Poster, 3rd iLEAPS Science Conference, Garmisch-Partenkirchen, Germany: 18.09.2011 - 23.09.2011
Abstract:
Eddy-covariance technique is applied worldwide to acquire the information of carbon exchange between a variety of ecosystems and atmosphere, but the data acquisition only covers averagely 65% of the whole year due to system failures and data rejection. Therefore, data gaps must be filled to provide the annual budget of the net ecosystem exchange (NEE). However, gap-filling strategies are still under discussion within the research community. Presently the major gap-filling strategies include mean diurnal variation (MDV), look-up tables (LUT), and nonlinear regression (NLR). They work quite well for long-time running sites over slowly-developed biosphere surfaces such as evergreen forests, but for fast-developing ecosystems or fast-growing croplands like potato or rice in the subtropical South Korea difficulties appeared. In this study, we developed a multi-step filter procedure to gain good-quality data as input for the different parameterizations, and tested several gap-filling strategies based on NLR in daytime NEE obtained from the long-term campaign during the TERrain and ECOlogical Heterogeneity (TERRECO) program in 2010. We used the software package TK2 (Univ. of Bayreuth) for calculation, including eddy-covariance correction strategies as high frequency spike check, Planar-Fit-rotation, spectral corrections, conversion from sonic temperature fluctuations to real temperature fluctuations, density correction for water vapour, iterative determination, and quality control. Also, a Footprint analysis showed that the target potato-field contributed the majority of the measured fluxes except a slight part which was contributed by the adjacent cabbage-field only during stable conditions. The new multi-step filter procedure was than applied to the TK2-resutls (30min fluxes) including a consistency check, a multi-step spike-check (quantile and standard deviation filter), an instrument-error check, and the TK2 quality-flag check, which rejects between 21 and 36% of the whole data set. These proofed data were now used in several approaches based on light-response function invented by Michealis & Menten. The comparison between simulation and observation indicated that the - today most used - temperature classification does not improve the model performance, while the day-binning routine could obviously influence the simulation. A vapor-pressure deficit (VPD) effect seems to improve the simulation esp. during the morning hours. Adding a leaf area index (LAI) factor to capture both the daily course and seasonal vegetation development, we were now able to fill the large gaps between observation periods when other models cannot be used.