Transcript User Vice Chair Report

Access to and Add Value of
Archived Data Methodology of Data Integration and Mining
for 1:1M Land Type Mapping of China
Prof. Liu Chuang
Prof. Shen Yuancen
Global Change Information and Research Center
IGSNRR/Chinese Academy of Sciences
PPF-WSIS Phase II, 14 November 2005, Tunis
1 China’s Scientific Data Sharing Program
2 Opportunities and Challenges: Access to
and Add Value of the Archived Data
3 Methodology of Adding Value of Archived
Data
4 Example:
1:1M Land Type Mapping of China
1 China’s Scientific Data Sharing Program
China has an implementation program in enhancing
open access to scientific data, a national long-term
(2005-2020) program: Scientific Data Sharing Program
(SDSP) which is initialed in 2003
About 40 data centers, 300 major databases covering
almost all of the basic sciences will be long term
supported, a series of data policies and data standards
will be established to meet the needs of open access to
the archived data.
Besides, e-Government
programs in agencies of China
and e-Sciences program in CAS
will promote the scientific data
sharing program greatly. For
example, the quick response
system of water resources
management system.
About 250 TB data archived with the
standard or near standard manners in
China (June 2005)
2 Opportunities and Challenges: Access to
and Add Value of the Archived Data
The progress makes great opportunities for
scientists in research:
• the location of data
• the way to access
• free or low costs
Two Major Challenges in China:
• Preservation and open access:
more stable, more open, more fast,
more easy and more low cost in
services, which is a long way to go
• Add Value: new methodology in
data integration and mining, which
is a new way to be created
3 Methodology of Adding Value of Archived
Data
The value of scientific data can be
divided into:
value for scientific research
value for social benefit
value for economic income
Relationship between data value and
data integration/mining
value
Dataset 3
Dataset 2
Dataset 1
time
Reference Hierarchical Model for
Data Integration and Data Mining
knowledge
Cal/Val
Object Simulating
Data Integration
data
Data
Selection
model
Distributed Information Infrastructure
• Data Selection: two important
issues in this stage
(1) how to select the necessary data
among the distributed data holders in
order to meet the need of modeling
for a specific objective
(2) how to determine the weights of
each selected datasets
• Data Integration:
one issue, very difficult issue, in this
stage has to be solved
- making the selected datasets
compatible
including data standard, termination,
definition, format, unit, resolution,
time period, method of capture the
data ….
• Object simulating:
two issue, the critical issues, in
this stage need to be solved
- establish a relationship between
the datasets selected (model)
- determine the parameters in the
model
• Cal/Val for the new dataset:
How the new dataset quality
could be:
- how quality is or what
conditions the new dataset or
knowledge could be high quality?
- Are there any way to help the
dataset quality enough?
• New knowledge/new dataset
created
go to publication and data
archiving process
Reference Hierarchical Model for
Data Integration and Data Mining
knowledge
Cal/Val
Object Simulating
Data Integration
data
Data
Selection
model
Distributed Information Infrastructure
Example:
Data Integration and Mining for
1:1M Land Type Mapping of
China
Land type research and 1:1M mapping in
China
There is a long history in China in land type
studies, the earlier record in 170 BC,
identified the China land into 9 types.
The most resent land type studies in 1:1M
mapping started in 1987, the first land type
classification system for 1:1M mapping of
China created in 1990 led by Prof. Zhao
Songqiao. landtypeclaSytemChina.doc
The stage of
completed
part of the
1:1M Land
Type Map of
China
Datasets：
(1)
(2)
(3)
(4)
(5)
(6)
The datasets used in this paper include:
Climate datasets in more than 600 climate stations from CMA
Soil map in 1:1M from CAS
MODIS-NDVI/EVI, 250m, 1kmresolution, 16-day and 10 days
composite 2002, from NASA and CAS
MODIS-NDSI, 1 km resolution, 10 days and monthly composite
2002, from CAS
SRTM in 90 Meters in USGS and DEM in 1:250k from
Geomatic Center of China
Ground truth survey datasets in Northeast China, Inner Mongolia,
Tibet, Gansu, Zhejiang, Guizhou …
(7) historical records including documentation and maps from CAS
(8) yearbooks of agriculture and land use from Statistic Bureau of China
MODIS-NDVI 16-days composite datasets, 2002, 1km
• Field sites
NDVI = (MODIS2-MODIS1)/ (MODIS2+MODIS1)
EVI = 2.5*(MODIS2-MODIS1)/(MODIS2+6*MODIS17.5*MODIS3+1)
NDSI = (MODIS4-MODIS6)/(MODIS4+MODIS6)
Forest (Betula)
0 NDVI 0.83
Single peak
10000
Location:
Far East Russia and
Daxingan Mountain in
Helongjian Province
10000*NDVI
8000
6000
4000
2000
0
1
2
3
4
5
6
7
-2000
Month
8
9
10
11
12
Location:
Great Hinggan Mt.
