Source code for prereise.gather.solardata.ga_wind.ga_wind

import dateutil
import h5pyd
import numpy as np
import pandas as pd
from pyproj import Transformer
from tqdm import tqdm

from prereise.gather.solardata.ga_wind.helpers import ll2ij, proj_string
from prereise.gather.solardata.helpers import get_plant_id_unique_location


def retrieve_data(
    solar_plant, hs_api_key, start_date="2007-01-01", end_date="2014-01-01"
):
    """Retrieves irradiance data from Gridded Atmospheric Wind Integration
    National dataset.

    :param pandas.DataFrame solar_plant: plant data frame.
    :param str hs_api_key: API key.
    :param str start_date: start date.
    :param str end_date: end date.
    :return: (*pandas.DataFrame*) -- data frame with *'Pout'*, *'plant_id'*, *'ts'* and
        *'ts_id'* as columns. Values are power output for a 1MW generator.
    """

    # Identify unique location
    coord = get_plant_id_unique_location(solar_plant)

    # Build query
    hs_endpoint = "https://developer.nrel.gov/api/hsds"
    hs_endpoint_fallback = "https://developer.nrel.gov/api/hsds/"
    hs_username = None
    hs_password = None

    try:
        f = h5pyd.File(
            "/nrel/wtk-us.h5",
            "r",
            username=hs_username,
            password=hs_password,
            endpoint=hs_endpoint,
            api_key=hs_api_key,
        )
    except OSError:
        f = h5pyd.File(
            "/nrel/wtk-us.h5",
            "r",
            username=hs_username,
            password=hs_password,
            endpoint=hs_endpoint_fallback,
            api_key=hs_api_key,
        )

    # Get coordinates of nearest location
    lat_origin, lon_origin = f["coordinates"][0][0]
    transformer = Transformer.from_pipeline(proj_string)
    ij = {key: ll2ij(transformer, lon_origin, lat_origin, *key) for key in coord.keys()}

    # Extract time series
    dt = f["datetime"]
    dt = pd.DataFrame({"datetime": dt[:]})
    dt["datetime"] = dt["datetime"].apply(dateutil.parser.parse)

    dt_range = dt.loc[(dt.datetime >= start_date) & (dt.datetime < end_date)]

    data = pd.DataFrame({"Pout": [], "plant_id": [], "ts": [], "ts_id": []})

    for key, val in tqdm(ij.items(), total=len(ij)):
        ghi = f["GHI"][min(dt_range.index) : max(dt_range.index) + 1, val[0], val[1]]
        data_loc = pd.DataFrame({"Pout": ghi})
        data_loc["Pout"] /= max(ghi)
        data_loc["ts_id"] = range(1, len(ghi) + 1)
        data_loc["ts"] = pd.date_range(start=start_date, end=end_date, freq="H")[:-1]

        for i in coord[key]:
            data_site = data_loc.copy()
            data_site["plant_id"] = i

            data = data.append(data_site, ignore_index=True, sort=False)

    data["plant_id"] = data["plant_id"].astype(np.int32)
    data["ts_id"] = data["ts_id"].astype(np.int32)

    data.sort_values(by=["ts_id", "plant_id"], inplace=True)
    data.reset_index(inplace=True, drop=True)

    return data