Dynamics of Carbon Dioxide Flux Over Africa: Insight from Observations and Model Simulations
No Thumbnail Available
Date
2020-12-29
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
The carbon cycle of tropical terrestrial vegetation plays a vital role in the storage
and exchange of atmospheric carbon dioxide (CO2). However, large uncertainties
surround the impacts of land-use change, climate warming, the frequency
of droughts, and CO2 fertilization. This culminates in poorly quantified carbon
stocks and carbon fluxes even for the major ecosystems in Africa’s carbon cycle
(savannas and tropical evergreen forests). Contributors to this uncertainty
are the sparsity of (micro-)meteorological observations across Africa’s vast land
area, a lack of sufficient ground-based observation networks and validation
data for CO2, and incomplete representation of important processes in numerical
models. Satellite retrievals are strongly influenced by land-use changes,
cloud cover, and aerosol loading. Moreover, Africa is a continent with wide extremes
in surface type (which ranges from desert, rainforest, and Savannah) and
aerosol loading. Therefore, the comparison of satellite observations with model
and available in-situ observations will be useful to prove the performance of
satellites and show how these systematic errors vary geographically over the
continent. In this thesis, GOSAT column-averaged dry-air mole fraction of carbon
dioxide (XCO2) is compared with the NOAA CT2016 and six flask observations
across Africa using five years of data covering the period from May 2009 to
April 2014. Besides, XCO2 from OCO-2 is compared with NOAA CT16NRT17
and eight flask observations across Africa using two years of data covering the
period from January 2015 to December 2016. The analysis shows that the XCO2
from GOSAT is higher than XCO2 simulated by CT2016 by 0.28 _ 1.05 ppm,
whereas OCO-2 XCO2 is lower than CT16NRT17 by 0.34 _ 0.9 ppm on African
landmass on average. The mean correlations of 0.83 _ 0.12 and 0.60 _ 0.41 and
an average RMSD of 2.30 _ 1.45 and 2.57 _ 0.89 ppm are found between the
model and the respective datasets from GOSAT and OCO-2 implying the existence
of a reasonably good agreement between CT and the two satellites over
Africa’s land region. However, significant variations were observed in some regions.
For example, OCO-2 XCO2 is lower than that of CT16NRT17 by up to 3
ppm over some regions in North Africa (e.g., Egypt, Libya, and Mali ) whereas
it exceeds CT16NRT17 XCO2 by 2 ppm over Equatorial Africa (10_S - 10_N).
This regional difference is also noted in the comparison of model simulations
and satellite observations with flask observations over the continent. For example,
CT shows a better sensitivity in capturing flask observations over sites
located in Northern Africa. In contrast, satellite observations have better sensitivity
in capturing flask observations in lower altitude island sites. CT2016
shows a high spatial mean of seasonal mean RMSD of 1.91 ppm during DJF
from GOSAT, while CT16NRT17 shows RMSD of 1.75 ppm during MAM from
OCO-2. On the other hand, the low RMSD of 1.00 and 1.07 ppm during SON in
model XCO2 from GOSAT and OCO-2, respectively, indicate better agreement
during autumn. The model simulation and satellite observations exhibit similar
seasonal cycles of XCO2 with a small discrepancy over Southern Africa (35_ -
10_S) and during wet seasons over all regions. Two remotely sensed vegetation
products that have been shown to correlate highly with Gross Primary Productivity
(GPP): Sun-Induced Fluorescence (SIF) and Near-Infrared Reflectance
of vegetation (NIRv) are also analyzed to further understand the dynamics of
carbon dioxide flux. A comparison against flux tower observations of daytimepartitioned
Net Ecosystem Exchange (NEE) from six major biomes in Africa
shows that SIF and NIRv reproduce the seasonal patterns of GPP well, resulting
in correlation coefficients of >0.9 (N=12 months, 4 sites) over savannas in the
northern and southern hemisphere. These coefficients are slightly higher than
for the widely used MPI-BGC GPP products and Enhanced Vegetation Index
(EVI). Similar to SIF signals in the neighbouring Amazon, peak productivity occurs
in the wet season coinciding with peak soil moisture, and is followed by an
initial decline during the early dry season that reverses when light availability
peaks. This suggests similar leaf dynamics are at play. Spatially, SIF and NIRv
show a strong linear relation (R >0.9, N=250+ pixels) with multiyear MPI-BGC
GPP even within single biomes. Both MPI-BGC GPP and EVI show saturation
relative to peak NIRv and SIF signals during high productivity months, which
suggests that GPP in the most productive regions of Africa might be larger than
suggested. Africa’s biome integrated productivity is strongly controlled by the
seasonality of soil moisture, with a weak influence of light availability superimposed,
indicating that the biome productivity of Africa strongly depends
on spatiotemporal drivers. Therefore, an understanding of the spatiotemporal
ecosystem dynamics together with its relation to meteorological variables is
paramount to quantify the responsiveness of the carbon cycle to climate variability.
