2006 ASTER DEM2000 SRTM DEM

Samuel Blanchard Clark University, Worcester, MA Faculty Advisors: Dr. John Rogan & Dr. Deborah Woodcock

DataA February 2000 SRTM DEM and October 2006 ASTER DEM were acquired forgeomorphic change analysis. An SRTM DEM rather than a ASTER DEM wasacquired for 2000 due to the unavailability of ASTER data for the time period andstudy area. ASTER spectral imagery for August 2001 and July 2006 were acquired aswell as a USGS topographic map, a Massachusetts state surficial geology map, aland-cover map, and orthophotos. Field data were collected for 78 drumlin locationswithin the study area recording both quantitative hypsometer data and qualitativedata.

MethodsDigital Elevation Model Accuracy Assessment:The SRTM and ASTER DEM datasets and topographic map were systematicallysampled for elevation differences in areas experiencing no change in landcomposition over the 6 year time period. A Normalized Difference Vegetation Index(NDVI) change image, produced from the ASTER spectral imagery, and the four mostdominant land-cover types were used to stratify the landscape. 200 points (50 pointsper land-cover type) were generated along the topographic map gradients. Thedifferences in elevation values between the datasets were used to determine RootMean Square Error (RMSE) and Mean Average Error (MAE).

Results

Study Area

Data Temporal Resolution Spatial Resolution *Vertical Accuracy *Horizontal AccuracySRTM DEM February 2000 30 m 6 - 16 m 20 mASTER DEM October 2006 30 m 12 - 25 m 10 – 20 mASTER Spectral Imagery August 2001 15 m -- -- -- 15 mASTER Spectral Imagery July 2006 15 m -- -- -- 15 mUSGS Topographic Map 2003 3 m 1.5 m -- -- --Surficial Geology 2007 -- -- -- -- -- -- 12 mLand-cover Map 2000 30 m -- -- -- -- -- --Orthophotos April 2005 0.5 m -- -- -- 3 mField Data 2007 to 2009 -- -- -- -- -- -- -- -- --

Geomorphic Change Analysis using ASTER and SRTM Digital Elevation Models in Central Massachusetts, USA

Data Specifications:

* Product specification and assessed accuracy

Geomorphic Change Analysis:Using data collected in the field, a stratified sample of 9 drumlins were selected forgeomorphic change detection (6 altered and 3 unaltered) based on date and extent ofalteration. The 2006 ASTER DEM was subtracted from the2000 SRTM DEM to create a geomorphic change imagecontaining only the 9 specified drumlin spatial extentsdetermined from the geology map.

Geomorphic Change Analysis- Map Validation:The results of the geomorphic change detection(type and spatial extent of change) were verifiedfor accuracy through a combination of ground fielddata and orthophotos.

SRTM-ASTER SRTM-USGS ASTER-USGSMAE 9.78 m 2.64 m 9.50 mRMSE 13.19 m 3.46 m 13.04 m

Elevation Variation and Land-cover Characteristics:

MAE and RMSE Statistics: SRTM, ASTER, and USGS Topographic Map

30

50

70

90

110

130

150

170

SRTM ASTER USGS

Data Set

Elev

atio

n (m

)

Forest

30

50

70

90

110

130

150

170

SRTM ASTER USGS

Developed

Data Set

Elev

atio

n (m

)

30

50

70

90

110

130

150

170

SRTM ASTER USGS

Agriculture & Grassland

30

50

70

90

110

130

150

170

SRTM ASTER USGS

Bare Soil

Data Set

Elev

atio

n (m

)

Data Set

Elev

atio

n (m

)

Geomorphic Change Analysis:

Sponsors:Henry David Thoreau Foundation

ASTER Spectral Imagery 2001

ASTER Spectral Imagery 2006NDVI 2006

NDVI 2001

2001-2006 NDVI Change

No change in land composition

Land-cover

USGS Topographic Map

Elevation Difference and Error

Suitable Sample Areas

Random Point Generation

Contour Line Locations

No change and Dominant

Land-cover

Dominant Land-cover

Field Data6 Altered

3 Unaltered

Geology MapSelected Drumlins for

Analysis

2000 SRTM DEM

2006 ASTER DEM

Geomorphic Change

Field Data

Results

Geomorphic Change

Orthophotos

Geomorphic change analysis examines the quantity and causation of landformalteration. Previous analyses of geomorphic change have relied on fine (<20 m)spatial resolution remotely sensed data or in situ data. Despite their noted accuracyand precision, these traditional approaches are limited in spatial extent and temporalfrequency (Smith et al. 2006). Within the past three decades, satellite- sensed datahave begun to supplant traditional methods of geomorphic change assessmentenabling change analysis over large, inaccessible areas with predictable repetition.

