Arctic Melt Pond Fraction and Binary Classification, 2021-2022 ...

• Main Authors: Lee, Sanggyun, Stroeve, Julienne
• Format: Dataset
• Language: English
• Published: NERC EDS UK Polar Data Centre 2023
• Subjects: "EARTH SCIENCE","CRYOSPHERE","SEA ICE","ICE GROWTH/MELT"; "EARTH SCIENCE","OCEANS","SEA ICE","ICE GROWTH/MELT"; "EARTH SCIENCE","SPECTRAL/ENGINEERING","INFRARED WAVELENGTHS","INFRARED IMAGERY"; "EARTH SCIENCE","SPECTRAL/ENGINEERING","VISIBLE WAVELENGTHS","VISIBLE IMAGERY"; "EARTH SCIENCE","CRYOSPHERE","SEA ICE"; "EARTH SCIENCE","OCEANS","SEA ICE"; Arctic; MODIS; melt pond; remote sensing; Sea ice
• Online Access: https://dx.doi.org/10.5285/834dc665-1ae6-464d-8c29-265a9be5229a; https://data.bas.ac.uk/full-record.php?id=GB/NERC/BAS/PDC/01786

This dataset provides daily, 8-day, and monthly Arctic melt pond fractions and binary classification, from 2021-05-01 to 2022-08-31. Level-2 MODerate resolution Imaging Spectroradiometer (MODIS) top-of-the-atmosphere (TOA) reflectances for bands 1-4 were obtained, to which two machine learning algorithms such as multi-layer neural networks and logistic regression were applied to map melt pond fraction and binary melt pond/ice classification. This work was funded by NERC standard grant NE/R017123/1. ... : Level-2 MODIS top-of-the-atmosphere (TOA) reflectances for bands 1-5 were used for the melt pond fraction and binary classification. Additionally, MODIS bands 5, 13, 16 and 19 were used to remove cloud shadows. The MOD35 data product was also used for cloud masking and MOD29 ice surface temperature product was used to flag refrozen melt ponds. Two machine learning algorithms were applied to the TOA band reflectances to map melt pond fraction and binary melt pond/ice/ocean classification. These included a Multi-Neural Network (MNN) and Multinomial Logistic Regression (MLR). Results were validated against high-resolution WorldView imagery, ship observations and other high resolution unclassified spy satellite data. ...