underestimate the NEP when the layered number less than 4, and the fewer layered numbers, the more serious underestimate tend. But there was a tendency to overestimate when layered number was more than 6, and the more layered numbers, the more serious overestimate tend.3. The least root mean square error (RMSE) was used to evaluate the different effects of layered numbers in canopy NEP simulation accuracy and to determine the best layered number. In July and August in 2003～2008, the best layered numbers were 4～7 when NEP was estimated by multi-layer model at mixed broadleaf-conifer forest ecosystem of Changbai Mountain. But when the layered number was less than 4～7, the NEP was underestimated, and when the layered number was more, NEP was overestimated.4. There were significant differences in the best layered number of multi-layer model exist in different seasons. In the Early and late periods of the growing season, the best layered numbers were about 2～3, and 6～8 in the vigorous growing season. The seasonal variation of the best layered number was the same to leaf area index. There was a significant positive exponential relationship between them(R2=0.9091, p<0.01). LAI was one of the main limiting factors in the best layered numbers.