Handout (1.2 MB)
In this ongoing study, we examine polar weather research and forecasting reanalysis data in conjunction with surface data. Through statistical analysis, we derive factors that significantly impact sea ice deviation and subsequently establish correlations with those defined factors, which are then plotted and used to train a deep neural network (DNN). Correlated data is input into a DNN in order to develop projections of sea ice concentration throughout annual growth cycles between the years 2000 and 2016.
The role of DNNs in climate and polar research is still being established; thus, it is critical to pursue this approach to sea ice concentration prediction. The intention of this research is to develop a DNN foundation for sea ice prediction that can be applied to additional Arctic and Antarctic regions. The resulting Laptev DNN predictions will eventually undergo comparative analysis against actual sea ice data in order to determine sea ice variable correlations and validate the accuracy of the study’s DNN predictions.
References:
Eicken, H., et al. “Sea-Ice Processes in the Laptev Sea and Their Importance for Sediment Export.” Continental Shelf Research, vol. 17, no. 2, Feb. 1997, pp. 205–233, https://doi.org/10.1016/s0278-4343(96)00024-6. Accessed 08 Sept. 2023.
“European State of the Climate 2020 | Copernicus.” Copernicus.eu, 2020, climate.copernicus.eu/esotc/2020. Accessed 08 Sept. 2023.
N., Meier, W., et al. Sea Ice. 2022, repository.library.noaa.gov/view/noaa/48539, https://doi.org/10.25923/xyp2-vz45. Accessed 09 Sept. 2023.

