Earth Explorer DEM Tips: Best Practices for GIS and Terrain AnalysisDigital Elevation Models (DEMs) are foundational datasets for GIS, hydrology, geomorphology, remote sensing, and many engineering applications. The USGS Earth Explorer portal is one of the most widely used sources for free DEM data, offering multiple products at different resolutions and coverage. This article collects practical tips and best practices for finding, downloading, preparing, and using DEMs from Earth Explorer, plus guidance on common pitfalls and recommended workflows for terrain analysis.
1. Choose the right DEM product and resolution
Selecting an appropriate DEM depends on the scale of your analysis and the landscape complexity.
- 30 m SRTM (Shuttle Radar Topography Mission): Good for regional studies, watershed delineation, and broad terrain mapping. Widely available globally but may contain voids and stair-step artifacts in steep terrain.
- 10 m or 1 m natl datasets / LiDAR-derived DEMs: Use high-resolution LiDAR (where available) for detailed engineering, urban modeling, floodplain mapping, and fine-scale morphological studies.
- ASTER GDEM: Global coverage with 30 m nominal resolution. Useful when SRTM has issues, but ASTER can have more noise and artifacts.
- USGS 3D Elevation Program (3DEP): High-quality US national DEMs at 10 m, 1 m, and other resolutions—preferred for U.S.-based work when available.
Tip: Match DEM resolution to the mapping unit and analysis—avoid using very high-resolution DEMs for broad-scale studies (unnecessary storage/processing) and avoid coarse DEMs for local engineering tasks (inaccurate slopes, volumes).
2. Use correct projection and vertical datum
DEMs are delivered in geographic or projected coordinate systems and referenced to specific vertical datums (e.g., NAVD88, EGM96).
- Confirm horizontal CRS (often WGS84) and reproject to a local projected CRS (e.g., UTM, State Plane) before performing distance, area, or slope calculations to avoid distortions.
- Confirm vertical datum and units (meters or feet). If combining multiple DEM tiles or other elevation sources, ensure they share the same vertical datum or apply datum conversions (geoid models as needed).
Tip: Use EPSG codes to avoid ambiguity (e.g., EPSG:4326 for WGS84 geographic, EPSG:32633 for UTM zone 33N). For vertical transformations, use PROJ or GDAL tools that support geoid grids where required.
3. Preprocess DEMs: fill sinks, remove artifacts, and mosaic tiles carefully
Raw DEMs often need cleanup before analysis.
- Fill voids and sinks: Use hydrologically sound sink-filling algorithms (e.g., Planchon–Darboux or Wang & Liu) to remove spurious depressions that break flow routing.
- Remove stripe or banding artifacts: Some satellite DEMs (e.g., SRTM, ASTER) may have striping; apply smoothing or destriping filters while preserving terrain features.
- Edge matching and mosaicking: When combining tiles, ensure overlapping areas are blended (feathering) or stitched to avoid seams. Use consistent resampling methods (cubic or bilinear for continuous elevation; nearest for categorical).
- Resampling: When changing resolution, choose an appropriate method—bilinear or cubic convolution for smooth elevation fields. For derived products like slope/aspect, resample before computing or compute at native resolution then aggregate appropriately.
Tip: Keep original DEMs and document preprocessing steps. Use a metadata file describing transformations, datums, and algorithms applied.
4. Use appropriate tools and software
Common tools for DEM handling and analysis:
- GDAL (gdalwarp, gdal_fillnodata, gdal_merge): Command-line utilities for reprojection, resampling, mosaicking, and filling nodata.
- WhiteboxTools and TauDEM: Robust hydrologic preprocessing and flow-routing tools with advanced algorithms for sink filling and stream delineation.
- QGIS & SAGA GIS: User-friendly GUI tools for visualization and many terrain analyses (slope, aspect, hillshade, watershed).
- ArcGIS Pro: Comprehensive toolbox for DEM management, 3D Analyst functions, and hydrologic modeling.
- Python (rasterio, pyproj, richdem, xarray): For scripting repeatable workflows and handling large datasets.
Tip: Use command-line or scripted tools for repeatability and batch processing, especially when dealing with many tiles or periodic updates.
5. Compute derivatives carefully: slope, aspect, hillshade, curvature
Derived layers are sensitive to noise and resolution.
- Slope and aspect: Compute in projected units; slope in degrees or percent depends on your application. High-resolution DEMs produce more detailed slope patterns — consider smoothing if too noisy for your objectives.
