Advanced

Constraining a terrestrial biosphere model with remotely sensed atmospheric carbon dioxide

Kaminski, T.; Scholze, M. LU ; Vossbeck, M.; Knorr, W. LU ; Buchwitz, M. and Reuter, M. (2017) In Remote Sensing of Environment 203. p.109-124
Abstract

We present two novel earth observation products derived from the BESD and EMMA XCO2 products which were respectively retrieved from SCIAMACHY and GOSAT observations within the GreenHouse Gas project of ESA's Climate Change Initiative (GHG-CCI). These products are inferred by a Carbon Cycle Data Assimilation System (CCDAS) and consist of net and gross biosphere-atmosphere fluxes of carbon dioxide on a global 0.5° grid. As a further dataset provided by the CCI, the burnt area product developed by its Fire忌i project was used in the CCDAS to prescribe the emission component from biomass burning. The new flux products are provided with per-pixel uncertainty ranges. Fluxes with uncertainty ranges can also be provided aggregated in... (More)

We present two novel earth observation products derived from the BESD and EMMA XCO2 products which were respectively retrieved from SCIAMACHY and GOSAT observations within the GreenHouse Gas project of ESA's Climate Change Initiative (GHG-CCI). These products are inferred by a Carbon Cycle Data Assimilation System (CCDAS) and consist of net and gross biosphere-atmosphere fluxes of carbon dioxide on a global 0.5° grid. As a further dataset provided by the CCI, the burnt area product developed by its Fire忌i project was used in the CCDAS to prescribe the emission component from biomass burning. The new flux products are provided with per-pixel uncertainty ranges. Fluxes with uncertainty ranges can also be provided aggregated in space and time, e.g. over given regions or as annual means. For both, posterior flux fields inferred from BESD and EMMA products, transport model simulations show reasonable agreement with the atmospheric carbon dioxide concentration observed at flask sampling stations. This means that the information provided by the terrestrial and transport models, the respective GHG ECV product, the burnt area ECV product, a product of the Fraction of Absorbed Photosynthetically Active Radiation used to drive the model, and the atmospheric flask samples is largely consistent. The most prominent feature in the posterior net flux is the tropical source of CO2 inferred from both products. But for the EMMA product this release, especially over South America, is with 300 gC/m2/year much more pronounced than for BESD. This confirms findings by a recent intercomparison of transport inversions using GOSAT data by Houweling et al. (2015). The reason for the larger net flux is increased heterotrophic respiration. For both products the posterior 2010 sink over Europe (without Russia) is in the range of a recent compilation of European flux estimates by Reuter et al. (2016b). The posterior 2010 uptake of Australia (including Oceania) inferred from the EMMA product is 1.3 ± 0.2 PgC/year and appears to confirm the high sink also derived from GOSAT by Detmers et al. (2015) over a slightly different period and area. While for some regions (USA, Canada, Europe, Russia, Asia) the one standard deviation uncertainty ranges derived from BESD and EMMA do overlap, for some other regions (Brazil, Africa, Australia) this is not the case. It is not clear yet whether this is due to the uncertainty specifications in the respective products or the handling of uncertainty in the assimilation chain. Assumptions on correlation of observational uncertainty in space and time have a considerable impact on the inferred flux fields (≈ 60 gC/m2/year). The effect of adding an uncertainty that approximates the error in the retrieval system is of similar size.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Carbon fluxes, Column integrated carbon dioxide, GOSAT, SCIAMACHY, Terrestrial carbon cycle, Uncertainty estimation, Variational data assimilation
in
Remote Sensing of Environment
volume
203
pages
16 pages
publisher
Elsevier
external identifiers
  • scopus:85019643532
  • wos:000418464200009
ISSN
0034-4257
DOI
10.1016/j.rse.2017.08.017
language
English
LU publication?
yes
id
a1c02cf9-10f1-4ef8-87f9-a4ae7cfa6b88
date added to LUP
2017-12-18 08:23:54
date last changed
2018-01-16 13:27:58
@article{a1c02cf9-10f1-4ef8-87f9-a4ae7cfa6b88,
  abstract     = {<p>We present two novel earth observation products derived from the BESD and EMMA XCO<sub>2</sub> products which were respectively retrieved from SCIAMACHY and GOSAT observations within the GreenHouse Gas project of ESA's Climate Change Initiative (GHG-CCI). These products are inferred by a Carbon Cycle Data Assimilation System (CCDAS) and consist of net and gross biosphere-atmosphere fluxes of carbon dioxide on a global 0.5° grid. As a further dataset provided by the CCI, the burnt area product developed by its Fire忌i project was used in the CCDAS to prescribe the emission component from biomass burning. The new flux products are provided with per-pixel uncertainty ranges. Fluxes with uncertainty ranges can also be provided aggregated in space and time, e.g. over given regions or as annual means. For both, posterior flux fields inferred from BESD and EMMA products, transport model simulations show reasonable agreement with the atmospheric carbon dioxide concentration observed at flask sampling stations. This means that the information provided by the terrestrial and transport models, the respective GHG ECV product, the burnt area ECV product, a product of the Fraction of Absorbed Photosynthetically Active Radiation used to drive the model, and the atmospheric flask samples is largely consistent. The most prominent feature in the posterior net flux is the tropical source of CO<sub>2</sub> inferred from both products. But for the EMMA product this release, especially over South America, is with 300 gC/m<sup>2</sup>/year much more pronounced than for BESD. This confirms findings by a recent intercomparison of transport inversions using GOSAT data by Houweling et al. (2015). The reason for the larger net flux is increased heterotrophic respiration. For both products the posterior 2010 sink over Europe (without Russia) is in the range of a recent compilation of European flux estimates by Reuter et al. (2016b). The posterior 2010 uptake of Australia (including Oceania) inferred from the EMMA product is 1.3 ± 0.2 PgC/year and appears to confirm the high sink also derived from GOSAT by Detmers et al. (2015) over a slightly different period and area. While for some regions (USA, Canada, Europe, Russia, Asia) the one standard deviation uncertainty ranges derived from BESD and EMMA do overlap, for some other regions (Brazil, Africa, Australia) this is not the case. It is not clear yet whether this is due to the uncertainty specifications in the respective products or the handling of uncertainty in the assimilation chain. Assumptions on correlation of observational uncertainty in space and time have a considerable impact on the inferred flux fields (≈ 60 gC/m<sup>2</sup>/year). The effect of adding an uncertainty that approximates the error in the retrieval system is of similar size.</p>},
  author       = {Kaminski, T. and Scholze, M. and Vossbeck, M. and Knorr, W. and Buchwitz, M. and Reuter, M.},
  issn         = {0034-4257},
  keyword      = {Carbon fluxes,Column integrated carbon dioxide,GOSAT,SCIAMACHY,Terrestrial carbon cycle,Uncertainty estimation,Variational data assimilation},
  language     = {eng},
  month        = {12},
  pages        = {109--124},
  publisher    = {Elsevier},
  series       = {Remote Sensing of Environment},
  title        = {Constraining a terrestrial biosphere model with remotely sensed atmospheric carbon dioxide},
  url          = {http://dx.doi.org/10.1016/j.rse.2017.08.017},
  volume       = {203},
  year         = {2017},
}