Skip to content

Latest commit

 

History

History
52 lines (39 loc) · 4.75 KB

DATA_README.md

File metadata and controls

52 lines (39 loc) · 4.75 KB

Authors

Overview

This repository contains all NASA ECCO data used for the AI4ER Guided Team Challenge oceans project. Data was downloaded from PO.DAAC using the entries in the table below. Download and preprocessing scripts can be found in the corresponding GitHub repository.

Data Type PO.DAAC Entry DOI
SSH ECCO_L4_SSH_05DEG_MONTHLY_V4R4 https://doi.org/10.5067/ECG5M-SSH44
SSS ECCO_L4_TEMP_SALINITY_05DEG_MONTHLY_V4R4 https://doi.org/10.5067/ECG5M-OTS44
SST ECCO_L4_TEMP_SALINITY_05DEG_MONTHLY_V4R4 https://doi.org/10.5067/ECG5M-OTS44
ZWS ECCO_L4_STRESS_05DEG_MONTHLY_V4R4 https://doi.org/10.5067/ECG5M-STR44
OBP ECCO_L4_OBP_05DEG_MONTHLY_V4R4 https://doi.org/10.5067/ECG5M-OBP44
Monthly-Mean Velocity ECCO_L4_OCEAN_VEL_05DEG_MONTHLY_V4R4 https://doi.org/10.5067/ECG5M-OVE44
Bolus Velocity ECCO_L4_BOLUS_05DEG_MONTHLY_V4R4 https://doi.org/10.5067/ECG5M-BOL44
Model Grid Geometry ECCO_L4_GEOMETRY_LLC0090GRID_V4R4 https://doi.org/10.5067/ECL5A-GRD44

Use and redistribution of this data is in line with NASA EarthData's Data and Information Policy and NASA PO.DAAC's Data Use and Citation Policy.

Data Structure

Latitudes of Interest

We primarily focus on abyssal circulation at four latitudes: 26N, 30S, 55S, and 60S. Data is monthly and derived from the 0.5 degree latitude-longitude interpolated product on PO.DAAC. When working in Python: we recommend using xarray for nc files and pickle/numpy for pickle files.

There is one directory for each latitude. All of these will contain the following:

  • [LATITUDE].nc: Satellite-observable variables for the latitude of interest, as well for the latitude directly above and below (plus/minus 0.5 degrees).
    • These include ample description embedded directly into the xarray object.
  • [LATITUDE]_moc_depth.pickle and [LATITUDE]_moc_density.pickle: The depth- and density-space MOC, taken as the maximum of the depth- and density-space streamfunctions, respectively.
    • This will read into Python as a numpy array, with shape [# TIME STEPS].
  • [LATITUDE]_sf_depth.pickle and [LATITUDE]_sf_density.pickle: The full overturning streamfunction in depth- and density-space, i.e., no maximum has been taken over depth/density yet.
    • This will read into Python as a numpy array, with shape [# VERTICAL LEVELS x # TIME STEPS].
  • [LATITUDE]_density_range.pickle: The density range for the density-space streamfunction. This is different for each latitude.
    • This will read into Python as a numpy array, with shape [# VERTICAL LEVELS].

30S includes three extra files for the depth- and density-space streamfunction for only the Atlantic Ocean (30S_atlantic_sf_depth.pickle and 30S_atlantic_sf_density.pickle, respectively) and the density range for the Atlantic Ocean streamfunction (30S_atlantic_density_range.pickle).

At the top-level directory, depth_range.pickle includes the actual depths for the depth-space streamfunction, which is constant across latitudes. This is taken directly from ECCO's vertical grid geometry profile. This will read into Python as a numpy array, with shape [# VERTICAL LEVELS].

All streamfunction and MOC strength measurements have units of Sverdrup, i.e., 10^6 m^3s^-1. Satellite-observable variables each have different units--see data variable descriptions embedded directly into the xarray object for more details.

Full Southern Ocean

We also experiment with machine learning models trained on the entire Southern Ocean, i.e., all latitudes below 30S. The two directories are inputs and moc:

  • inputs: Contains southern_ocean.nc, which holds all satellite-observable variables for all latitudes in the Southern Ocean, as well as southern_ocean_floor.nc, which holds the bottom depth for the entire Southern Ocean.
  • moc: Contains a single file for each latitude's MOC strength through time. Filenames are in the format [LATITUDE]_moc.pickle.
    • These will read into Python as numpy arrays, with shape [# TIME STEPS].