Skip to main content

​
Overview

The gdal_baseline_slope.py script calculates specialized slope values from Digital Elevation Models (DEMs) using variable baseline lengths. Unlike standard slope algorithms, this tool measures slope across specific pixel distances, making it particularly useful for rover mission planning where traversability depends on slope over specific ground distances.
This tool was developed for Mars rover mission planning but can be applied to any planetary body with DEM data.

​
Installation

Requires Anaconda Python environment with GDAL, NumPy, and SciPy: 
conda install gdal numpy scipy

​
Usage

python gdal_baseline_slope.py [-baseline INTEGER] [-ot Byte] [-crop] infile outfile.tif

​
Parameters

​
-baseline
integer
Baseline length in pixels (e.g., 1, 2, 5). If omitted, uses Horn’s method (standard 3x3 slope calculation).
​
-ot
string
Output data type. Set to Byte to scale 32-bit floating point slopes to 8-bit using: DN = (Slope * 5) + 0.2. A 32-bit version is still generated as 32bit_outfile.tif.
​
-crop
boolean
Trim 1-5 pixels from image edges based on selected baseline amount.
​
infile
string
required
Input DEM file (any GDAL-readable format).
​
outfile
string
required
Output slope file (GeoTIFF format).

​
Mathematical Method

The baseline slope algorithm uses a moving window that samples only the corner pixels, measuring slope across a specified baseline distance.

​
Baseline = 2 Example

Uses a 3×3 pixel window, sampling only the 4 corners: 
[a, ·, b,
 ·, ·, ·,
 c, ·, d]
Calculation: 
dz_dx = ((b + d) - (a + c)) / (2.0 * baseline * x_cellsize)
dz_dy = ((a + b) - (c + d)) / (2.0 * baseline * y_cellsize)
slope = sqrt(dz_dx² + dz_dy²)
slope_degrees = arctan(slope) * 180 / π

​
Baseline = 5 Example

Uses a 6×6 pixel window, sampling only the 4 corners: 
[a, ·, ·, ·, ·, b,
 ·, ·, ·, ·, ·, ·,
 ·, ·, ·, ·, ·, ·,
 ·, ·, ·, ·, ·, ·,
 ·, ·, ·, ·, ·, ·,
 c, ·, ·, ·, ·, d]
Same equation as above, with baseline = 5.

​
Horn’s Method (Default)

When no baseline is specified, the script uses Horn’s Method (equivalent to gdaldem slope): 
# From 3×3 window:
[a, b, c,
 d, e, f,
 g, h, i]

dz_dx = ((c + 2*f + i) - (a + 2*d + g)) / (8 * x_cellsize)
dz_dy = ((g + 2*h + i) - (a + 2*b + c)) / (8 * y_cellsize)
For more information on spatial filters, see the ISIS3 Spatial Filters Guide.

​
Examples

​
Basic Slope with Baseline = 5

python gdal_baseline_slope.py -baseline 5 input_DEM.tif slope_5m.tif
Calculates slope using a 5-pixel baseline.

​
8-bit Output with Cropping

python gdal_baseline_slope.py -baseline 2 -ot Byte -crop DEM_20m.tif slope_2m_8bit.tif
Generates:
  • 32bit_slope_2m_8bit.tif - Full 32-bit floating point slopes in degrees
  • slope_2m_8bit.tif - 8-bit scaled version (1-255)

​
HiRISE DEM Processing

python gdal_baseline_slope.py -baseline 5 -ot Byte -crop \
  HiRISE_DEM_1m.tif rover_slope_5m.tif
8-bit Scaling Formula: The Byte output uses DN = (Slope * 5) + 0.2, which deliberately maps slopes >50° to DN value 255.

​
Standard Slope (Horn’s Method)

python gdal_baseline_slope.py mars_DEM.tif standard_slope.tif
This tool loads the full image into memory and can be slow for large DEMs.

​
Output Files

Output TypeDescription
32-bit FloatSlope values in degrees (full precision)
8-bit ByteScaled slopes: DN = (Slope × 5) + 0.2, offset = -0.2, scale = 0.2
CroppedEdge pixels removed based on baseline (prevents edge artifacts)

​
Rover Mission Applications

1

Choose Baseline

Select baseline based on rover wheelbase or critical distance:
  • Baseline = 1: ~1m for small rovers
  • Baseline = 2: ~2m for MER-class rovers
  • Baseline = 5: ~5m for larger vehicles or obstacle detection
2

Process DEM

Run gdal_baseline_slope.py with appropriate baseline and output options.
3

Apply Thresholds

Use slope output to identify traversable terrain:
  • Safe: < 15°
  • Caution: 15-25°
  • Avoid: > 25° (typical rover limits)
4

Generate Products

Create visualization products with color ramps and integrate with path planning software.

​
Script Location

~/workspace/source/gdal_baseline_slope/gdal_baseline_slope.py
The repository includes helper scripts:
  • gdal_hist.py - Generate slope histograms
  • slope_histogram_cumulative_graph.py - Create cumulative slope distribution graphs
  • gdal_HiRISE_RoverSlope_crop.pl - Batch processing for HiRISE DEMs
  • gdal_CTX_RoverSlope_crop.pl - Batch processing for CTX DEMs

​
Performance Notes

Memory Usage: Current implementation loads the entire DEM into memory. For large files, ensure sufficient RAM is available.
  • Processing speed depends on DEM size and baseline length
  • Larger baselines require larger filter windows but process similar times
  • Edge pixels are set to NoData when using mode='constant'

​
References