Evaluating Regional Land Use Cover Trend
with Biophysical Settings: A Remote Sensing
and GIS integrated Approach
R. N. Sahoo a ,K. Suzanchi b, S. Pandey c , S. Arekhi d, R.K. Tomar a, P.Chandna c
a
Division of, Agricultural Physics
Division of Environmental Sciences,
Indian Agricultural Research Institute, New Delhi (IARI), 110012, India,
C Rice-Wheat Consortium for Indo-Gangetic Plains, CIMMYT, New Delhi, 110012, India,
d Collage of Agriculture, Ilam University, Ilam, Iran
b
INTRODUCTION
• The Indo-Gangetic Plain is a rich, fertile and ancient land
encompassing most of Northern and Eastern India where,
land use particularly agriculture has undergone dramatic
changes in the last four decades. Of late evidence is
accumulating to suggest that high productivity and growth
rates achieved during the Green Revolution era are no
longer being sustained to meet the needs of still
increasing population in the region.
•
There has been effort to search for unattended regions to
scale up the agricultural productivity to meet these needs.
Keeping this in view, an attempt was made to undertake
land use land cover change analysis over two decade
and evaluate different land use particularly agriculture
and its change with respect to its spatially varying
biophysical parameters such as topography, land form,
soil, weather, water resources of the regions.
Objectives
•
To undertake land cover change analysis over
two decades and evaluate different land use
particularly agriculture and its change
with
respect to its spatially varying biophysical
parameters such as topography, land form, soil,
weather, water resources of the regions.
•
This would form a frame work to predict possible
land use type or change in land use based on
single or composite effect of these biophysical
parameters at regional scale.
Study Area
Indo-Gangetic Plain Indian States
Spot vegetation 21th February 2005 (R:MIR, G: NDVI, B:B0)
Study Area
Indo-Gangetic
Plain India States
Spot vegetation 21th may 2005 (R:MIR, G: NDVI, B:B0)
Material
• Global land cover images:
– 1 Kilometer Land Cover Map Derived From AVHRR Data
(University of Maryland: The Global Land Cover Facility)
– MODIS/Terra Land Cover Type Yearly L3 Global 1km SIN Grid
(MOD12Q1)
• Biophysical map of IGP:
–
–
–
–
–
DEM Data (GTOPO30 )
Soil Data (FAO)
Agro Ecological zones (NBSS & LUP )
Irrigated Areas at Different Levels (IWMI)
State and district maps
• Statistical Data
– Land use statistics (India Harvest, CMIE)
Methodology
• Land use/cover change analysis was done using
multitemporal satellite data of 1km resolution
available in public domain.
• Spatially varying land forms and topography of
the region were derived from DEM data of
GTOPO30 available at 1km resolution.
• Information on soil and water resources (i.e.
irrigation area) were taken from available soil
map of FAO and IWMI respectively.
Flow diagram of methodology
South Asia
IWMI Map
South Asia
DEM
FAO soil
zone
Agro ecological
Zones
Georeferencing
Retrieval
for IGP
Retrieval
for IGP
Digitization
Geo-coding
Classified
soil Type
map
Classified
AESR zones
Retrieval for
IGP
Classified
elevation
AVHHR land cover
1984
Modis land cover
2005
Geo-referencing and
retrieval for IGP
Resize and
modification
Extract class type 1 and
merge to IGBP type
Change analysis
Map
differencing
Classification
comparison
Changes
thresholds
Table and
Histogram
Overlay analysis
Finding zones and possible factor/s governing
agricultural land cover change
Change detection
statistics
Change type
& direction
One Kilometer Land Cover Map Derived From AVHRR Data (IGBP)
Deriving a global land cover classification product.
•
•
The product was derived by testing several metrics that describe
the temporal dynamics of vegetation over an annual cycle.
These metrics have the potential to be used as input variables to a
global land cover classification. Tested metrics are based on :
1)
2)
3)
4)
•
the ratio between surface temperature and NDVI,
seasonal metrics derived from the NDVI temporal profile such as length of
growing season,
a rule-based approach that determines cover type through a series of
hierarchical trees based on surface temperature and NDVI values, and
annual mean, maximum, minimum, and amplitude values for all optical and
thermal channels in the AVHRR Pathfinder Land (PAL) data.
