Analysis of Arctic peak-season carbon flux estimations based on four MODIS vegetation products

Sammanfattning: Increased temperatures in high latitudes may alter the carbon dynamics throughout the Arctic. Modelled CO2 simulations show that current climate conditions constitute the Arctic a net carbon sink, though the large extent and fine landscape heterogeneity raise an uncertainty about the carbon sink/source status of the region. The understanding of Arctic CO2 fluxes can be improved through integration of remote sensing techniques and environmental modelling. In this study, vegetation indices, i.e. LAI and NDVI from four MODIS products are used in the Pan-Arctic Net Ecosystem Exchange (PANEEx) model to calculate NEE at 12 Arctic study sites. The main objective was to determine the impact of the vegetation indices at 250 m, 500 m and 1 km resolution on the precision of NEE estimations. Data from eddy covariance towers (EC) were used to identify similarities and discrepancies between modelled and in situ LAI and NEE scores in July 2008-2010. Google Earth Engine (Google Inc.), a powerful geospatial platform, was implemented for data acquisition and quantitative analysis. Linear correlations on 1:1 scatter plots and inferential statistics were used to assess the relationships between the modelled and in situ estimations. The model run using the 250 m MOD13Q1 LAI product simulated 78% of the measured NEE fluxes (R2 = 0.73, p < 0.001) throughout the study sites. Overall, utilization of the PANEEx model with 250 m MODIS products indicates a potential for future modelling in the Arctic. Data analysis generated considerable differences in modelled NEE outputs and hence, their application in environmental modelling needs to be considered. The model simulations also demonstrate the potential of employing vegetation indices on much finer scale, i.e. 10-30 m in order to capture the Arctic heterogeneous landscape.