Using ArcMap to Extract Shorelines from Landsat TM & ETM+ Data or “Remote Sensing for the Masses” Thirty‐second ESRI International Users Conference Proceedings, San Diego, CA
Richard C. Daniels, GISP Office of Information Technology Washington State Department of Transportation P.O. Box 47430 Olympia, WA 98504‐7430 Phone: (360) 705‐7654 FAX: (360) 705‐6817
[email protected]
Problem: Remote Sensing (RS) data is not being used by the GIS community to its full extent to support data creation. Why not? • Remote Sensing software & Data seen as to expensive, • Large software packages are hard to maintain and are only needed occasionally, and • Converting raster data products to vector information is difficult. ArcMap 10’s newly added Image Window and raster analysis tools now gives the GIS User ready access to RS tools that have been “known” for 30 years, but were not readily available to the GIS user till now.
Goal To develop a rapid low cost methodology for extracting shorelines and quantifying regional shoreline change using Landsat TM and ETM+ data in ArcGIS. The research tasks conducted were: 1. 2. 3. 4. 5.
Identify the best Landsat band/transform combination to use for shoreline extraction; Develop a Landsat Toolbox to automate the image classification and shoreline extraction process; Create ocean shorelines for the years 1989, 1995, 1999, 2010, 2011, 2012 from Landsat imagery; Calculate change rates for the period 1995 to 1999 and compare them to rates calculated from shorelines previously digitized from orthophotography for the same period; and Conduct a shoreline change rate analysis at 1 km intervals along a 102 km long section of coast
A little bit about Landsat ‐ Landsat Thematic Mapper (TM) is a multispectral scanning radiometer that was carried on board Landsats 4 and 5. The TM sensors have provided nearly continuous coverage from July 1982 to present. The Landsat Enhanced Thematic Mapper (ETM+) was introduced with Landsat 7. TM and ETM+ data cover the visible, near‐infrared, shortwave, and thermal infrared spectral bands of the electromagnetic spectrum. The Landsat Project is a joint initiative of the U.S. Geological Survey (USGS) and the National Aeronautics and Space Administration (NASA). For more information visit: http://landsat.gsfc.nasa.gov/
Bands/Transforms Evaluated Combination Name
Combination
Suitable for Shoreline Extraction?
a. Band Slicing ‐ NDVI
NDVI
No, breaking waves
b. Band Slicing – Band 5
Band 5 (Infrared)
No, breaking waves
c. 4‐Band Method
Band 1‐3, Band 6 (resampled)
No, Time Intensive, waves
d. 3‐Band and NDVI
Band 1‐3, NDVI
No, breaking waves
e. Tasseled Cap
Brightness, Greenness, Wetness, uses bands 1‐7
Yes, difficult to differentiate clouds
f. Tasseled Cap, NDVI
Brightness, Greenness, Wetness, NDVI
Yes
Comparison of the classification results for Point Disappointment, Washington
Comparison of Classification Results between Years
Landsat Band Selection Best bands to use are actually transforms. Transforms both normalize the data and reduce variability between years. This simplifies interpretation of the results of the land cover classification. The Tasseled Cap brightness, greenness, wetness, and Normalized Difference Vegetation Index (NDVI) were selected for used in this study.
Data Processing – Landsat Toolbox Process Step
Process Name
Automated?
