Geography Reference
In-Depth Information
8.1.2 Classification of LUCC Data for Climate Models
There are a number of schemes that have been proposed for land cover categori-
zation from regional to global-scale, including the International Geosphere-Bio-
sphere Programme (IGBP)—Data and Information Systems: Land Cover Working
Group land cover categorization scheme (Belward and Loveland
1995
), a six-class
biome categorization, the Simple Biosphere Model scheme (Sellers et al.
1996
), and
the Federal Geographic Data Committee vegetation characterization and informa-
tion standards (F. G. D. Committee
1996
). Many parameters in land surface model
are identified based on land cover types; for instance, time-invariant model variables
(e.g., vegetation reflectance, canopy top height, canopy base height, root depth, and
leaf respiration factor) in the Simple Biosphere Model 2 (SIB2) (Sellers et al.
1996
)
and the Common Land Model (CoLM) (Dai et al.
2001
). Thus, the specific land
cover classification units must not only be discernible (with high accuracy) from
remotely sensing image and ancillary data but also be directly related to the physical
characteristics of land surface. The IGBP scheme embraces the same philosophy but
with modifications to be compatible with existing schemes used by environmental
models, to incorporate land use in addition to land cover and to represent mosaics
(Belward and Loveland,
1995
).
In order to enhance the studies on land use/cover change, data have to be updated to
increase the accuracy. Time series land cover datasets have been widely used in
numerous climate simulation projects. Most attention has been paid on effects of the
accuracy of land cover data on climate simulation. Though there are temporal land use
data with accuracy higher than 90 % (Wu et al.
2013
), the high-precision land cover
data is still absent. Therefore, there is an urgent need to reclassify the LUCC dataset to
feed into Global Climate Models (GCMs) and Regional Climate Models (RCMs). For
example, in a case study on North China, Wu et al. (
2013
) overlaid the land cover
maps of the IGBP-DIS, GLC (Loveland et al.
2000
), University of Maryland Data
(UMD) and Data Center for West China (WESTDC), and selected the compatible
grids with classification as sample grids. They then combine land cover data with land
use data to generate new land cover data of high accuracy for climate simulation.
Their study showed that the C4.5 algorithm was suitable for converting land use data
to land cover data of IGBP classification. The temporal land cover data produced by
their method can meet the accuracy requirement of climate simulation and can be used
as the parameters of dynamical downscaling in regional climate simulation, which
constitutes a significant improvement in data processing.
8.1.3 Data Resolution and Reliability
Land surface has considerable heterogeneity because of the existence of different
land cover types such as bare area, water bodies, urban land, trees, and snow/ice,
which vary over small distance. This surface variability not only determines the
