Automating Multi-Directional Oblique Weighted Hillshade in QGIS with the Graphical Modeler

1. Background

Hill shading mimics how light interacts with a digital elevation model (DEM) from a hypothetical source, creating a greyscale image that enhances the perception of landforms (Horn, 1981; Yoeli, 1965). It produces a visual effect that helps improve understanding of terrain features. In many GIS projects, hillshade is the first product created from a DEM and often serves as the basis for subsequent cartographic work. Numerous derivatives, such as slope and curvature, are routinely derived from DEMs for geomorphometric analysis (Zevenbergen & Thorne, 1987). Standard hillshade, however, uses a single illumination azimuth (usually from the northwest) and altitude. This can cause directional bias, where slopes facing away from the light source become excessively dark, while illuminated slopes may lose micro-topographic detail. Multi-directional techniques address this by combining multiple hillshades generated from different azimuths. The MDOW approach uses oblique light directions and assigns weights to each, creating a more balanced, visually rich representation of terrain. Single-direction hillshades often overemphasize slopes facing away from the light source, causing noticeable directional artifacts. Multi-directional and oblique-weighted hillshades, like the Multi-Directional Oblique Weighted (MDOW) method originally popularized in the ArcGIS Terrain Tools toolbox, mitigate this issue by integrating hillshades from multiple azimuths (ESRI, 2015; Jenness, 2015). Recent work has also investigated alternative visualization methods that enhance or substitute traditional hillshade, especially for bathymetric and coastal data (Šolín et al., 2023). QGIS provides tools for MDOW hillshade, hillshade calculation, and raster algebra, but doing each step manually can be tedious and error-prone, especially when working with multiple DEM tiles. The QGIS Graphical Modeler allows users to connect algorithms into a single, custom workflow, enabling automated and reproducible creation of MDOW hillshades.

2. Data and software

The workflow is designed to be data-agnostic and can be applied to any single-band elevation raster with appropriate units and resolution. Typical inputs include national DEMs, LiDAR-derived surfaces, or SRTM-based products.

For this method, we used:

• Software: QGIS 3.44.3 with the Processing framework enabled.
• Input: DEM rasters in a projected coordinate system suitable for terrain analysis.
• Processing tools:
  • Raster terrain analysis → Hillshade (QGIS or GDAL implementation)
  • Raster → Raster calculator (GDAL)

The same steps can be adapted to other coordinate systems and DEM resolutions by adjusting the z-factor or scale parameters in the hillshade algorithm. The Graphical modeler in QGIS can be found by going to processing > Model Designer, and the Model Designer window should open as shown in Figure 1.

Figure 1: QGIS Model Designer Interface

METHODS

3.1 Concept of MDOW hillshade

MDOW hillshading is produced by:

1. Computing several hillshade rasters from the same DEM, each with a different azimuth but similar altitude.
2. Combining these individual hillshades into a composite raster using a weighted arithmetic expression.

Illumination azimuths are chosen to provide complementary perspectives, typically spanning the northwest, west, southwest, and north directions. In this scenario, azimuths of 360°, 315°, 270°, and 225° were used to emphasise oblique lighting while avoiding the strong directional artefacts of a single light source.

Weights can be uniform (simple average) or adjusted to emphasize certain directions; the decision depends on cartographic preference and the landscape being depicted.

3.2 Building the model in QGIS

All steps are implemented inside the QGIS Graphical Modeler (Processing → Model Designer). The main components are an input parameter, four hillshade processes, and a final raster-calculator step.

3.2.1 Define the model input

1. Open Model Designer and create a new model.
2. Under Inputs, add a Raster Layer parameter.
3. Name it DEM and provide a descriptive label such as “Input DEM for MDOW hillshade”.
4. This parameter becomes the single user-supplied input when the model is executed.

Figure 2 shows the Model Designer with an input parameter

3.2.2 Add multiple hillshade algorithms

For each illumination direction, add a hillshade algorithm as shown in Figure 3:

1. From Algorithms, drag Hillshade (under Raster terrain analysis) into the model canvas.
2. Set the Elevation layer to the DEM input.
3. Specify an azimuth and altitude value consistent with your standard.
4. Leave the z-factor and scale at defaults unless your DEM is in degrees or has non-metric elevation units.
5. Name the output.

Repeat this step three more times, creating hillshades with azimuths 315°, 270° and 225° and giving each a distinct output name (e.g., hls_315, hls_270, hls_225). The resulting model canvas contains the DEM input branching into four hillshade nodes, mirroring the graphical layout shown in the original tutorial.

Figure 3: Snapshot of model designer after creating all four hillshades

3.2.3 Combine hillshades with the Raster Calculator

1. From Algorithms, add Raster calculator (GDAL) to the model.
2. In the Raster calculator dialogue, add each hillshade output as a raster layer.
3. Define an expression that combines the four hillshades into a single MDOW raster. A simple example with equal weighting is:

MDOW = (hls_360 + hls_315 + hls_270 + hls_225) / 4

where each term corresponds to the associated hillshade band. Practitioners can assign different weights, like giving slightly higher weight to 315° and 270° azimuths, to highlight a preferred lighting effect. The main idea is that all weighting is incorporated into the model, ensuring transparent and repeatable results.

4. Set the output layer name, choose an appropriate data type (usually Float32 or Int16), and specify the output.
5. Connect the four hillshade outputs to the raster calculator node in the model canvas. The final model includes a single DEM input, four parallel hillshade branches, and a raster calculator node, as shown in Figure 4, which produces the MDOW hillshade in Figure 5.

Figure 4 Graphical modeler for MDOW.

Figure 5 MDOW.