These metrics were applied to 1984 PAL data to derive a global
land cover classification product using a decision tree classifier
(University of Maryland).
AYHHR Land Cover Type (IGBP):
Class
Type 1
0
Water
1
Evergreen Needleleaf Forest
2
Evergreen Broadleaf Forest
3
Deciduous Needleleaf Forest
4
Deciduous Broadleaf Forest
5
Mixed Forest
6
Woodland
7
Wooded Grassland
8
Closed Shrubland
9
Open Shrubland
10
Grasslands
11
Croplands
12
Bare Ground
13
Urban and Built-Up
MODIS/Terra Land Cover Type Yearly L3 Global 1km SIN Grid
MOD12Q1 - Type 1(IGBP)
•
The MODerate-resolution Imaging Spectroradiometer (MODIS) Land Cover
Classification products contain 5 types classification schemes describing
land cover properties such as
Type 1 : IGBP (International Geosphere-Biosphere Programme) global
vegetation classification scheme
Type 2: (UMD) University of Maryland modification of the IGBP scheme
Type 3: MODIS LAI/fPAR
Type 4: MODIS Net Primary Production
Type 5: Plant Functional Types (PFT) (in consideration of the Community Land Model
(CLM) used in climate modeling).
These classes are distinguished with a supervised decision tree classification method.
•
The primary land cover scheme IGBP identifies 17 classes which includes
-
11 natural vegetation classes
03 developed land classes
permanent snow or ice,
barren or sparsely vegetated
water.
The MOD12 classification schemes are multitemporal classes describing land cover
properties as observed during the year (12 months of input data).
MODIS (MODI12Q1) Land Cover Type:
class
IGBP (Type 1)
UMD (Type 2)
LAI/FPAR (Type 3)
NPP (Type 4)
PFT (Type 5)
0
Water
water
water
water
Water
1
evergreen needle leaf forest
evergreen needle leaf
forest
grasses/cereal crops
evergreen needle leaf
vegetation
Evergreen needle leaf
trees
2
evergreen broadleaf forest
evergreen broadleaf
forest
shrubs
evergreen broadleaf
vegetation
Evergreen broadleaf trees
3
deciduous needle leaf forest
deciduous needle leaf
forest
broadleaf crops
deciduous needle leaf
vegetation
Deciduous need leleaf
trees
4
deciduous broadleaf forest
deciduous broadleaf
forest
savanna
deciduous broadleaf
vegetation
Deciduous broadleaf
trees
5
mixed forests
mixed forests
broadleaf forest
annual broadleaf
vegetation
Shrub
6
closed shrub lands
closed shrub lands
Needle leaf forest
annual grass vegetation
Grass
7
open shrub lands
open shrub lands
unvegetated
non-vegetated land
Cereal crop
8
woody savannas
woody savannas
urban
urban
Broadleaf crop
9
savannas
savannas
Urban and built up
10
grasslands
grasslands
Snow and Ice
11
permanent wetlands
12
croplands
croplands
13
urban and built-up
urban and built-up
14
cropland/natural vegetation
mosaic
15
permanent snow and ice
16
barren or sparsely vegetated
Barren or sparse
vegetation
barren or sparsely
vegetated
Comparison of Land Cover Type of MODIS and AVHRR
Class
MODIS-IGBP (Type 1)
AYHHR Land Cover Type
Class
0
water
water
0
1
evergreen needle leaf forest
Evergreen Needle leaf Forest
1
2
evergreen broadleaf forest
Evergreen Broad leaf Forest
2
3
deciduous needle leaf forest
Deciduous Needle leaf Forest
3
4
deciduous broadleaf forest
Deciduous Broad leaf Forest
4
5
mixed forests
Mixed Forest
5
6
closed shrub lands
