Effects of satellite spatial resolution on gross primary productivity estimation through light use efficiency modeling

Abstract

Terrestrial Gross Primary Productivity (GPP) describes the total amount of CO2 assimilated by plants in an ecosystem during photosynthesis and is considered the largest flux component of the global carbon cycle. One of the most prominent techniques for estimating GPP at ecosystem scale is the Light Use Efficiency (LUE) approach, taking advantage of the spatiotemporal capabilities that satellite data provide. LUE expresses GPP as the product of absorbed photosynthetically active radiation (APAR) and the efficiency (Îμ ) that APAR is converted to biomass. Although satellite imagery is the key component of such models, the effects of image spatial resolution on model performance have not been thoroughly investigated. The emergence of new satellite instruments with enhanced spatial, spectral and temporal capabilities (i.e. Copernicus Sentinels) provides the opportunity for GPP estimation in high spatial resolution and comparison with low resolution GPP products (i.e. MODIS). In this study, a LUE model is applied to three satellite instruments with different spatial resolution: MODIS (500 m), Sentinel-3 (300 m) and Sentinel-2 (10 m). The GPP estimates of the three instruments are compared over six forest sites in Greece: Two deciduous (Quercus sp., Fagus sylvatica), two coniferous (Pinus nigra, Pinus halepensis) and two mixed (Pinus nigra with Fagus sylvatica). The results demonstrate that spatial resolution is not a crucial parameter for LUE modeling in wide, homogenous and fully covered forested areas. The spatial resolution is more important when applying LUE in mixed canopies or partially covered forested areas due to the effects of the different land cover types. To that purpose, Sentinel-2 presents a unique potential for accurate characterization of the land cover type and dynamics, due to the increased spatial resolution and frequent coverage, appearing as a prominent tool for future large scale GPP monitoring. © 2018 SPIE.