Load function for weather stations

[Balinus]

We need a function that will load weather stations data.

The problem is that there is no standard for such data.

Objectives

• Canadian weather stations
• NCEI weather stations (see NCEI package)
• A hardcoded way to import csv data (columns is station, rows are time)

[Zeitsperre]

I've done a fair amount of this (badly) in python. There might be a way to adapt some of my work in https://github.com/Zeitsperre/canada-climate-python to perform this.

While this work didn't use pandas, I'm wondering if an implementation of it or something similar to it exists for Julia. Environement Canada station data is less-than-optimally formatted, so a portion of my code simply deals with workarounds that pandas might be better suited for. I'll start looking into this Monday.

[Balinus]

DataFrames.jl and Pandas.jl could be a solution, but would add other dependencies. The other option would be to continue with AxisArrays.jl which can index Categorical axis (station number for example).

[Balinus]

@houton199 Any code that could be used here?

[Zeitsperre]

@Balinus

I've retackled this issue in a few places, converting either ECCC flat files or CSVs to NetCDF with "loose" CF conventions. The code is the work of several people (@sbiner, @RondeauG) and is currently being integrated into Ouranosinc/miranda. See https://github.com/Ouranosinc/miranda/tree/master/miranda/eccc for some examples of how we're dealing with hourly flat files. Feel free to take inspiration or raise issues or suggest edits if you see something.

[Balinus]

Thanks! Will have a look.