Automatisk detektion av diken och vattendrag från digitala höjdmodeller med hjälp av djupinlärning Automatic Detection of Ditches and Natural Streams from Digital Elevation Models Using Deep Learning 2024-57-1 SLU.seksko.2024.4.4.IÄ-1 https://doi.org/10.5878/jrex-z325 Swedish National Data Service Svensk nationell datatjänst Landing page Automatisk detektion av diken och vattendrag från digitala höjdmodeller med hjälp av djupinlärning Automatic Detection of Ditches and Natural Streams from Digital Elevation Models Using Deep Learning 2024-57-1 SLU.seksko.2024.4.4.IÄ-1 https://doi.org/10.5878/jrex-z325 oai:DiVA.org:uu-494403 urn:nbn:se:uu:diva-494403 10.1007/s13280-022-01770-8 oai:DiVA.org:umu-201888 10.1061/jidedh.ireng-9796 urn:nbn:se:umu:diva-201888 dos Santos Toledo Busarello, Mariana dos Santos Toledo Busarello, Mariana Lidberg, William Lidberg, William Ågren, Anneli Ågren, Anneli Westphal, Florian Westphal, Florian Swedish National Data Service Svensk nationell datatjänst Landing page RIVERS/STREAMS WATER CHANNELS digital elevation model Hydrography Hydrografi ditches diken machine learning maskininlärning artificial intelligence artificiell intelligens lidar laserskanning Imagery / Base Maps / Earth Cover Arealtäckande bilder och bakgrundskartor Geoscientific Information Geovetenskap Elevation Höjddata Location Positionering Inland Waters Sjöar och vattendrag This data contains the digital elevation models and polyline shapefiles with the location of channels from the 12 study areas used in this study. It also has the code to generate the datasets used to train the deep learning models to detect channels, ditches, and streams, and calculate the topographic indices. The code to train the models is also included, along with the models with the highest performance in 0.5 m resolution. The channels were mapped differently based on their type: ditches were manually digitized based on the visual analysis of some topographic indices and orthophotos obtained from the DEM. Streams were mapped by initially detecting all natural channel heads, then tracing the downstream channels, and finally manually editing them based on orthophotos. Den här datamängden innehåller digitala höjdmodeller och polyline-shapefiler med placeringen av kanaler från de 12 studieområdena som används i denna studie. Den innehåller också koden för att generera datamängderna som används för att träna djupinlärningsmodellerna som används för att kartera kanaler, diken och vattendrag, samt beräkna topografiska index. Koden för att träna modellerna ingår också, tillsammans med modellerna med bäst prestanda i 0,5 meters upplösning. Kanalerna kartlades på olika sätt: diken digitaliserades manuellt baserat på visuell analys av vissa topografiska index och ortofoton som erhållits från en digital höjdmodell. Vattendrag kartlades genom att först upptäcka alla naturliga kanalhuvuden, sedan spåra nedströmskanalerna och slutligen redigera dem manuellt baserat på ortofoton. Sweden Sverige Numeric Text Geospatial Professionals from the Swedish Forest Agency manually digitized the ditches within the 12 study areas spread across Sweden based on the hillshade and high-pass median filter obtained from the DEM. Historical photos and current ortophotos (resolution ranging from 0.17-0.5 m), the ditches were manually digitized. Streams were mapped by initially detecting all natural channel heads, then tracing the downstream channels, and finally manually editing them based on ortophotos.Professionals from the Swedish Forest Agency manually digitized the ditches within the 12 study areas spread across Sweden based on the hillshade and high-pass median filter obtained from the DEM. Historical photos and current ortophotos (resolution ranging from 0.17-0.5 m), the ditches were manually digitized. Streams were mapped by initially detecting all natural channel heads, then tracing the downstream channels, and finally manually editing them based on ortophotos. Computer-based observationComputer-based observation Datorbaserad observationDatorbaserad observation Access to data through SND. Data are freely accessible. Åtkomst till data via SND. Data är fritt tillgängliga. openAccess Paul, S. S., Maher Hasselquist, E., Jarefjäll, A., & Ågren, A. (2023). Virtual landscape-scale restoration of altered channels helps us understand the extent of impacts to guide future ecosystem management. In Ambio (Vol. 52, Issue 1, pp. 182–194). https://doi.org/10.1007/s13280-022-01770-8 Paul, S. S., Maher Hasselquist, E., Jarefjäll, A., & Ågren, A. (2023). Virtual landscape-scale restoration of altered channels helps us understand the extent of impacts to guide future ecosystem management. In Ambio (Vol. 52, Issue 1, pp. 182–194). https://doi.org/10.1007/s13280-022-01770-8 urn:nbn:se:uu:diva-494403 10.1007/s13280-022-01770-8 oai:DiVA.org:uu-494403 2023 oai:DiVA.org:uu-494403 Lidberg, W., Paul, S. S., Westphal, F., Richter, K.-F., Lavesson, N., Melniks, R., Ivanovs, J., Ciesielski, M., Leinonen, A., & Ågren, A. M. (2023). Mapping drainage ditches in forested landscapes using deep learning and aerial laser scanning. In Journal of irrigation and drainage engineering (No. 04022051; Vol. 149, Issue 3). https://doi.org/10.1061/jidedh.ireng-9796 Lidberg, W., Paul, S. S., Westphal, F., Richter, K.-F., Lavesson, N., Melniks, R., Ivanovs, J., Ciesielski, M., Leinonen, A., & Ågren, A. M. (2023). Mapping drainage ditches in forested landscapes using deep learning and aerial laser scanning. In Journal of irrigation and drainage engineering (No. 04022051; Vol. 149, Issue 3). https://doi.org/10.1061/jidedh.ireng-9796 10.1061/jidedh.ireng-9796 urn:nbn:se:umu:diva-201888 oai:DiVA.org:umu-201888 2023 oai:DiVA.org:umu-201888