0
Download and Extract Data
Partial, from USGS EROS Data Center
1
Clip Multiple Rasters to AOI
Yes
2
Fix Landsat 7 ETM+ Scanline Errors
Yes
3
Landsat TM/ETM+ Tasseled Cap
Yes
4
Normalized Difference Vegetation Index (NDVI)
Yes
5
Category Creation for Land and Sea
Yes
6
Classify Land and Sea
Partial, manual review required
7
Create Shore Boundary
Yes
8
Manual Shoreline Review
No, manual editing required
0. USGS – Global Visualization Viewer & Data Ordering
1. Clip Multiple Rasters to Area of Interest
2. Fix Landsat 7 ETM+ Scanline Errors
3. Calculate Landsat TM/ETM+ Tasseled Cap
4. Calculate Normalized Difference Vegetation Index (NDVI)
5. Category Creation (10 classes)
6. Reclassify into Land and Sea
7. Create Shore Boundary
8. Manually Review and Edit Shoreline
Shoreline Creation Shoreline creation and extraction following the eight step process
Landsat Scene Selection Approximate Tide Level at Over flight Toke Point, Time WA*
Last High Last Tide High Time Tide
Lower Low Tide Lower Time Low Tide
Sensor
Scene Path/Row
Date
ETM+
047, 028
12‐May‐2012
9:44
1.41
5:24
6.85
12:06
TM
047, 028
25‐Oct‐2011
9:34
7.43
11:06
9.06
5:06
0.17 flooding
ETM+
047, 028
7‐May‐2010
9:44
4.16
7:18
5.48
13:48
0.81 ebbing
TM
047, 028
15‐Apr‐1999
9:34
4.04
12:36
7.93
6:18
‐1.27 flooding
TM
047, 028
22‐May‐1995
9:34
4.97
7:12
6.8
13:30
0.79 ebbing
TM
047, 028
10‐Sep‐1989
9:34
5.9
9:30
5.91
14:18
High Tide 4.23 ebbing
Tide Stage
‐0.65 ebbing
*The maximum difference between tide elevations in all scenes is 6.02 ft (2 m). On beaches with 2‐3% slope this would translate into a 30‐50 m offset in shoreline position due to tide elevation. This would need to be addressed or controlled for in annual or seasonal analyses.
Change Rate Calculation: Digital Shoreline Analysis System (DSAS) An ArcGIS extension for calculating shoreline change: U.S. Geological Survey The DSAS extension to ESRI’s ArcGIS v.10 was developed by USGS to enable a user to calculate shoreline rate‐of‐change statistics from multiple historic shoreline positions. The extensions requires a baseline for measurement generates transects at user defined intervals. Change rates are then calculated for each transect. DSAS provides 4 options for calculating shoreline rates : (1) End point rate (EPR), (2) linear regression, (3) weighted linear regression, and (4) least median of squares. The standard error, correlation coefficient, and confidence intervals are also computed for the simple and weighted linear‐regression methods. The results of all rate calculations are output to a table that can be linked to the transect file by a common attribute field for display. For more information about DSAS U.S. Geological Survey Open File Report 2008‐1278. Available online at http://woodshole.er.usgs.gov/project‐pages/DSAS/version4/.
Summary Statistics ‐1995 to 1999 Shoreline distances calculated with Landsat TM derived shorelines compared to shorelines derived from orthorectified air photography previously collected by the USGS Sponsored Southwest Washington Coastal Erosion Study Data Set
Horizontal Accuracy (m)
Within Pair Difference (m)
Between Pair Difference (m)
Landsat 5 TM 6/22/1995 1:12,000 Scale Othophotography 9/22/1995
30
Mean= 41.96 Minimum= 0.76 Maximum= 245.04 Corrected Max= 131.97
Landsat 1995 to 1999 Mean= 9.34 Minimum= -227.31 Maximum= 667.04
Mean= 34.70 Minimum= 2.56 Maximum= 94.73
Air Photo 1995 to 1999 Mean= -0.25 Minimum= -250.77 Maximum= 610.72
Landsat 5 TM 4/15/1999 1:24,000 Scale Othophotography 5/26/1999
28
30 23
Correlation Results – 1995 to 1999 The relatively high correlation indicates that when the standard error and horizontal accuracy of the data sources are considered, that Landsat derived shorelines can be used to obtain valid shoreline change rates and that these rates are comparable to those derived from air photography.
Decadal Change Analysis
Annual Analysis
Annual‐Seasonal Analysis
Conclusion 1. This study successfully demonstrated a low cost methodology for extracting shorelines from Landsat TM and ETM+ data using ArcGIS. 2. Change rates were calculated using these shorelines in DSAS. The results correlate well (R2 of 0.79) with change rates previously derived for the same period from shorelines manually digitized from orthorectified air photography. 3. The near‐continues 29 year record of TM and ETM+ data makes this a rich dataset for both decadal, annual, and seasonal change analysis.
