Development of Aerial Photos and LIDAR Data Approaches to Map Spatial and Temporal Evolution of Ditch Networks in Peat-Dominated Catchments

Abstract

Spatiotemporal information on historical peatland drainage is needed to relate past land use to observed changes in catchment hydrology. Comprehensive knowledge of historical development of peatland management is largely unknown at catchment scale. Aerial photos and LIDAR data enlarge the possibilities for identifying past peatland drainage patterns. Here, our objectives are: (1) to develop techniques for semi-automatically mapping the location of ditch networks in peat-dominated catchments by using aerial photos and LIDAR data, and (2) to generate time series of drainage networks. Our approaches provide open-access techniques to systematically map ditches in peatdominated catchments through time. We focused on the algorithm in such a way that we can identify the ditch networks from raw aerial images and LIDAR data based on the modification of multiple filters and number of threshold values. Such data are needed to relate spatiotemporal drainage patterns to observed changes in many northern rivers. We demonstrate our approach using data from the Simojoki river catchment (3160 km 2 ) in northern Finland. The catchment is dominated by forests and peatlands that were almost all drained after 1960. For two representative locations in cultivated peatland (downstream) and peatland forest (upstream) areas of the catchment; we found total ditch length density (km/km 2 ) estimated from aerial images and LIDAR data based on our proposed algorithm varied from 2% to 50% compared against the monitored ditch length available from National Land survey of Finland (NLSF) in 2018. A different pattern of source variation in ditch network density was also observed for whole catchment estimates and for available drained peatland database from Natural Resources Institute Finland (LUKE). Despite such differences no significant differences were found using the non-parametric Mann-Whitney U-test with 0.05 significance level based on the samples of pixel based identified ditches between (i) aerial images & NLSF vector files and (ii) LIDAR data & NLSF vector files.

Mapping Drainage Ditches in Forested Landscapes Using Deep Learning and Aerial Laser Scanning

“…Still, it will be worth exploring whether adding more indices besides HPMF can improve the prediction accuracy. For example, a recent study in Finland by Bhattacharjee et al (2021) demonstrated that aerial photos may contain important information for mapping drainage ditches on peatlands. However, in areas hidden beneath canopy, it might be better to include topographical indices such as impoundment size index implemented in Whitebox tools 1.4.0 (Lindsay 2018), which can be included as multiband images.…”

Utilizing historical maps in identification of long-term land use and land cover changes

“…For example, for a river catchment level hydrology analysis, the knowledge of watercourse network expansion, both downstream and upstream, is needed (Bhattacharjee et al. 2021 ). Such knowledge of drainage systems is also of importance for the assessment of maintenance needs or candidate mires for restoration (Hasselquist et al.…”

“…However, the coarse spatial resolution of old satellite images and the inability to detect features below the canopy layer in satellite and aerial images limit their utilization in mapping linear features (Bhattacharjee et al. 2021 ).…”

Virtual landscape-scale restoration of altered channels helps us understand the extent of impacts to guide future ecosystem management

“…Previous ditch studies utilizing LiDAR-derived DEM have mapped smaller areas up to 150 ha (Cazorzi et al 2013 ; Rapinel et al 2015 ; Roelens et al 2018a ) while we mapped 1708 km of ditches and 248 km of natural streams in 11 study regions covering a total of 415 km 2 . Only recently, Bhattacharjee et al ( 2021 ) conducted a ditch detection analysis in a large forested peatland in Finland; this study was highly reliant on aerial imagery, and notable uncertainties were reported when aerial images were unavailable for parts of the study area. While the strength of manual interpretation of high-resolution DEM for ditch detection was demonstrated in this study, we also highlight the extensive investment of man-hours for manual digitization.…”