Karst Landscape Characterization

Why it matters • What it is • Mapping lineaments & sinkholes

Why Karst Characterization?

  • ~20% of global land surface shows karst features; many aquifers are karstic.
  • Rapid, anisotropic flow → flashy recharge, contaminant vulnerability, sinkhole hazards.
  • Management needs spatially explicit maps: recharge zones, flow paths, hazard areas.
  • Goal: reproducible workflow to identify lineaments & delineate sinkholes and derive metrics (depth, volume, density).

Karst Landscape Global Karst Extent

What is “Karst Characterization”?

  • Integrated mapping of surface & subsurface indicators:
    • Sinkholes (dolines), dry valleys, uvalas, caves/springs
    • Lineaments (fractures/joints/faults) guiding groundwater flow
    • Dye Traces Identify flowpaths of water from sinks to springs, caves, streams
  • Products:
    • Inventories (points/polygons/lines), density maps, terrain derivatives
    • Suitability/recharge potential layers for decision support

Karst Dye Trace

What is “Karst Characterization”?

Sinks

Faults/Lineaments

Datasets

  • Elevation: 3DEP (10–30 m) or local LiDAR (≤1 m)
  • Imagery: NAIP/Planet/Sentinel-2 (false color helpful)
  • Bedrock/Soils: geology units, GLHYMPS, SSURGO
  • Hydrography: NHD springs/streams; field GPS
  • Optional: UAV-SfM for cm-scale morphology (lecture #2)

3DEP NHD National Geologic Map Compilations —

DEM Preprocessing

  • Reproject to target CRS (e.g., UTM 12N)
  • Fill small artifacts conservatively (avoid erasing true depressions)
  • Derive slope, aspect, curvature, TPI/TRI
  • Multi-scale analysis (window sizes tuned to expected sinkhole diameter) Terrain Attributes

:::

Notes: This is what we did (will do) in the lab exercise ::: —

Mapping Sinkholes (Raster-first)

  • Closed-depression detection
    • DEM fill vs. original → depth raster (Cut/Fill logic)
    • Threshold minimum depth & minimum area
  • Morphology filters
    • TPI/TRI, curvature (concavity), circularity metrics
  • Vectorization
    • Raster → polygons; clean with dissolve/simplify
  • Attributes
    • Depth, area, volume, elongation, circularity (4πA/P²)

Mapping Lineaments (Raster/Imagery)

DEM-based (shaded relief): - Multi-azimuth hillshades; combine (min/mean) - Edge detection (Canny), non-max suppression - Hough Transform for straight segments; merge/extend with angular tolerance

Optical imagery (optional): - Structural traces in bare or sparse vegetation - PCA/band ratios; NIR enhances lithologic contrasts

Postprocess: - Snap/merge segments, remove short/noisy lines - Compute orientation (rose diagrams), length density (km/km²)

Lineament Detection Lineament Density

Typical Pitfalls

  • Differing definitions of lineaments
  • data scale/resolution
  • Not all lineaments visible at surface
  • some lineaments are result of modern geomorphologic processes not underlying geology
  • fill technique only works with closed sinks
  • False sinks, and lineaments
  • Anthropogenic depressions (quarries, tanks) → false positives

Minimal Workflow (Concept)

  1. Acquire & preprocess DEM (reproject, gentle fill)
  2. Derive terrain attributes (slope, curvature, TPI/TRI)
  3. Sinkholes: depression depth → threshold → polygons → metrics
  4. Lineaments: multi-azimuth hillshade → edges → Hough → merge → stats
  5. Densities/heatmaps; MCDA layers for recharge/hazard
  6. Validate; tune thresholds; document parameters & uncertainty

Outputs & End-uses

  • Sinkholes: polygon inventory with depth, area, volume, circularity

Sinkhole Density Maps
  • Lineaments: polyline set with azimuth, length; density rasters

Arizona Lineaments

Arizona Lineament Density
  • Dye Trace: polyline from introduction point to detection point

Dye Trace- Flowpath Map
  • Maps: suitability/recharge potential, hazard zones

Vulnerability Map

Sankey Diagram

Map of Recharge potential

Going Further

Once a database of validated sinkholes and lineaments are created:

Utilize the database for training a Deep Learning Model to Identify Sinks

Image Segmentation

Use Terrain attributes as input to DL model.

  • increase ability to automate karst landscape Characterization
  • identify sinks that are not closed or subtle lineaments (hard for current tools)

Toward Decision Support

  • Combine layers (weights/MCDA) for Opportunistic Recharge Enhancement
  • Prioritize corridors where:
    • Sinkhole density is high and dolines are deep
    • Lineament density/orientation aligns with recharge pathways
    • Geology/soil support high permeability
  • Flag conflicts: infrastructure, contamination sources, protected areas

Suitability for Thinning to enhance Recharge in Ponderosa Pine Forests Map of Recharge potential —

Next Lecture (UAV-SfM)

  • cm-scale morphology of sinkholes

Acknowledgments & Notes

  • This deck summarizes standard karst/geomorphometry practice adapted to DEM workflows.
  • Always validate with field observations where feasible.
  • Document versions of data, parameters, and CRS for reproducibility.