- Hillshade: Use multiple azimuths and altitudes to reveal different terrain features. Combine multi-directional hillshades for clearer visualization.
- Topographic Position Index (TPI) and curvature: Useful for landform classification and geomorphology; results change with analysis window size—choose a scale that matches the feature size of interest.
- Viewshed and visibility: Compute in projected coordinates and account for observer height and earth curvature for large areas.
Tip: Document the window sizes, units, and algorithms used for derivatives; include these in metadata.
6. Hydrologic conditioning and stream network extraction
DEM conditioning ensures realistic flow routing.
- Burn-in streams: For accurate channel routing, “burn” vector stream networks into the DEM (lower elevation along known channels) when raw DEMs do not capture channel depth.
- Depression filling vs. carving: Decide whether to fill depressions (common for generalized watershed delineation) or carve channels/burn streams (better for preserving real depressions like lakes).
- Flow direction algorithms: D8 is simple and efficient but may be less accurate in complex terrain. MFD (Multiple Flow Direction) methods model flow dispersion better for soil erosion and sediment transport studies.
- Stream threshold selection: Choose threshold area or accumulated flow values carefully when defining stream networks; validate against known channels.
Tip: Validate extracted networks with high-resolution imagery or field data where possible.
7. Error, uncertainty, and validation
Every DEM has errors—quantify and account for them.
- Sources of error: sensor noise, voids, vegetation/urban canopy, processing artifacts, datum mismatches.
- Validate with ground control points (GCPs), GPS survey points, or LiDAR where available. Compute RMSE and bias statistics to understand vertical accuracy.
- Vegetation and buildings: For “bare-earth” analyses you need DEMs processed to remove canopy/structures (often available as Digital Terrain Models, DTM). Otherwise, you’ll be working with Digital Surface Models (DSM).
- Propagate uncertainty into downstream models (e.g., hydrologic routing, slope stability) by sensitivity analyses or Monte Carlo simulations.
Tip: Report vertical accuracy and limitations in any deliverable or publication.
8. Storage, performance, and large-area workflows
Working with high-resolution DEMs at large extents is storage- and compute-intensive.
- Use cloud-optimized formats: Cloud Optimized GeoTIFF (COG) and cloud-native data stores (Zarr, COGs in S3) allow efficient access and tiling for large datasets.
- Tile and pyramid: Build overviews (pyramids) to speed visualization; process in chunks for memory-constrained environments.
- Use spatial indexing and efficient libraries (rasterio, xarray, Dask) to parallelize processing for very large mosaics.
- Compression: Use lossless compression (DEFLATE, LZW) for GeoTIFFs to reduce storage without losing precision.
Tip: For repeated large processes, consider cloud compute (AWS, GCP) where datasets may already reside to avoid large downloads.
9. Common pitfalls and how to avoid them
- Mixing vertical datums: Always check and harmonize vertical references—mistakes here can create meters-level errors.
- Using geographic CRS for distance-based calculations: Reproject to an appropriate projected CRS first.
- Blind reuse of derivatives: Slope/aspect calculated from noisy DEMs will propagate errors into models—preprocess and validate.
- Ignoring metadata: DEM product metadata often includes known caveats (voids, acquisition date, processing level). Read it.
Tip: Create a short checklist for each project: product chosen, CRS/datum, preprocessing steps, derivatives computed, validation methods.
10. Example end-to-end workflow (concise)
- Search Earth Explorer and download DEM tiles covering your AOI; choose the best available product (e.g., 3DEP LiDAR if in the U.S.).
- Inspect metadata for CRS, vertical datum, resolution, and known issues.
- Reproject to a suitable projected CRS if needed.
- Mosaic tiles with edge blending; fill voids or apply destriping where necessary.
- Hydrologically condition (fill sinks or burn streams) depending on objectives.
- Compute derivatives (slope, aspect, hillshade) at the appropriate scale; smooth if required.
- Validate elevations and derived products against ground control or higher-quality data.
- Package outputs with metadata and quality notes.
11. Additional resources and learning
- GDAL, PROJ, and EPSG documentation for reprojection and datum transformations.
- WhiteboxTools and TauDEM manuals for hydrologic preprocessing.
- QGIS and ArcGIS tutorials for DEM derivatives and visualization.
- Peer-reviewed literature on DEM error propagation and hydrologic modeling.
Final practical note: Always preserve the original downloaded DEMs and keep detailed records of preprocessing and analysis steps. That reproducibility and transparency are as important as the technical choices themselves.
Leave a Reply