Woodland
6
7
open shrub lands
Wooded Grassland
7
8
woody savannas
Closed Shrub land
8
9
savannas
Open Shrub land
9
10
grasslands
Grasslands
10
11
permanent wetlands
Croplands
11
12
croplands
Bare Ground
12
13
urban and built-up
Urban and Built-Up
13
14
cropland/natural vegetation mosaic
15
permanent snow and ice
16
barren or sparsely vegetated
Merging classes based on IGBP-DIS definition (Hansen et al, 2000)
Land Cover Map of 1 km derived from AVHRR Data (IGBP)
1984
MODIS/Terra Land Cover Type
Yearly L3 Global 1km SIN Grid MOD12Q1 Type 1(IGBP)
2005
W
at
er
land cove r
U
rb
an
an
d
Ba
re
Bu
il t
-
U
p
G
ro
un
d
C
ro
pl
an
d
AVHHR 1984
G
ra
ss
la
nd
W
oo
W
dl
oo
an
d
de
d
G
ra
ss
la
C
nd
lo
se
d
S
hr
ub
la
nd
O
pe
n
Sh
ru
bl
an
d
N
ee
Ev
dl
e
er
le
gr
af
ee
n
Br
D
oa
ec
d
id
le
uo
af
us
N
ee
D
dl
ec
e
id
le
uo
af
us
Br
oa
d
le
af
M
ix
ed
Fo
re
st
Ev
er
gr
ee
n
Area '000 ha
Comparison of Land Cover area
Land cover Comparison
MODIS 2005
50,000,000
45,000,000
40,000,000
35,000,000
30,000,000
25,000,000
20,000,000
15,000,000
10,000,000
5,000,000
0
Biophysical data
Digital Elevation Model
1000 m
1m
Biophysical data
Elevation classes
Biophysical data
Soil Zone
Biophysical data
Irrigated Areas at Different Levels
Biophysical data
Agro Ecological Sub Region
Biophysical data
Agro Ecological Sub Region
AGRO_ZONE
AGRO_ECO_ZONE
REGION
2.1
M9Eh1
Western Plains Hot arid
2.3
M9Et2
Western Plains Hot arid
4.1
N8Dd3
Northern plain and Central high lands
4.3
N8Dm4
Northern plain and Central high lands
4.4
16Dm4
Northern plain and Central high lands
9.1
N8Dm(cd)4
Northern, plain hot subhumid (dry)
9.2
N8Cd5
Northern, plain hot subhumid (dry)
10.3
16Cd5
Central high lands(malwa,bundelkhand,eastern satpura,hot subhumid(dry
11.0
J3Cd(cm)5
12.3
J2cd5
13.1
O8Cd(cm)6
Eastern plain,hot subhumid (moist)
13.2
B10Cm6
Eastern plain,hot subhumid (moist)
14.2
A15Cd(cm)6
Western Himalayas,warm,moist semiarid to dry sub humid
14.5
A10B(A)9
Western Himalayas,warm,moist semiarid to dry sub humid
15.1
O8Cm7
Bengal and Assam Plain,hot subhumid(moist) to humid
15.3
Q8A9
Bengal and Assam Plain,hot subhumid(moist) to humid
16.2
C11A10
Eastern Himalayas,warm perhumid
18.5
S7Cm7
Gangetic Delta, hot, moist-subhumid
sloping chattisgarh/Mahanadi Basin, hot,moist-subhumid
Chhotanagar Plateau and Garhjat Hills,hot,dry and moist,subhumid tran
Biophysical data
Topography
Soil Map
Different Irrigated Areas
Agro ecological Sub Region
Change Detection Analysis
Map difference
• Difference map of two classified images was
prepared with thresholds -0.5 to +0.5 having five
classes.
• This map show the extend of change that happened
in all classes of land cover
• Four classes of change include change[+] and no
change[-] as shown in next slide.
Map difference
Change Detection Statistics
• While generating image having change detection
statistics, we compared two classified image class
by class
• Related to two classified Images, any of produced
band shows exact how any land use type in first
image changed to other land use in last one.
• This method shows the direction of change exactly
and it is useful for understanding the land use
dynamics and possible driving forces acting
upon it.