Forest (Larix+Betula, up)
Meadow steppe (down)
Location:
Huang-Huai-Hai Plain
Rotated crops land with
winter wheat and maize
Location:
North Korea
Forest (purple)
Rice (white)
Wetland (reed)
0 NDVI 0.53
0 EVI 0.42
EVI Time Series of Phragmites Australis
0.45
0.4
0.35
0.25
0.2
0.15
0.1
NDVI Time Series of Phragmites Australis
0.05
0.6
0
1
2
3
4
5
6
7
8
9
10
11
12
0.5
month
0.4
NDVI
EVI
0.3
Location: Yellow River Delta
0.3
0.2
0.1
0
1
2
3
4
5
6
7
month
8
9
10
11
12
9000
7000
8000
6000
5000
6000
Temperate Meadow
MODIS_NDVI*10000
MODIS_NDVI*10000
7000
0 NDVI 0.8
4000
3000
2000
2000
0
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
4500
6000
Temperate Meadow
5000
0 NDVI 0.6
4000
MODIS_EVI*10000
MODIS_EVI*10000
0 NDVI 0.6
3000
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
7000
4000
3000
2000
3500
Temperate Steppe
3000
0 NDVI 0.4
2500
2000
1500
1000
1000
0
4000
Temperate Steppe
1000
1000
0
5000
500
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
0
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
Location: Xilingol, Inner Mongolia
5000
3000
4500
2000
4000
Temperate Desert
MODIS_NDVI*10000
MODIS_NDVI*10000
2500
0 NDVI 0.25
1500
1000
0 NDVI 0.45
3000
2500
2000
1500
1000
500
500
0
3500
Sand Steppe
0
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
4000
2500
3500
3000
Temperate Desert Steppe
1500
MODIS_EVI*10000
MODIS_EVI*10000
2000
0 NDVI 0.2
1000
Sand Steppe
2500
0 NDVI 0.35
2000
1500
1000
500
500
0
0
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
Months
Jan. Feb. Mar. Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.
Location: Xilingol, Inner Mongolia
Fig.2 互花米草盐沼，S.alterniflora
4000
3500
10000*EVI
3000
2500
2000
1500
1000
500
0
1
2
3
4
5
6
7
月份，Month
8
9
10
11
12
Location: Coastal area in
Northern Jiangsu province
Fig.1 互花米草盐沼，S.alterniflora salt march
7000
10000*NDVI
6000
5000
Wetland
4000
3000
0 NDVI 0.52
2000
1000
0
1
2
3
4
5
6
7
月份，Month
8
9
10
11
12
0 EVI 0.35
9000
8000
10000*NDVI
7000
6000
5000
4000
3000
2000
1000
0
1
2
3
4
5
6
7
8
9
10
11
12
alpine meadow,month
7000
Location:
6000
Qinghai Province
Alpine Meadow
10000*EVI
5000
4000
3000
2000
1000
0
1
2
3
4
5
6
7
8
alpine meadow,month
9
10
11
12
1400
1200
10000*NDVI
1000
800
600
400
200
0
1
2
3
4
5
6
7
8
Gobi,Month
9
10
11
12
900
Gobi in arid region in northwestern China
800
Location: MinQin County,
Gansu Province
10000*EVI
700
600
500
400
Gobi
300
200
100
0
1
2
3
4
5
6
7
Gobi,Month
8
9
10
11
12
7000
Location:
6000
10000*NDVI
5000
MinQin County (Oasis),
Gansu Province
4000
3000
2000
Spring Wheat Crop
1000
0
1
2
3
4
5
6
7
8
9
10
11
12
6
7
8
Wheat,Month
9
10
11
12
Wheat,Month
6000
10000*EVI
5000
4000
3000
2000
1000
0
1
2
3
4
5
Land
Location: Gongbujiangda area located at the Eastern Tibet
April 2001
June 2001
August 2001
Location:
Nyainqntanglha Mountains
NDSI >0.4 and MODIS2 >
0.11
Up left: Feb.2002
Up right: June 2002
Down left: Sep. 2002
Conclusion:
The reference Hierarchical mode of data
integration and mining is very important
for innovated knowledge development,
the computational science plays a
critical role in the new methodology.
The new methodology in data
integration and mining will take China
land type studies into a new milestone.
Thank you !