7.3.3.19. Visual_iced Dataloader

Visual_iced is a dataloader for .tiff images, which are outputs from the visual_iced library developed by Martin Rogers at the British Antarctic Survey’s AI Lab. These visual_iced images are ice/water binary files, generated from a combination of MODIS and SAR satellite imagery.

In the source data, 0s are representative of open water, and 1s are representative of ice. In the dataloader, we map these values to sea ice concentration, in the range of 0 to 100. Values between 0 and 100 are generated by the aggregation of the 0s and 1s within each cell.

Note

The visual_iced dataloader only supports loading in single files, as the visual_iced datasets are not temporally continuous within a given boundary.

Name in config: 'visual_iced'

class meshiphi.dataloaders.scalar.visual_iced.VisualIcedDataLoader(bounds, params)

import_data(bounds)

Reads in data from Visual_Ice NetCDF files. Renames coordinates to ‘lat’ and ‘long’.

Parameters:

bounds (Boundary) – Initial boundary to limit the dataset to

Returns:

visual_ice dataset within limits of bounds. Dataset has coordinates ‘lat’, ‘long’, and variable ‘SIC’

Return type:

xr.Dataset

import_from_nc(xarray_dataset)

applys transformations need to import a netcdf file

Parameters:

xarray_dataset (xr.Dataset) – dataset to be transformed

Returns:

transformed dataset

Return type:

xr.Dataset

import_from_tiff(xarray_dataset)

applys transformations need to import a tiff file

Parameters:

xarray_dataset (xr.Dataset) – dataset to be transformed

Returns:

transformed dataset

Return type:

xr.Dataset