Land Use/ Land Cover Mapping Initiative Watershed Kansas Biological Survey

advertisement
Land Use/ Land Cover Mapping Initiative
for Kansas and the Kansas River
Watershed
Kansas Biological Survey
Kansas Applied Remote Sensing Program
April 2008
Previous Kansas LULC Projects
Kansas LULC Map (1990)
10 Level I classes
1989-1991 Landsat TM data
Single-date (summer)
Three-year project
Kansas GAP Map (1996)
43 classes; 40 are natural
vegetation
1992-1995 Landsat TM data
Multi-seasonal (spring, summer,
fall)
~5-year project
Summary of LULC products for EPSCoR
– Phase 1:
• Modified Level I land use/land cover product derived from
30-meter Landsat Thematic Mapper data
– Phase 2:
• Change detection products created by analysis of the new
LULC maps with past LULC maps;
• Cool-season/warm season grasslands mapping using
multitemporal 250-meter MODIS data;
• Crop type identification and crop rotation products, also
created from multitemporal MODIS data.
Level 1 Land Use/Land Cover Map Description
Phase 1: Modified Level I LULC mapping
– Multitemporal Landsat TM (30-meter resolution)
• Seasonal data: Spring, summer, & fall
– Maximum separability of classes
• Image dates: 2004-05
– Classes Mapped
• Forest, water, cropland, grassland, rural developed, and
urban commercial/industrial, urban residential, urban
open, and urban woodland.
– Comparable to 1990 LULC map (classes & MMU)
• Allows change detection
Land Cover Classification Methodology
Data Acquisition
• 2004/2005 Terrain-corrected
TM imagery
• 3-date: Spring, Summer, & Fall
Image Processing:
• Layer Stack
• Clip Spatial Extents
• Mask Clouds
Unsupervised
Classification:
ISODATA & Maximum
Likelihood Classifiers
• Assign and recode
spectral classes to
LULC classes
• Cluster Busting
Generalize to MMU
Mosaic Processing
Units
Using existing KS GAP
and USDA databases
and aerial imagery
interpretation.
Map
Refinements/Editing
Accuracy Assessment
Map & Deliverables
Map Refinement: Addition of Conservation Reserve lands
Grass/Crop Image
CLU data (CRP land in red)
CLU data added
• From a mapping perspective: CRP land was accurately
mapped as grasslands
• From an ecological perspective: A large number of grasslands
are at risk of being converted back to cropland
Generalization using CLU data
Pre-generalized
Traditional Generalization Techniques
Generalized Using Common Land Unit
Land Cover Change: Nemaha County
1990 Land Cover Map
2005 Land Cover Map
Rural Land Cover Change: Nemaha County
1990 Land Cover Map
2005 Land Cover Map
Urban Land Cover Change: Johnson County
1990 Land Cover Map
2005 Land Cover Map
Phase 1 Land Use/Land Cover Map
Accuracy Assessment: In Progress
• Accuracy Levels reported: Overall, User’s, Producer’s, and
the KAPPA statistic
• Stratified random sampling design
– Sample size proportionate to the area mapped for each
land cover class
• Using existing databases as reference or ground-truth to
assess accuracy levels for cropland, grassland, and
woodland.
– Kansas GAP database
– USDA database
• Using aerial imagery (NAIP) interpretation techniques to
assess accuracy levels for water and urban.