Digital elevation models (DEMs) are the most widely used data to derive details onsurface terrain such as volume, mass, slope, and aspect. Fine spatial resolutionairborne data are the most commonly used source of information in DEM generation(Schiefer and Gilbert 2007). Coarse spatial resolution (>20 m) DEMs from spacebornesensors such as Advanced Spaceborne Thermal Emission and Reflection Radiometer(ASTER) and Shuttle Radar Topography Mission (SRTM) are less common in changedetection studies owing to relatively coarse spatial resolution (Stevens et al. 2004).

The geomorphic change literature has primarily focused on assessing changeresulting from natural processes and not from anthropogenic processes (Toutin 2008).However, landform change resulting from human activities has greatly outpacedchange from natural processes and is a major component in sediment movement andsurface alterations (Rivas et al. 2006). Human induced geomorphic change has onlybeen examined indirectly through land use effects on sediment transportation,excavation, and erosion but remotely sensed data in the form of coarse spatialresolution DEMs have not been extensively utilized to assess human geomorphicchange over time.

Southern New England, in particular central Massachusetts, is currently undergoingtremendous landscape changes resulting from increased urbanization. The state’srigorous environmental protection laws protecting low lands from development hasmade uplands a target for new development. The rate of geomorphic changeoccurring from this demand makes the region an ideal study area to examine theutility of coarse resolution DEMs in anthropogenic geomorphic change analysis(Woodcock et al. In Progress).

The objective of this study is to examine the accuracy of ASTER based DEMs, incombination with the SRTM DEM, compared to a ground based reference mapand to use these DEMs to map anthropogenic geomorphic change in a 135 km²area in Central Massachusetts, USA, over a 6 year period (2000-2006).

Introduction & Background

ReferencesRivas, V., A. Cendrero, M. Hurtado, M. Cabral, J. Gimenez, L. Forte, L. del Rio, M. Cantu, and A. Becker. (2006) Geomorphic consequences of urban development and mining activities;

an analysis of study areas in Spain and Argentina. Geomorphology. 73: 185-206Schiefer, E. and R. Gilbert. (2007) Reconstructing morphometric change in a proglacial landscape using historical aerial photography and automated DEM generation. Geomorphology. 88:

167-178Smith, M. J., J. Rose, and S. Booth. (2006) Geomorphological mapping of glacial landforms from remotely sensed data: An evaluation of the principal data sources and an assessment of

their quality. Geomorphology. 76: 148- 165Toutin, T. (2008) ASTER DEMs for geomatic and geoscientific applications: a review. International Journal of Remote Sensing. 29 (7): 1855-1875Woodcock, D. W., J. Rogan, and S. D. Blanchard. (In Progress). Humans as an Agent of Geomorphological Change: Drumlin Alteration and Loss in Central Massachusetts. Unpublished.

Study Area: Massachusetts, USA

Alte

red

Dru

mlin

Alte

red

Dru

mlin

Una

ltere

d D

rum

lin

Discussion & ConclusionPreliminary results indicate DEM errors are sufficiently low to provide an accurateassessment of landform alterations. The SRTM DEM was more accurate than missionspecified (3.46 m RMSE compared to 16 m RMSE {Global scale} and 6 m RMSE{Local scale}) and the ASTER DEM was more accurate than mission specified (13 mRMSE compared to 25 m RMSE). The level of agreement between the SRTM andASTER DEMs was also reasonable at 13.9 m RMSE. The low variability in elevationvalues over the four land-cover types indicates these data can accurately representcomplex landscapes in the study area with minimal error.

The ASTER DEM in combination with the SRTM DEM successfully detectedgeomorphic change on known, altered drumlin features. The 0 to 44 m elevation lossand the spatial extent and alteration characteristics were consistent with fieldmeasurements. The accuracy of the ASTER On-Demand Level 3 product to detectfine scale geomorphic changes resulting from anthropogenic alteration anddevelopment provides another tool for monitoring human interactions with theenvironment. ASTER DEMs, as well as other coarse spatial resolution DEM- capablespaceborne sensors, offer accurate data sources with the advantages of low cost,large extent, and predictable temporal frequency, as alternatives to traditional data inanthropogenic geomorphic change analyses.