Automatic detection of clefts and corridors within the sandstone plateau - Szczeliniec Wielki & Szczeliniec Mały mesas, Poland

Automatic detection of clefts and corridors within the sandstone plateau 1.0 Kacper Jancewicz Department of Geomorphology University of Wrocław 2024 Concept introduction INPUT DATA Airborne-LiDAR-based Digital Elevation Model (high resolution grid of at least 1 m x 1 m resolution). EXPECTED RESULT Grid –cells representing zones enclosed by at least two opposite rock walls of height exceeding certain threshold value. ASSUMPTIONS •Software environment: ArcGIS 10.8 •Analysis of four pairs of directions: N–S, NE–SW, E–W, SE–NW. •Two crucial variables: •Cleft Width Threshold value (CWT) that defines the maximum width of detected pit, cleft or corridor (measured in metres or number of cells) •Depth Threshold value (DT) that defines the minimum depth of pits, clefts or corridors (measured in metres) So… how does it work? (pt. 1) Firstly, we need to find maximum elevation within a certain neighbourhood of every DEM cell. We use Focal Statistics tool: •Neighborhood = IRREGULAR (kernel file must be prepared) •Statistics type = MAXIMUM The irregular shape of neighbourhood is defined by CWT and direction (one of eight values: 0°, 45°, 90°, 135°, 180°, 225°, 270°or 315°) –see example  The calculation is done for each of eight direction values, resulting eight grids: Hmax000, Hmax045, (...) Hmax315. Above: Example of kernel file designed for CWT = 25 cells and direction = 0°(N). Central cell marked in green, while its neigbourhood is marked in orange. How does it work? (pt. 2) Secondly, we use raster calculator and subtract values of maximum elevation grids from initial DEM: 𝐷𝐸𝑀 − 𝐻𝑚𝑎𝑥000 = Δ𝐻000 𝐷𝐸𝑀 − 𝐻𝑚𝑎𝑥045 = Δ𝐻045 𝐷𝐸𝑀 − 𝐻𝑚𝑎𝑥090 = Δ𝐻090 (...) 𝐷𝐸𝑀 − 𝐻𝑚𝑎𝑥315 = Δ𝐻315 Example of the ΔH000 grid –the Szczeliniec Wielki study site (Stołowe Mts, Poland) How does it work? (pt. 3) Thirdly, we have to define the DT value. For instance, if we search for clefts deeper than 1 meter, then DT = -1 m. Each ΔHxxx grid is reclassified as follows: <DT → 1 ≥DT → 0 As a result, we get eight grids: 000_below DT, 045_below DT, (...), 315_belowDT (see right for examples) How does it work? (pt. 4) We’re looking for landforms consisting of at least two parallel rock walls, so we sum grids representing the opposite direction values: 000_below DT + 180_below DT = sum000_180 045_below DT + 225_below DT = sum045_225 090_below DT + 270_below DT = sum090_270 135_below DT + 315_below DT = sum135_315 Results should be reclassified: 0(no walls detected) → NoData 1(one rock wall detected) →NoData 2 (two parallel walls detected) → 1 After reclassification we get four grids: clefts000_180, clefts045_225, clefts090_270, clefts135_315, which should be mosaicked to new raster Clefts_final (mosaic operator = MAXIMUM) reclassification WARNING! CWT and DT values should be defined by user individually for each studied site in order to achieve optimal results. Clefts and pits of Szczeliniec Wielki mesa (CWT = 25 cells, DT = -2 m) That’s all! Choose your study site and give the procedure a try .