DATA REFERENCE

Deb, P., Orr, A. & Turner, J. "Polar Weather Research and Forecasting (WRF) simulation of West Antarctic climate, summer 1980-2015" (2018) Polar Data Centre, Natural Environment Research Council, UK doi:10.5285/9536f22e-37dd-4f37-948b-e19c70e15292

LITERATURE REFERENCE

Deb et al., 2016 and Deb et al., 2018.

Methodology:

The model uses the recommended configuration of version 3.5.1 of the polar modified Weather Research and Forecasting model (known as "Polar WRF"; Bromwich et al., 2013) for West Antarctica described by Deb et al. (2016). The model physics employed are - the WRF Single Moment 5-Class (WSM5) cloud microphysics scheme, the Mellor-Yamada-Janjic (MYJ) boundary layer scheme, the Rapid Radiative Transfer Model for General Circulation Models (RRTMG), and the Noah land surface model.

The model outer domain encompasses the West Antarctic ice sheet and a large part of the surrounding ocean with 74 x 61 grid points at 45 km horizontal grid spacing centred on the point 72.7°S, 118.5°W, and the nested (one-way) inner domain covers the Amundssen Sea Embayment with 132 × 111 grid points at 15 km grid spacing centred on the point 73.5°S, 118.4°W. Both grids are defined on a polar stereographic projection with straight vertical longitude 120.5°W and latitude of true scale 70°S. The inner domain has longitude extremes at 164.1°W and 73.1°W, and latitude extremes at 81.1°S and 64.3°S (though the grid, being rectangular in polar stereographic space, does not completely fill this extent). In the vertical, both domains extend from the surface to 50 hPa, with 30 vertical levels.

The model employs spectral nudging (outer domain and wave numbers 1-5 only, from approximately 1.5 km to the model top), high-resolution orography based on the Bedmap2 dataset, and meteorological forcing from ERA-Interim reanalysis (Dee et al., 2011). Additionally, the surface boundary forcing of the model includes daily satellite observations of sea ice concentration (based on the 25 km resolution Bootstrap data set; Comiso, 2000) and sea surface temperature (based on the 0.25° Advanced Very High Resolution Radiometer (AVHRR) data set; Casey et al., 2010). The hindcast was produced by running separate simulations for each of the summer months from December 1979 to February 2015 (with 24 hr spin up).

Python 2.7.3 version was used for post-processing. The netCDF4 library was used to extract the hourly model data and produce the monthly means (using the NumPy Python library). The monthly means were stored in binary npz format which can be easily loaded on Python. See the following link on how to save and load npz files on Python:

https://docs.scipy.org/doc/numpy/reference/routines.io.html#numpy-binary-files-npy-npz