peatland in both global land cover data sets. It may, therefore,
suggest that the precision wetland water based on computer
automatic extraction methods could meet the demands of
research at the global scale. However, the very low precision of
identification of peatland means that there are still substantial
uncertainties in these wetland-related landcover types data
when used directly as wetland type data; on the other hand,
improvements in the classification algorithms related to
peatland extraction should be made in the future.
(II) The agreement between the GLC2000 data and reference
data is higher than that of MODIS data and reference data from
both the class area and the spatial location. This is possibly
because of the different classification algorithms between the
global data sets. Since the GLC2000 is developed by the
coordination of the more than 30 groups, utilizing different
classification algorithms, different evaluation methods and
various regional schemes from region to region, the precision
of GLC2000 is available at regional scale. In comparison, the
MODIS global landcover data set is generated at global scales
aiming at various global research areas including climate
change, biodiversity conservation, ecosystem assessment, and
environmental modeling and they adopt the uniform
classification algorithm for the convenience of periodic
updating. Therefore, the MODIS data set is more practical for
research at global scale.
The class-specific accuracies of 38.1% and 45.9% in the
MOD12Q1 and GLC2000’ peatlands by Herold et. al(Herold et
al., 2008) have been calculated from the original samples using
documented theory for stratified random sampling and
considering the map area proportions for each class. Given the
different samples and validation frameworks, it is inappropriate
to compare the absolute numbers of accurate directly.
(IIT) The main reasons for low precision in these two global
land cover products include (a)the different aims of various
products and therefore the inconsistent wetland definitions
in their systems; (b) the coarse spatial resolution of satellite
images used in global data, which leads to the existence of
substantial mixed pixels that could greatly, reduce the
classification precision of global data sets especially for
fragmentized and heterogeneous landscapes; (c) Discrepancies
among the image data used in global data sets and reference
data. Because of the highly dynamic characteristics of wetlands,
the difference in image acquisition date usually leads to
discrepancies among data sets, especially in areas with distinct
seasonal variation.
Much more attention must be paid during the application of
existing global land cover products in global /wetland-related
research due to their low precisions. At the same time, it is
necessary to develop wetland-specific landcover classification
schemes and image classification methods by using
multi-sources and multi-classifiers in future.
Acknowledgement:
We are greatly thankful for careful revision by Arthur
Cracknell. This work was funded by the National High-tech
R&D Program of China (863 Program 2009AA122003-7).
References:
Costanza R.A., Groot R., et al., 1997, the value of the world's
ecosystem services and natural capital. Nature, 386, 253-260.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
Friedl M.A., Mclver D.K., Hodges J. C. F., et al., 2002, Global
land cover mapping from MODIS: algorithms and early results.
Remote Sensing of Environment, 83, 287-302.
Giria C., Zhu Z.L. and Reed B., 2005, A comparative analysis
of the Global Land Cover 2000 and MODIS land cover data
sets. Remote Sensing of Environment, 94, 123-132.
Gong P., 2009, Accuracy evaluation of the global land cover
based on the global flux observation sites. Progress in Natural
Science, 19, 754-759. (In Chinese)
Gong P., Niu Z.G., Cheng X., et al, 2010, China's wetland
change (1990-2000) determined by remote sensing. Scince
china earth sciences, 53(7), 1036-1042.
Herold M., Mayaux P., Woodcock C.E., et al., 2008, Some
challenges in global land cover mapping: An assessment of
agreement and accuracy in existing 1 km datasets. Remote
Sensing of Environment, 112, 2538-2556.
Hong Z.G., Dai E.F., Lin Z.J., et al., 2006, The method on
image recognizing and interpreting surface hydrologic features
with remote sensing on qinghai tibetan plateau. Science of
Surveying and Mapping, 3, 82-84.
Li GL, Du PJ, Wang XM. et al, 2009, Consistency
Evaluation and Integration Application of Land Cover
Classification Results from Multi-source Remotely Sensed
Images. Geography and Geo-Information Science, 25, 68-71.
Lehner B. and Petra, 2004, Development and validation of a
global database of lakes, reservoirs and wetlands. Journal of
hydrology, 296, 1-22.
Loveland T.R, Reed B.C, Brown JF. et al, 2000,
Development of a global land cover characteristics database
and IGBP DISCover from 1 km AVHRR data. International
Journal of Remote Sensing, 21, 1303-1330.
Niu H.G.&Ma X.H., 1985, The swamp wetland in china. The
Commercial Publishing Service, Beijing, china
Niu Z.G., Gong P., et al., 2009, The preliminary mapping of
remote sensing over china and corresponding geography
analysis. Scince china earth sciences, 39, 188-203.
Pflugmacher D., Krankina O., Cohen W. B, 2007,
Satellite-based peatland mapping: Potential of the MODIS
sensor. Global and Planetary Change, 56, 248-257.
Latifovica R., Olthof L, et al., 2004, Accuracy assessment
using sub-pixel fractional error matrices of global land cover
products derived from satellite data. Remote Sensing of
Environment, 90, 153-165.
Ran Y.H. Li X., Lu L., 2010, Evaluation of four remote
sensing based land cover products over China. International
Journal of Remote Sensing, 31, 391-401.
Wu W.B. Yang P. Zhang L., et al, 2009, Accuracy
assessment of four global land cover datasets in China.
Transactions of the Chinese Society of Agricultural
Engineering, 25, 167-173.
