Abstract ID: 3.13052 | Accepted as Talk | Talk/Oral | TBA | TBA

Title: Integrating Sentinel NDVI data into eddy covariance postprocessing Authors: Torben Callesen, Anna Candotti, Leonardo Montagnani Mountain eddy covariance (EC) sites are often characterised by a high degree of heterogeneity in land cover. Combined with other factors like diel variation in wind direction and atmospheric decoupling due to slope wind circulation, this can cause measurement bias in the integrated flux signal. While current footprint partitioning tools focus on interpretation of processed data, the increasing availability of high-resolution remote sensing products may facilitate a more automated and fundamental inclusion of site surface information in the data processing pipeline. This study combines a 10 m resolution NDVI site map derived from Sentinel-2 Level 2-A bands with footprint projections calculated using the open-source tool provided by Kljun et al. (2015) to obtain weighted surface characteristics for each half-hourly averaging period. Data comprising three months of EC measurements from three different sites with complex topographies and heterogenous land cover was used for analysis. Postprocessing was performed using the REddyProc R package (Wutzler et al., 2018) following otherwise common flux network procedure (Pastorello et al., 2020). We demonstrate that remote sensing information such as NDVI may be integrated into data quality flagging, gap-filling and flux partitioning with marked impacts on these processes due to the improved ability to explain variation in net ecosystem exchange (NEE). Although, as with many filtering approaches, segregation by footprint-weighted NDVI may suffer from bias due to concurrent wind and meteorological conditions, appropriate grouping has proven sufficient to allow existing algorithms enough data to function reliably in each of the three very different site cases. We therefore recommend that, with future refinement, this and similar remote sensing products be integrated into standard EC postprocessing methodology.