Geography Reference
In-Depth Information
Table 4.3
Schemes of the simulation test
Test
Test period
Land cover data used in the WRF model
Control test
2005.10-2007.12
Land cover data of 1985
Sensitivity test
2005.10-2007.12
Land cover data of 2005
the data is converted from 1 to 20 km. The land use/cover data used in this study
are extracted from the Chinese subset of the Global Land Cover Characteristics
database, which is developed based on the AVHRR data with the support of IGBP-
DIS in 1992, and the China subset of the MODIS land cover data product in 2005.
The two datasets have the spatial resolution of 1 km and adopt the IGBP classi-
fication. On the basis of the data of the IGBP land cover classification, we for-
mulate the transformation method from the IGBP land cover classification to the
USGS land cover classification (Table
4.4
), and establish the land-use and land-
cover dataset of the North China Plain of the USGS classification.
The LUCC data are further upscaled on the basis of the data mentioned earlier so
as to embed the high resolution underlying surface data into the large-scale climate
model. In this study, the 1 km resolution land-cover and land-use data of USGS
classification are upscaled into the 20 km resolution data. Besides, the three kinds of
data are integrated in a system of seven land cover types, and their area consistency
and spatial consistency are analyzed so as to check the change of the classification
accuracy of the land-cover and land-use data before and after the reclassification and
upscaling. The result shows the difference of the total area between the land-cover
and land-use data after reclassification and upscaling the initial data (Table
4.4
).
There is a high overall consistency between the 1 km resolution initial data of
the IGBP classification and USGS classification in year 1992 and 2005, except the
slight difference in the total area of grassland, water bodies, and unused land.
Besides, the 20 km resolution data of USGS classification differed from both of the
other two kinds of Land-cover and land-use data. The result of the comparison
between the initial data and the upscaled USGS data of year 1992 indicates the
area of irrigated cropland and pasture, grassland, and water bodies decreased by
2.69, 7.78, and 41.42 %, respectively; while the area of dryland cropland and
pasture, deciduous broadleaf forest, urban and built-up land, and unused land
increased by 0.62, 0.46, 1.68, and 153.66 %, respectively. By contrast, that of year
2005 indicates the area irrigated cropland and pasture, deciduous broadleaf forest,
water bodies, and urban and built-up land decreased by 6.46, 7.43, 36.35 and
4.87 %, respectively; while the area of dryland cropland and pasture, grassland,
and unused land increased by 1.41, 17.29, and 101.11 %, respectively (Table
4.5
).
The error matrix is used to assess the spatial consistency between the initial data
and the data after reclassification in this study. The result indicates that the con-
sistency of the land use/cover types except the unused land has all exceeded 95 %
(Table
4.6
). The consistency of the urban and built-up land is the highest, reaching
99.25 %, followed by that of dryland cropland and pasture, which is 98.76 %. The
overall consistency has reached 96.84 % and the Kappa coefficient is 0.95, indi-
cating the reclassification result has a high classification accuracy.
