Electronic archive of data on leads in the Arctic seas.

Sea ice leads are narrow linear-shaped openings enclosed in ice. They may be covered with nilas or young ice. Climatological values of modal lead orientation and specific lead length are important characteristics for the study of spatial and temporal variability of leads both for needs of navigation in the Arctic and in the context of the Arctic climate change. To calculate climatological lead characteristics, a long-period data archive is required. The main source of observations on leads is satellite imagery. Manual interpretation of leads in satellite images is an extremely labor-intensive process. In 2025, we developed a method for automatic identification of leads retrieved from SuomiNPP infrared satellite images using a convolutional neural network. As a result of image processing with this method, each lead is represented as a polyline and defined by a set of geographic coordinates for the ends of the line segments. This kind of data representation provides calculation of all the main lead characteristics. Using the method, we compiled a data archive on leads in the Laptev and East Siberian Seas for 2012–2025. Data on leads for spring months were derived from visible-range SuomiNPP (for April–May 2012–2025) and Terra (for March–May 2000–2011) images. For this type of images, the model was retrained. The results obtained were verified by comparing the modal orientation and specific length of leads calculated from automatically identified data with those from expert interpretation data. It was found that the difference in modal orientation of leads was minor in a vast majority of cases: the average difference was 7–8°. Differences in the specific lead length were also small, averaging 1–10 m/km2. Data on leads in the winter months of 2000–2011 were retrieved from the free available results of processing the MODIS ice surface temperature data by applying the “multiway” algorithm, which is part of our method for automatic identification of leads. Verification of the results obtained in this way showed that the average difference in modal orientation values was 11°. Therefore, using lead data from all the datasets produced will not introduce significant error in the subsequent determination of generalized lead characteristics, provided that differences in the spatial resolution are taken into account. This enables us to aggregate all the datasets produced into a single data archive on leads in the Laptev and East Siberian seas in 2000–2025.