Estimation of Land-Surface-Atmosphere Carbon Dioxide Exchanges over Tropical Savanna Ecosystems.

Abstract

The African savanna ecosystems have been identified as major sources of inter-annual variability of the global atmospheric carbon dioxide (CO2). To quantify CO2 fluxes between the land surface and atmosphere of terrestrial ecosystems, micro-meteorological measurements based on the eddy covariance (EC) techniques are used. However, only a limited number of EC measurements have been performed for the variety of savanna ecosystems in West Africa. In an effort to ascertain the contributions of land use change to the carbon cycle over the West African Sudanian Savanna on different temporal scales (diurnal to the annual scale), three EC stations were established along transect of changing land use characteristics, near the border between Ghana and Burkina Faso. The ecosystems included: grassland (GR1), mixture of fallow and cropland (CR2), and a nature reserve (NR3). The Marginal Distribution Sampling (MDS) technique was used for gap-filling of the net ecosystem exchange of CO2 (Fc). The Fc was apportioned into its composite signals, gross primary production (Fg) and ecosystem respiration (Fe). An inter-comparison of the temporal variability of the CO2 fluxes (Fc, Fg and Fe) from the three EC sites was performed for the period between January and December 2013. Numerical simulations of observed momentum, water, energy and CO2 fluxes at the three different ecosystems, using a multi-layer atmosphere-SOil-VEGetation model (SOLVEG), were also carried out to clarify crucial processes in savanna ecosystems. The results over the study period showed that the ecosystems responded to physiological and environmental variables such as: soil moisture, Vapour Pressure Deficit (VPD), Leaf Area Index (LAI), albedo, Water Use Efficiency (WUE) and Photosynthetic Photon Flux Density (PPFD), by influencing CO2 exchange over the land. The ecosystems of the three sites, were found to be net sinks of CO2 in the rainy season (May to October). However, in the dry season (November to April), they became net sources of CO2 into the atmosphere, except, NR3 which served as a net sink of CO2 during the wet to dry transition period (November to December). On an annual timescale, only NR3 served as a net sink of CO2 from the atmosphere into the ecosystem. The annual Fc budgets were 127.8, 108.0 and -387.3 g C m-2 yr-1 for GR1, CR2 and NR3 respectively. The diurnal to annual estimates of CO2 fluxes from our findings compared well with outcomes of selected former flux measurements from across Africa, most especially with ecosystems that shared similar characteristics. Numerical simulations using SOLVEG for selected days of rainy and dry seasons revealed that the model reproduced well the diurnal changes of observed net radiation (Rn), sensible heat flux (H), latent heat flux (λE), friction velocity (u∗), soil surface temperature (Ts) and the net ecosystem exchange of CO2 (Fc). The simulation of Fc by the model was appreciably modulated by soil moisture, as well as soil and plant respiratory rates. Comparison of the model statistics for the three study sites, revealed an improvement in the model performance, increasing in the order of GR1, CR2 and NR3. The study revealed that the three contrasting ecosystems responded to environmental and physiological factors based on the ecosystem functional types and these affected CO2 exchanges processes over the land. This suggests that land use and management may play significant role in the diurnal to annual sequestration and efflux patterns of net ecosystem exchange of CO2 and its composite fluxes (Fg and Fe), over the West African Sudanian Savanna.