Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

First assessment of the plant phenology index (PPI) for estimating gross primary productivity in African semi-arid ecosystems

Abdi, A. M. LU orcid ; Boke-Olén, N. LU ; Jin, H. LU ; Eklundh, L. LU orcid ; Tagesson, T. LU ; Lehsten, V. LU and Ardö, J. LU orcid (2019) In International Journal of Applied Earth Observation and Geoinformation 78. p.249-260
Abstract


The importance of semi-arid ecosystems in the global carbon cycle as sinks for CO
2
emissions has recently been highlighted. Africa is a carbon sink and nearly half its area comprises arid and semi-arid ecosystems. However, there are uncertainties regarding CO
2
fluxes for semi-arid ecosystems in Africa, particularly savannas and dry tropical woodlands. In order to improve on existing remote-sensing based methods... (More)


The importance of semi-arid ecosystems in the global carbon cycle as sinks for CO
2
emissions has recently been highlighted. Africa is a carbon sink and nearly half its area comprises arid and semi-arid ecosystems. However, there are uncertainties regarding CO
2
fluxes for semi-arid ecosystems in Africa, particularly savannas and dry tropical woodlands. In order to improve on existing remote-sensing based methods for estimating carbon uptake across semi-arid Africa we applied and tested the recently developed plant phenology index (PPI). We developed a PPI-based model estimating gross primary productivity (GPP) that accounts for canopy water stress, and compared it against three other Earth observation-based GPP models: the temperature and greenness (T-G) model, the greenness and radiation (GöR) model and a light use efficiency model (MOD17). The models were evaluated against in situ data from four semi-arid sites in Africa with varying tree canopy cover (3–65%). Evaluation results from the four GPP models showed reasonable agreement with in situ GPP measured from eddy covariance flux towers (EC GPP) based on coefficient of variation (R
2
), root-mean-square error (RMSE), and Bayesian information criterion (BIC). The GöR model produced R
2
= 0.73, RMSE = 1.45 g C m
−2
d
−1
, and BIC = 678; the T-G model produced R
2
= 0.68, RMSE = 1.57 g C m
−2
d
−1
, and BIC = 707; the MOD17 model produced R
2
= 0.49, RMSE = 1.98 g C m
−2
d
−1
, and BIC = 800. The PPI-based GPP model was able to capture the magnitude of EC GPP better than the other tested models (R
2
= 0.77, RMSE = 1.32 g C m
−2
d
−1
, and BIC = 631). These results show that a PPI-based GPP model is a promising tool for the estimation of GPP in the semi-arid ecosystems of Africa.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; and
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Drylands, Eddy covariance, Remote sensing, Earth observation, Gross primary productivity, Land surface temperature, Africa, Plant phenology index
in
International Journal of Applied Earth Observation and Geoinformation
volume
78
pages
12 pages
publisher
Elsevier
external identifiers
  • scopus:85062888348
ISSN
1569-8432
DOI
10.1016/j.jag.2019.01.018
project
Global Savannah Phenology: Integrating Earth Observation, Ecosystem Modeling, and PhenoCams
language
English
LU publication?
yes
id
29b684b8-7638-4bb6-b4c2-9bfc81300c40
date added to LUP
2019-03-21 08:44:48
date last changed
2023-01-03 22:00:09
@article{29b684b8-7638-4bb6-b4c2-9bfc81300c40,
  abstract     = {{<p><br>
                                                         The importance of semi-arid ecosystems in the global carbon cycle as sinks for CO                             <br>
                            <sub>2</sub><br>
                                                          emissions has recently been highlighted. Africa is a carbon sink and nearly half its area comprises arid and semi-arid ecosystems. However, there are uncertainties regarding CO                             <br>
                            <sub>2</sub><br>
                                                          fluxes for semi-arid ecosystems in Africa, particularly savannas and dry tropical woodlands. In order to improve on existing remote-sensing based methods for estimating carbon uptake across semi-arid Africa we applied and tested the recently developed plant phenology index (PPI). We developed a PPI-based model estimating gross primary productivity (GPP) that accounts for canopy water stress, and compared it against three other Earth observation-based GPP models: the temperature and greenness (T-G) model, the greenness and radiation (GöR) model and a light use efficiency model (MOD17). The models were evaluated against in situ data from four semi-arid sites in Africa with varying tree canopy cover (3–65%). Evaluation results from the four GPP models showed reasonable agreement with in situ GPP measured from eddy covariance flux towers (EC GPP) based on coefficient of variation (R                             <br>
                            <sup>2</sup><br>
                                                         ), root-mean-square error (RMSE), and Bayesian information criterion (BIC). The GöR model produced R                             <br>
                            <sup>2</sup><br>
                                                          = 0.73, RMSE = 1.45 g C m                             <br>
                            <sup>−2</sup><br>
                                                          d                             <br>
                            <sup>−1</sup><br>
                                                         , and BIC = 678; the T-G model produced R                             <br>
                            <sup>2</sup><br>
                                                          = 0.68, RMSE = 1.57 g C m                             <br>
                            <sup>−2</sup><br>
                                                          d                             <br>
                            <sup>−1</sup><br>
                                                         , and BIC = 707; the MOD17 model produced R                             <br>
                            <sup>2</sup><br>
                                                          = 0.49, RMSE = 1.98 g C m                             <br>
                            <sup>−2</sup><br>
                                                          d                             <br>
                            <sup>−1</sup><br>
                                                         , and BIC = 800. The PPI-based GPP model was able to capture the magnitude of EC GPP better than the other tested models (R                             <br>
                            <sup>2</sup><br>
                                                          = 0.77, RMSE = 1.32 g C m                             <br>
                            <sup>−2</sup><br>
                                                          d                             <br>
                            <sup>−1</sup><br>
                                                         , and BIC = 631). These results show that a PPI-based GPP model is a promising tool for the estimation of GPP in the semi-arid ecosystems of Africa.                         <br>
                        </p>}},
  author       = {{Abdi, A. M. and Boke-Olén, N. and Jin, H. and Eklundh, L. and Tagesson, T. and Lehsten, V. and Ardö, J.}},
  issn         = {{1569-8432}},
  keywords     = {{Drylands; Eddy covariance; Remote sensing; Earth observation; Gross primary productivity; Land surface temperature; Africa; Plant phenology index}},
  language     = {{eng}},
  pages        = {{249--260}},
  publisher    = {{Elsevier}},
  series       = {{International Journal of Applied Earth Observation and Geoinformation}},
  title        = {{First assessment of the plant phenology index (PPI) for estimating gross primary productivity in African semi-arid ecosystems}},
  url          = {{http://dx.doi.org/10.1016/j.jag.2019.01.018}},
  doi          = {{10.1016/j.jag.2019.01.018}},
  volume       = {{78}},
  year         = {{2019}},
}