Title: Determination of Arctic melt pond fraction and sea ice roughness from Unmanned Aerial Vehicle (UAV) imagery.

Journal：Advances in Polar Science, 29 (3), doi: 10.13679/j.advps.2018.3.00003.

Authors: WANG M. -F., J. Su *, T. Li, X. -Y. Wang, Q. Ji, Y. Cao, L. Lin, and Y. -L. Liu

Abstract: Melt ponds on Arctic sea ice are of great significance in the study of the heat balance in the ocean mixed layer, mass and salt balances of Arctic sea ice, and other aspects of the earth-atmosphere system. During the 7th Chinese National Arctic Research Expedition, aerial photographs were taken from an Unmanned Aerial Vehicle over an ice floe in the Canada Basin. Using threshold discrimination and three-dimensional modeling, we estimated a melt pond fraction of 1.63% and a regionally averaged surface roughness of 0.12 for the study area. In view of the particularly foggy environment of the Arctic, aerial images were defogged using an improved dark channel prior based image defog algorithm, especially adapted for the special conditions of sea ice images. An aerial photo mosaic was generated, melt ponds were identified from the mosaic image and melt pond fractions were calculated. Three-dimensional modeling techniques were used to generate a digital elevation model allowing relative elevation and roughness of the sea ice surface to be estimated. Analysis of the relationship between the distributions of melt ponds and sea ice surface roughness shows that melt ponds are smaller on sea ice with higher surface roughness, while broader or deeper melt ponds usually occur in areas where sea ice surface roughness is lower.