Advancing Spatial Pattern Integration in Species Distribution Modelling With Computer Vision: How Grain Size and Extent Concepts Shape Predictions of Emerging Aquatic Insects

Deep SDMs for predicting Plecoptera and Trichoptera from drone and satellite imagery show performance shifts up to 0.26 AUC with grain size and up to 0.16 AUC with extent.

• Why it matters: Traditional SDMs only reflect grain size via raster-derived variables, leaving uncertainty about how both spatial resolution and window extent shape predictions for emerging aquatic insects.
• What they did: Using multispectral and LiDAR data at 31 cm to 10 m resolution and window extents of 40 m to 100 m, the study applied Deep SDMs and used Grad-CAM to quantify variable importance and spatial agreement.
• The result: Best results used 2.5 m multispectral at 40 m extent for Plecoptera and the same multispectral resolution with LiDAR at 100 m extent for Trichoptera, with variable importance changing moderately across grain sizes.

AI summary of the abstract · may contain errors · not medical advice · read the paper.