Change Detection Statistics
Change in Grassland area to other
classes (cropland and urban)
Change Detection Statistics
Change in Woody Grassland area
to other classes ( Cropland )
Change Detection Statistics
Change in Woodland area to other
classes ( cropland and mixed forest )
Change Detection Statistics
Change in Open shrubland area to
other classes( cropland and urban)
Change Detection Statistics
Change in Closed shrubland area to
other classes (cropland and urban)
Change Detection Statistics
Change in Cropland area to
other classes (Urban Areas)
Change Detection Matrix
Class Total
Row Total
Urban and Built-up
Bare Ground
Cropland
Grassland
Woodland
Wooded Grassland
Open Shrubland
Closed Shrubland
Mixed Forest
Deciduous Broadleaf
Forest
Deciduous Needleleaf
Forest
Evergreen Broadleaf
Foreset
Water
2005
Evergreen Needleleaf
Forest
1984
Unclassified
3.343
3.517
3.875
0
4.358
3.271
0.567
0.635
0.928
1.464
0.683
0.383
6.25
0.057
0.233
100
Water
42.21
0.306
0.165
0
0.459
0
1.299
3.583
1.02
0.452
4.094
0.636
37.5
1.71
96.869
100
Evergreen Neadleleaf Forest
0.173
20.948
2.391
0
6.651
17.29
0.023
0.04
0.034
0.257
0.07
0.005
6.25
0.2
95.626
100
Evergreen Braodleaf Forest
0.207
20.183
36.109
0
26.491
36.449
0.041
0
0.4
6.298
0.098
0.038
0
0
98.936
100
0.007
0
0
0
0
0
0
0
0
0
0.006
0.001
0
0
100
100
Deciduous Broadleaf Forest
0.055
7.187
6.513
0
5.275
6.075
0.007
0
0.067
0.601
0.015
0.019
0
0
98.312
100
Mixed Forest
0.552
20.795
17.89
0
13.532
13.551
0.1
0
0.718
15.144
0.249
0.079
0
0.542
98.457
100
Closed Shrubland
0.463
0.765
0.824
0
0.573
0.935
0.079
0.06
0.086
0.139
0.19
0.04
0
0.171
98.829
100
Open Shrubland
2.548
0.306
0.577
0
0.688
0
0.921
6.537
0.299
0.241
0.858
0.151
25
0.342
96.991
100
Woody grasslan
0.242
20.489
22.836
0
20.528
18.692
0.691
0.147
3.387
8.692
0.597
0.45
0
1.34
97.753
100
woodland
0.331
1.682
2.226
0
2.179
1.402
0.138
0.114
0.409
0.745
0.212
0.095
0
0.228
97.702
100
Grassland
0.207
1.529
2.143
0
1.261
0.935
0.038
0.06
0.256
0.611
0.118
0.068
0
0.057
98.833
100
cropland
45.2
2.294
4.369
0
18.005
1.402
92.617
83.818
91.258
64.43
89.633
96.692
12.5
36.517
99.453
100
Urban and Built-Up
2.355
0
0.082
0
0
0
3.226
3.823
0.904
0.627
2.461
1.267
12.5
58.637
99.899
100
Bare Ground
2.106
0
0
0
0
0
0.253
1.183
0.234
0.298
0.715
0.075
0
0.2
97.412
100
100
100
100
0
100
100
100
100
100
100
100
100
100
100
0
0
57.79
79.052
63.891
0
94.725
86.449
99.921
93.463
96.613
99.255
99.882
3.308
100
41.363
0
0
5.449
-23.089
132.481
0
-45.642
2200.935
-99.018
-74.901
-91.38
-93.07
-99.088
70.394
11975
268.016
0
0
Deciduous Needleleaf Forest
Class Total
Class Changes
Image Difference
Discussion: Difference Map with biophysical settings
Irrigation Area
Difference map
Agro ecological Sub
Region
Topography
Soil Map
Discussion: Comparison of Changed areas with topography
Woody Grassland Change
Grassland Change
Close shrubland Change
Discussion: Comparison of Changed areas with soil types
Woody Grassland Change
Grassland Change
Close shrubland Change
Discussion:
Woody Grassland Change
Comparison of Changed areas with different irrigated areas
Grassland Change
Close shrubland Change
Discussion: Comparison of Changed areas with Agro Ecological Sub Regions
Woody Grassland Change
Grassland Change
Close shrubland Change
Conclusions
• Land cover products at 1km resolution were found
to be very useful for its monitoring at regional
scale leading to identify hotspots and update land
use cover statistics.
• Major biophysical parameter/s of the region could
be found correlated with land cover change hot
spots.
• Major land use classes have been changed to
crop land to meet the demand of increasing
population trend of the region. But, Qs is next
what if the same trend continues ?
k.suzanchi@gmail.com
rnsahoo@iari.res.in