snowex-hackweek / space-station Goto Github PK

Need a shape file with pertinent site meta data for all SNOTELs and SnowEx Sites - create a gist (geojson) and export a binary .shp as zip. This can then be handed off to @arienzo to manipulate in arcGIS

• Spatial plot of difference
• Map of what datasources are included colored by distance to snotel site
• Histogram of depth differences on log scale
• box and whisker for each snotel
• Flow chart figure of the pipeline

Windroses, etc. Check out metloom package

We need mock objects of the data structure to allow plotting and analysis work to happen in tandem

What we want, what we need to avoid.

max distance from site should change depending on different parameters such as vegetative cover, elevation, aspect, etc...

as a developer I need to know what n # of met sites are closest to any given snow ex sites.

The function should take a lat/long of a snow ex site, a input geojson (mostly snotels, but any sites) and a n # of sites and return a geojson representation of the n # of snotels that are closest to that snow ex site. The calculation does not need to be exact but should be accurate enough to be trusted and with low computational overhead.

A snippet which may help taken from another project takes a input gdf and returns other sites within that gdf that are closest to a given site. Just need to modify it to use a lat and long as input rather than a reference to another site

def nearest_sites(df_meta, resp_site=None, pred_sites=None, num_sites=10):
    try:
        num_sites = int(num_sites)
    except ValueError:
        num_sites = 10
    tx = df_meta[df_meta["triplet"] == resp_site]["latitude"]
    ty = df_meta[df_meta["triplet"] == resp_site]["longitude"]

    def calc_dist(row, tx=tx, ty=ty):
        return (row["latitude"] - tx) ** 2 + (row["longitude"] - ty) ** 2

    df_meta["proximity"] = df_meta.apply(calc_dist, axis=1)
    df_meta.sort_values(by="proximity", inplace=True)
    df_nearest = df_meta.head(num_sites + 1).copy()
    triplets_in_use = tuple(set([i for i in (resp_site,) + pred_sites if i]))
    df_in_use = df_meta[df_meta["triplet"].isin(triplets_in_use)]
    df = pd.concat([df_nearest, df_in_use]).drop_duplicates(ignore_index=True)
    return df

As a developer I need a dataloader for both snowEx, AWDB, and metloom data (and any other sources that are identified). The methods or classes should take a site_id, date range and sensor/element type and return a datetime indexed dataframe with the data. Perhaps all data sources should be called from a single method and return all data sources in a single dataframe.

@Ibrahim-Ola

We need to determine the exact and accurate location of station data

A reminder to decide on how exactly we want to analyze our data