For that reason, an Empirical Ensemble Mode Decomposition (EEMD)
was applied on 17 years monthly time series of natural CO2 flux covering the
period from January 2000 to December 2016. The EEMD depicts natural CO2
flux has six periodicities over tropical Africa corresponding to seasonal, interannual,
and decadal-scale variabilities which are likely driven by atmospheric and
oceanic processes. Seasonal variabilities at quasi-3 months, quasi-6 months, and
quasi-12 months contribute about 91.41% of the variability of natural CO2 flux,
suggesting that CO2 flux has a strong variability at the seasonal scale. Moreover,
high atmospheric CO2 flux was observed during warm and dry conditions. Precipitation
is found to be a dominating driver of CO2 flux at the seasonal scale
over the west coast of tropical Africa and East Africa. In addition to the six
periodicities, the application of EEMD to a monthly time series of CO2 flux indicates
the existence of either a nonlinear downward trend or a possible multidecadal
periodicity that cannot be captured by the limited length of the current
data set. The later is more likely as revealed by a slight reversal at the beginning
of 2013. Moreover, analysis of different regions of tropical Africa shows
reduced CO2 uptake over most regions since 2000, with exception for tropical
North Africa which is found to have increased CO2 uptake most likely due to
enhanced vegetation which exceeds deforestation. At the interannual scale, a
quasi-2 year and quasi-5 year fluctuations were obtained from the EEMD with
a contribution of 6.93% to the total CO2 flux variability. This interannual fluctuation
has a significant correlation with Niño 3.4 index, El Niño induced temperature,
precipitation, soil moisture, and enhanced vegetation index. A significant
positive correlation between a warming temperature and interannual CO2
flux over tropical North Africa and rainforest regions suggests that temperature
is the major driver of CO2 fluctuation at the interannual scale over these
regions. Conversely, over Western and Tropical East Africa, precipitation was
found as the most dominant driver. The anomalously high interannual CO2
flux was found in response to strong El Niño (Niño 3.4 index greater than 1.0)
in the years 2009 and 2015/16 over most of Equatorial Africa. During the peak
of 2015/16 El Niño, tropical Africa releases 0.2 mol/m2/month CO2 into the
atmosphere due to interannual variability. The strongest (0.5 mol/m2/month)
contribution was from the tropical rainforest, most likely driven by the rising
temperature. Besides, Ethiopian highlands also release 0.4 mol/m2/month CO2
flux due to dry and warm conditions during this strong El Niño event. The
CO2 flux mean over 17 years (2000-2016) shows that tropical Africa is a net CO2
sink (-7.02 gC/m2/year). However, during the 2015/16 El Niño years, tropical
Africa releases 29.12 gC/m2/year leading to 487.49 TgC/year which is twice
the estimated carbon flux of Africa (240 Tg C y1 ) for the period covering from
2000 to 2005.
Description
Keywords
Dynamics, Carbon Dioxide, Over Africa, Insight, Observations, Model, Simulations