ASTER DEM Differencing and Image Analysis - Appendix

DEM differencing and image analysis was also conducted using The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data for Santiaguito.
ASTER has a high spatial and spectral resolution with 4 VNIR (15m), 6 SWIR (30m), and 5 TIR (90m) bands. Information on ASTER can be found here.  ASTER is unique in that the satellite has a backward looking band (3B) which allows for stereoscopic viewing capabilities for creating DEMs.  The ASTER DEM analysis resulted in similar elevation changes both inside and outside of the volcanic zone, so they were not used in this research.  Many necessary insights were gained through this analysis which is described below.  In addition, I will discuss how ASTER DEMs were created and analyzed in this study.  

ASTER analysis was conducted for Santiaguito and five images were found that were cloud free.  Of those five, three were nearly the same date:  Jan 4, Jan. 20, and Feb 12, 2005, while the others were obtained on Dec. 21, 2002 and March 13, 2004.  The December and March images had a plume, but much of the plume was north of the dome and was masked out. Again, due to the lack of GCP’s, relative DEMs were created by the ASTER 3N(Nadir) and 3B(Backward-looking) bands using an automated stereo-correlation method, creating 15 meter resolution DEMs, for Dec. 21, 2002, March 13, 2004 and Jan. 4, 2005 using SILCAST 1.0 developed by Sensor Information Laboratory Corp (SILC).  SILCAST 1.0 was available through the Michigan Tech Research Institute and the DEMs were created by the staff and sent to MTU.  The DEMs created from SILCAST, were relative, meaning no ground control points were used.

Note:  ENVI has a DEM extraction tool which was initially used for this study.  This tool takes the user through the process of choosing common points on the backward (3B) and forward (3N) looking bands and then creates a DEM.  This tool created poor DEMs and was not used in this study. In addition I tried using ERDAS Imagine to create DEMs using stereo pairs and poor DEMs also resulted. The DEMs created through SILCAST were the best quality of ASTER DEMs I have seen. The quality of ASTER DEMs is going to be poor when compared to aerial derived methods, but in this case I wanted to analyze changes from 2000-present and these were the only DEMs available.  The SILCAST processing used only the entire image was used to create the DEM, where masked regions were used in ENVI.    

Once the DEMs were created, the Jan. 4, 2005 and Dec. 21, 2002 DEMs were used for DEM differencing.  The images are shown below with the 3N band for each date and the corresponding hillshade (derived from the DEM).  The difference in quality to the aerial derived DEMs is clear, as shown here (hillshade of Santiaguito).  The ASTER derived DEMs result in reduced resolution and appear to be grainy.  I used the DEM differencing procedures to create the elevation change which is overalyed on the Jan. 4, 2005 image.

ASTER image, hillshade, and difference
The area which is masked out contains the plume from the Dec. 2002 image. Although some changes are noticeable within the dome region, many of these changes are similar to the surrounding areas. Ultimately, the volume change analysis between ASTER DEMs for 2002, 2004, and 2005 were not significant due to regions outside of the volcanic zone having similar elevation changes to regions within the volcanic zone.  No significant changes were found within the volcanic zone. This is not surprising due to the short temporal scale of DEMS (2002 through 2005) and the 15 m resolution of the DEMs.  Significant activity would have to occur during this time period for elevation changes to be present.  

In addition to the above analysis, the three ASTER generated DEMs were analyzed with the aerial derived 2001 DEM to obtain elevation and volume change. The three ASTER DEMs were shifted in the x and y directions to align with the 2001 DEM using the UTM Zone 15 WGS 1986 coordinate system.  Elevation change for the 2001 DEM and the three ASTER DEMs are shown below, revealing changes on the Caliente dome.  

ASTER difference at Caliente from 2001 to 2002, 2004, and 2005

Also shown below are the elevation differences outside of the Caliente region:

Aster differences outside of Caliente
The clouds and plumes are masked out in each image.  The elevation changes inside and outside of the volcanic regions were very similar.   The mean elevation was lower for regions outside, but not significantly. Due to similar elevation changes inside and outside the volcanic area and the higher error for ASTER DEMs, vertical error of 10 -30 meters (Lang et al. 1999) the ASTER DEMs were not used for this analysis. For 2001-2005 a total volume change on Caliente was 0.19 km3 (1.49 m3 s-1), with a possible volume error estimate of 0.53 – 1.6 km3.  Using the error range for ASTER DEMs, errors were much larger than the calculated volume change and were therefore not used.
Although the ASTER DEM analysis was not successful, ASTER images were still analyzed and visible changes at the volcano are shown below:

ASTER Visible Bands

 The images acquired on Dec. 21, 2002, March 13, 2004, and February 12, 2005 are shown above with the visible, near-IR, and thermal IR bands. Changes on the Caliente dome are seen within the highlighted boxes. Activity during this period was consistent with previous activity including explosions, ash emissions, blocky lava flows, pyroclastic flows, and lahars (BGVN 2003, 2004, 2005).  From the second row of imagery, deposition from a blocky lava flow is shown, while row 3 shows a change in channel morphology.  Lastly the thermal IR band is shown, revealing activity in each of the images.

Conclusions
Creating ASTER DEMs are going to yield somewhat grainy results, when compared with those from aerial derived methods.  If ground control points exist for a study site the ASTER DEMs should greatly improve and have decreased error.  Unfortunately for this study, there were few images available over a large temporal scale and no ground control points were available. Ultimately ASTER imagery was used to chronicle changes happening at the volcano, which can help with further analysis.

I have created many references related to ASTER and ASTER DEMs which can be found here.

Copyright information for ASTER images is located under the JPL Image Use Policy located here.