Assessment of Wet Land Cover Types from Global Land Cover Products - Implications for the Global Methane Balance
Erasmi, Stefan1; Jungkunst, Hermann2; Grunwald, Dennis3
1Georg-August-University Göttingen, GERMANY; 2University of Koblenz-Landau, GERMANY; 3Thünen Institut Braunschweig, GERMANY

The knowledge about the global spatial coverage and latitudinal distribution of wetlands is of utmost importance for a reliable estimate of the global methane resources in soils. Remote sensing is a valuable means to map and monitor the status and changes of land surface properties at scales ranging from regional to global. Coarse to medium resolution remote sensing data deliver a globally consistent and objective source of information for a spatially explicit mapping of the global distribution of potential carbon stocks in terms of land cover type maps. However, there is still considerable uncertainty in estimates of the area and distribution of the relevant land cover types (e.g. wetlands for methane sources) and hence of the stored methane globally.
These uncertainties are attributed to a number of limitations that are either determined by the technical specification of the sensor (wavelength, spectral and spatial resolution), the accuracy of mapping or the thematic representation of the relevant land cover types within the existing global land cover data products. Several studies underlined the limitations in consistency and accuracy of the existing global land cover products (e.g. IGBP DISCover, UMD, MODIS 1-km, GLC2000, GlobCover) for mapping land cover classes (e.g. Herold et al. 2008) and especially for estimating the amount and spatial distribution of wetlands (e.g. Krankina et al. 2008). Further uncertainty is given by the fact that in many cases even the judgement of land cover types in being sources or sinks for atmospheric greenhouse gases is contradictory. E.g. Grunwald et al. (2012) showed the impact of treating boreal needle-leaved evergreen forest with a low tree cover as a source (instead of a sink) for methane.
In the present study, we compared three different global land cover products (MODIS 1-km, GLC2000, GlobCover) with regard to their ability to map land cover types that are considered as methane source in the global methane cycle. Based on a validation study in the Taiga zone of Northwest Russia, we estimated the uncertainty (error of commission) in mapping wetland areas and the consequences of mapping errors for modelling the net methane balance. The areal extent of wetlands in the three products is considerably different (267 to 833 km2). The commission errors range between 11.8 % (MODIS 1-km) and 35.2 % (GLC2000). However, only the classification system of the MERIS-based GlobCover product provides more detailed information about potential wetland cover types (e.g. boreal open needle-leaved forests) and thus seems to overcome some limitations in class definitions. In our study, we showed that the consideration of wet forest ecosystems considerably increases the net methane balance estimates for Europe with a bias of more than 100 % compared to conventional estimates based only on wetland areas.
Overall, the preliminary results indicate the necessity of proper class definitions for wet land cover types in the context of modelling net greenhouse gas balances. Furthermore it is pointed out that the existing global land cover types show strong inconsistencies and inaccuracies in wetland delineation and must undergo precise validation. Only recently, different approaches have been proposed for future validation strategies of global land cover maps, e.g. stratified random sampling (Ollofson et al. 2012) or crowdsourcing (Fritz et al. 2012). They provide a basis for improved global wetland mapping from future SENTINEL missions.

References:
Fritz, Steffen; McCallum, Ian; Schill, Christian; Perger, Christoph; See, Linda; Schepaschenko, Dmitry et al. (2012): Geo-Wiki: An online platform for improving global land cover. In: ENVIRONMENTAL MODELLING & SOFTWARE 31, S. 110-123.

Grunwald, Dennis; Fender, Ann-Catrin; Erasmi, Stefan; Jungkunst, Hermann F. (2012): Towards improved bottom-up inventories of methane from the European land surface. In: ATMOSPHERIC ENVIRONMENT 51, S. 203-211.

Herold, M.; Mayaux, P.; Woodcock, C. E.; Baccini, A.; Schmullius, C. (2008): Some challenges in global land cover mapping: An assessment of agreement and accuracy in existing 1 km datasets. In: REMOTE SENSING OF ENVIRONMENT 112 (5), S. 2538-2556.

Krankina, O. N.; Pflugmacher, D.; Friedl, M.; Cohen, W. B.; Nelson, P.; Baccini, A. (2008): Meeting the challenge of mapping peatlands with remotely sensed data. In: BIOGEOSCIENCES 5 (6), S. 1809-1820.

Olofsson, Pontus; Stehman, Stephen V.; Woodcock, Curtis E.; Sulla-Menashe, Damien; Sibley, Adam M.; Newell, Jared D. et al. (2012): A global land-cover validation data set, part I: fundamental design principles. In: INTERNATIONAL JOURNAL OF REMOTE SENSING 33 (18), S. 5768-5788.