Generating a DEM for research
Generating a DEM for research
For my thesis research in archaeology, I am modeling potential settlement locations (suitability analysis) and movements of certain populations historically located in present-day Ecuador and Colombia (historically, the viceroyalties of Peru and New Grenada) based off of a variety of variables. In order for these models to produce useful results in mountainous and densely forested regions, I need a DEM/DSM with a spatial resolution of at least 10m, if not better, that can accurately capture elevations in these terrains. None of the publicly available DEMs fit these requirements and I cannot find LiDAR data on the region (there are some granules from GEDI, but I do not think they cover enough area). I have a DEM from the government of Ecuador with variable 3-5m spatial resolution, but there are significant gaps and it was made with photogrammetry so it does not capture the elevations in forested areas. After going through various satellite datasets, from my limited knowledge I should be able to generate a DEM of this area from the PALSAR L1.1 product through differential interferometry that fits the upper-end of my requirements (10m spatial resolution). I am also interested in forest/vegetation density and canopy height as well as hydrology data (locations and characteristics of river systems, and also looking into the possibility of predicting potential crossing points at certain times of the year. I don't even know if that is possible.) I will be using a variety of georeferenced historical maps to understand and possibly correct differences between the historical and contemporary landscape, by for example, interpolating forest densities across deforested areas. Will the PALSAR L1.1 product work for generating the DEM that I need? Is there a better dataset that I could use for this purpose? Any other advice or suggestions would be much appreciated. I apologize in advice if this is the wrong forum for this type of question.
Hi @gdervin,
To generate a DEM from SAR, it's best to have two images that are acquired very close together in time, but from slightly different places in space.
You could try generating DEMs from ALOS PALSAR L1.1 products, but there are a number of issues that may be problematic:
- ALOS had a repeat cycle of 46 days, and a lot of surface characteristics can change in that time, especially in areas with vegetation or strong seasonality.
- The accuracy of the orbit files are also important for DEM generation, and that may also be problematic with ALOS data.
- The most recent acquisitions were in 2011, so there may also have been topographic changes since then.
- You must make sure to account for atmospheric impacts.
If you're working in vegetated areas, L-band data is a reasonable choice, as it has a better chance of penetrating through the canopy. If you have access to L-band data other than ALOS, such as SAOCOM, that could be another option.
If you do generate a DEM using ALOS PALSAR data (or other SAR resources), it would be important to validate it against trusted sources, and perhaps compare it to the other resources you mentioned (GEDI, the DEM from the government of Ecuador) to see if it meets the needs of your project.
If your work is funded by NASA, you could also look into accessing the Maxar Precision DSM through NASA's CSDA program.
Go to full postTo generate a DEM from SAR, it's best to have two images that are acquired very close together in time, but from slightly different places in space.
You could try generating DEMs from ALOS PALSAR L1.1 products, but there are a number of issues that may be problematic:
- ALOS had a repeat cycle of 46 days, and a lot of surface characteristics can change in that time, especially in areas with vegetation or strong seasonality.
- The accuracy of the orbit files are also important for DEM generation, and that may also be problematic with ALOS data.
- The most recent acquisitions were in 2011, so there may also have been topographic changes since then.
- You must make sure to account for atmospheric impacts.
If you're working in vegetated areas, L-band data is a reasonable choice, as it has a better chance of penetrating through the canopy. If you have access to L-band data other than ALOS, such as SAOCOM, that could be another option.
If you do generate a DEM using ALOS PALSAR data (or other SAR resources), it would be important to validate it against trusted sources, and perhaps compare it to the other resources you mentioned (GEDI, the DEM from the government of Ecuador) to see if it meets the needs of your project.
If your work is funded by NASA, you could also look into accessing the Maxar Precision DSM through NASA's CSDA program.
Filters:
-
ASF - hjkristenson
- Subject Matter Expert
- Posts: 6
- Joined: Mon Aug 08, 2022 1:41 pm America/New_York
Re: Generating a DEM for research
Hi @gdervin,
To generate a DEM from SAR, it's best to have two images that are acquired very close together in time, but from slightly different places in space.
You could try generating DEMs from ALOS PALSAR L1.1 products, but there are a number of issues that may be problematic:
- ALOS had a repeat cycle of 46 days, and a lot of surface characteristics can change in that time, especially in areas with vegetation or strong seasonality.
- The accuracy of the orbit files are also important for DEM generation, and that may also be problematic with ALOS data.
- The most recent acquisitions were in 2011, so there may also have been topographic changes since then.
- You must make sure to account for atmospheric impacts.
If you're working in vegetated areas, L-band data is a reasonable choice, as it has a better chance of penetrating through the canopy. If you have access to L-band data other than ALOS, such as SAOCOM, that could be another option.
If you do generate a DEM using ALOS PALSAR data (or other SAR resources), it would be important to validate it against trusted sources, and perhaps compare it to the other resources you mentioned (GEDI, the DEM from the government of Ecuador) to see if it meets the needs of your project.
If your work is funded by NASA, you could also look into accessing the Maxar Precision DSM through NASA's CSDA program.
To generate a DEM from SAR, it's best to have two images that are acquired very close together in time, but from slightly different places in space.
You could try generating DEMs from ALOS PALSAR L1.1 products, but there are a number of issues that may be problematic:
- ALOS had a repeat cycle of 46 days, and a lot of surface characteristics can change in that time, especially in areas with vegetation or strong seasonality.
- The accuracy of the orbit files are also important for DEM generation, and that may also be problematic with ALOS data.
- The most recent acquisitions were in 2011, so there may also have been topographic changes since then.
- You must make sure to account for atmospheric impacts.
If you're working in vegetated areas, L-band data is a reasonable choice, as it has a better chance of penetrating through the canopy. If you have access to L-band data other than ALOS, such as SAOCOM, that could be another option.
If you do generate a DEM using ALOS PALSAR data (or other SAR resources), it would be important to validate it against trusted sources, and perhaps compare it to the other resources you mentioned (GEDI, the DEM from the government of Ecuador) to see if it meets the needs of your project.
If your work is funded by NASA, you could also look into accessing the Maxar Precision DSM through NASA's CSDA program.
Heidi Kristenson
Senior GIS Specialist
Alaska Satellite Facility
Senior GIS Specialist
Alaska Satellite Facility