Use of remote sensing and bio-geochemical models to estimate the net carbon fluxes of managed mountain grasslands

• Published in: Ecological modelling Vol. 474; p. 110152
• Main Authors: Argenti, G., Chiesi, M., Fibbi, L., Maselli, F.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.12.2022
• Subjects: BIOME-BGC; C-Fix; GPP; MODIS; NDVI; NEP; Pasture; Pasture; GPP; NEP; BIOME-BGC; NDVI; C-Fix; MODIS
• ISSN: 0304-3800; 1872-7026
• DOI: 10.1016/j.ecolmodel.2022.110152

•Two models are combined to estimate the net carbon fluxes of managed pastures.•The first is an NDVI-driven parametric model, the second a biogeochemical model.•The simulation takes into account the effects of mowing and grazing.•The carbon flux estimates are assessed versus eddy covariance observations.•The approach is able to simulate the carbon balance of managed mountain pastures. The current paper presents a methodology which can simulate the net carbon fluxes of semi-natural grasslands based on the combination of a parametric and a bio-geochemical model. The first model (modified C-Fix) estimates grassland gross primary production (GPP) by driving the light use efficiency approach with remotely sensed normalized difference vegetation index (NDVI) data. The latter model (BIOME-BGC) simulates all main ecosystem processes based on meteorological and ancillary data. The outputs of the two models are finally combined and corrected for the effects of human-induced biomass changes (i.e. mowing, grazing, etc.). The modelling strategy, fed with ancillary and MODIS NDVI data, is tested in two semi-natural grasslands, the first (Amplero) on the Apennines and the second (Monte Bondone) on the Alps (Central and Northern Italy, respectively). The two sites, which show diversified eco-climatic conditions and are managed differently (i.e. Amplero by spring mowing plus cattle grazing and Monte Bondone by summer mowing), were selected for the availability of eddy covariance observations of GPP and net ecosystem production (NEP) collected during several years (from 2003 to 2012). These observations were used to assess the estimates of the modelling strategy, obtaining promising results for both study sites. The greatest estimation errors are found in the Alpine site and are mostly induced by the inaccurate detection of local grassland NDVI values, which is due to the insufficient spatial and temporal resolutions of the used MODIS NDVI imagery. The results of the study highlight the importance of information on the local management practices applied, which is decisive for quantifying medium-term changes in net carbon fluxes.