Arctic gully identification using high-resolution lidar

Hillslope flow networks in the upland Arctic include unchannelized water tracks, curvilinear zones of increased saturation, and channelized gullies which can exist along the same longitudinally connected flow network. However, controls on patterns of channelization on continuous permafrost landscapes remain poorly constrained in part due to the difficulty of remotely detecting discontinuous channelized segments. Here we introduce a novel method to identify gullies within Arctic hillslope flow networks using high-resolution LiDAR verified with field observations on Alaska’s North Slope. This method combines slope, tangential curvature, normalized elevation, and a delineated flow network to model and detect gullies on the landscape. Our best-fit model accurately identifies 79% of gullies (n=11/14) and 80% of water tracks (n=31/39) observed in the field. For the 401.5 km2 study region, we found that 26.6% of hillslope flow networks contain gullies. We detected 13439 water track networks (8686 km in length, 94.5% of hillslope network) and 3863 gully networks (563 km in length, 6.1% of hillslope flow network). Gully networks were most abundant in Holocene-aged sediments and did not show clear topographic patterns with slope, aspect, or drainage area, suggesting that localized thermal processes may be the primary control on initial gully formation. While the exact location of new gullies on Arctic hillslopes may be difficult to predict, we expect their flow networks to transition from primarily unchannelized to channelized as permafrost thaws, with direct impacts on water, nutrient, and sediment transport.