Getting the land cover data …
http://www.kars.ku.edu/
Getting the land cover data …
http://www.kars.ku.edu/projects/ecoforecasting/
Data contents
• Arc GRID files of watershed land cover (actual
and buffered boundaries)
• Watershed boundary files: actual and buffered
– Raster and vector formats
• County boundaries for watershed area
• Ancillary data
– ArcMap .mxd file (Arc 9.2 document)
– Arc layer (.lyr) file
– Name files for land cover classes (.doc & .xls)
Level II Land Cover Mapping
Phase 2: Map Grasslands and Land Management
– LULC Mapping:
•
•
•
•
Crop type
Crop rotation practices
Warm season/cool season grasslands mapping
Irrigated vs. non-irrigated croplands*
– Data:
• Time-series 250m MODIS imagery
• 6 years MODIS data in KBS-KARS archive
– 2005 data as the target year
– 16-day composites, 23/year
Comparison of MODIS and Thematic Mapper imagery
Landsat TM 30m resolution
MODIS 250m resolution
Dense Time-Series Maximum Value Composites
• More is better: multi-date maximum-value image
composites.
– Select the “best” pixel, i.e., the pixel with the
highest green vegetation response, over a given
time period (7, 10, 14, 16 days) – this tends to
eliminate pixels contaminated by clouds, noise, etc.
– Create a new image consisting of the maximum
values for the given time period.
– Thus, if bi-weekly (14-day) composites are used,
you get 26 composite images per year → “dense
time series” or “hyper-temporal” imagery.
• How does this look over the course of a year?
Alfalfa
Fallow
Summer Crops
Winter Wheat
Dec. 19
Dec. 3
Nov. 17
Nov. 1
Oct. 16
Sept. 30
Sept. 14
August 29
August 13
July 28
July 12
June 26
June 10
May 25
May 9
April 23
April 7
March 22
March 6
Feb. 18
Feb. 2
Jan. 17
Jan. 1
NDVI
General Crop Types
Average multi-temporal NDVI profiles for Kansas in 2001
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
Corn
Sorghum
Soybeans
Dec. 19
Dec. 3
Nov. 17
Nov. 1
Oct. 16
Sept. 30
Sept. 14
August 29
August 13
July 28
July 12
June 26
June 10
May 25
May 9
April 23
April 7
March 22
March 6
Feb. 18
Feb. 2
Jan. 17
Jan. 1
NDVI
Summer Crop Types
Average multi-temporal NDVI profiles for Kansas in 2001
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Irrigated and Non-Irrigated Crops
Average multi-temporal NDVI profiles for Kansas in 2001
0.9
0.8
0.7
0.5
0.4
0.3
0.2
Corn (Irrigated)
Corn (Non-Irrigated)
Winter Wheat (Irrigated)
Dec. 19
Dec. 3
Nov. 17
Nov. 1
Oct. 16
Sept. 30
Sept. 14
August 29
August 13
July 28
July 12
June 26
June 10
May 25
May 9
April 23
March 22
March 6
Feb. 18
Feb. 2
Jan. 17
0
April 7
0.1
Jan. 1
NDVI
0.6
Winter Wheat (Non-Irrigated)
0
Dec. 19
Dec. 3
Nov. 17
Nov. 1
Oct. 16
Sept. 30
Sept. 14
August 29
August 13
Winter Wheat
July 28
July 12
June 26
June 10
May 25
May 9
April 23
April 7
March 22
March 6
Feb. 18
0.8
Feb. 2
Jan. 17
Jan. 1
NDVI
Double Cropping
Average Multi-Temporal NDVI Profiles for Southeast Kansas
1
0.9
Alfalfa
Double Crop
0.7
0.6
Summer Crop
0.5
0.4
0.3
0.2
0.1
Mapping Cool and Warm Season Grasslands
Flint Hills
Warm Season Grass
Cool Season
Grass Area
Acknowledgments
• Project Personnel
–
–
–
–
–
–
–
–
–
–
Ed Martinko
Josh Campbell
Kevin Dobbs
Steve Egbert
Mark Jakubauskas
Jude Kastens
John Lomas
Iwake Masialeti
Dana Peterson
Jerry Whistler
• Kansas GIS Policy Board
– Funding for Land Cover
Mapping Outside the
Kansas River Basin
– 2005 Landsat imagery
database
• USGS AmericaView /
KansasView Program
– Imagery discount and
research support
Download