PyGeoHydro: Retrieve Geospatial Hydrology Data#

PyPi Conda Version CodeCov Python Versions Downloads

CodeFactor black pre-commit Binder

Features#

PyGeoHydro (formerly named hydrodata) is a part of HyRiver software stack that is designed to aid in hydroclimate analysis through web services. This package provides access to some public web services that offer geospatial hydrology data. It has three main modules: pygeohydro, plot, and helpers.

PyGeoHydro supports the following datasets:

  • NWIS for daily mean streamflow observations (returned as a pandas.DataFrame or xarray.Dataset with station attributes),

  • CAMELS for accessing streamflow observations (1980-2014) and basin-level attributes of 671 stations within CONUS.

  • Water Quality Portal for accessing current and historical water quality data from more than 1.5 million sites across the US,

  • NID for accessing the National Inventory of Dams web service,

  • HCDN 2009 for identifying sites where human activity affects the natural flow of the watercourse,

  • NLCD 2019 for land cover/land use, imperviousness, imperviousness descriptor, and canopy data. You can get data using both geometries and coordinates.

  • WBD for accessing Hydrologic Unit (HU) polygon boundaries within the US (all HUC levels).

  • SSEBop for daily actual evapotranspiration, for both single pixel and gridded data.

Also, it has two other functions:

  • interactive_map: Interactive map for exploring NWIS stations within a bounding box.

  • cover_statistics: Categorical statistics of land use/land cover data.

  • overland_roughness: Estimate overland roughness from land use/land cover data.

The plot module includes two main functions:

  • signatures: Hydrologic signature graphs.

  • cover_legends: Official NLCD land cover legends for plotting a land cover dataset.

  • descriptor_legends: Color map and legends for plotting an imperviousness descriptor dataset.

The helpers module includes:

  • nlcd_helper: A roughness coefficients lookup table for each land cover and imperviousness descriptor type which is useful for overland flow routing among other applications.

  • nwis_error: A dataframe for finding information about NWIS requests’ errors.

You can find some example notebooks here.

Moreover, under the hood, PyGeoHydro uses AsyncRetriever for making requests asynchronously with persistent caching. This improves the reliability and speed of data retrieval significantly. AsyncRetriever caches all request/response pairs and upon making an already cached request, it will retrieve the responses from the cache if the server’s response is unchanged.

You can control the request/response caching behavior by setting the following environment variables:

  • HYRIVER_CACHE_NAME: Path to the caching SQLite database. It defaults to ./cache/aiohttp_cache.sqlite

  • HYRIVER_CACHE_EXPIRE: Expiration time for cached requests in seconds. It defaults to -1 (never expire).

  • HYRIVER_CACHE_DISABLE: Disable reading/writing from/to the cache. The default is false.

For example, in your code before making any requests you can do:

import os

os.environ["HYRIVER_CACHE_NAME"] = "path/to/file.sqlite"
os.environ["HYRIVER_CACHE_EXPIRE"] = "3600"
os.environ["HYRIVER_CACHE_DISABLE"] = "true"

You can also try using PyGeoHydro without installing it on your system by clicking on the binder badge. A Jupyter Lab instance with the HyRiver stack pre-installed will be launched in your web browser, and you can start coding!

Please note that since this project is in early development stages, while the provided functionalities should be stable, changes in APIs are possible in new releases. But we appreciate it if you give this project a try and provide feedback. Contributions are most welcome.

Moreover, requests for additional functionalities can be submitted via issue tracker.

Citation#

If you use any of HyRiver packages in your research, we appreciate citations:

@article{Chegini_2021,
    author = {Chegini, Taher and Li, Hong-Yi and Leung, L. Ruby},
    doi = {10.21105/joss.03175},
    journal = {Journal of Open Source Software},
    month = {10},
    number = {66},
    pages = {1--3},
    title = {{HyRiver: Hydroclimate Data Retriever}},
    volume = {6},
    year = {2021}
}

Installation#

You can install PyGeoHydro using pip after installing libgdal on your system (for example, in Ubuntu run sudo apt install libgdal-dev). Moreover, PyGeoHydro has an optional dependency for using persistent caching, requests-cache. We highly recommend installing this package as it can significantly speed up send/receive queries. You don’t have to change anything in your code, since PyGeoHydro under-the-hood looks for requests-cache and if available, it will automatically use persistent caching:

$ pip install pygeohydro

Alternatively, PyGeoHydro can be installed from the conda-forge repository using Conda:

$ conda install -c conda-forge pygeohydro

Quick start#

We can explore the available NWIS stations within a bounding box using interactive_map function. It returns an interactive map and by clicking on a station some of the most important properties of stations are shown.

import pygeohydro as gh

bbox = (-69.5, 45, -69, 45.5)
gh.interactive_map(bbox)
Interactive Map

We can select all the stations within this boundary box that have daily mean streamflow data from 2000-01-01 to 2010-12-31:

from pygeohydro import NWIS

nwis = NWIS()
query = {
    **nwis.query_bybox(bbox),
    "hasDataTypeCd": "dv",
    "outputDataTypeCd": "dv",
}
info_box = nwis.get_info(query)
dates = ("2000-01-01", "2010-12-31")
stations = info_box[
    (info_box.begin_date <= dates[0]) & (info_box.end_date >= dates[1])
].site_no.tolist()

Then, we can get the daily streamflow data in mm/day (by default the values are in cms) and plot them:

from pygeohydro import plot

qobs = nwis.get_streamflow(stations, dates, mmd=True)
plot.signatures(qobs)

By default, get_streamflow returns a pandas.DataFrame that has a attrs method containing metadata for all the stations. You can access it like so qobs.attrs. Moreover, we can get the same data as xarray.Dataset as follows:

qobs_ds = nwis.get_streamflow(stations, dates, to_xarray=True)

This xarray.Dataset has two dimensions: time and station_id. It has 10 variables including discharge with two dimensions while other variables that are station attitudes are one dimensional.

We can also get instantaneous streamflow data using get_streamflow. This method assumes that the input dates are in UTC time zone and returns the data in UTC time zone as well.

date = ("2005-01-01 12:00", "2005-01-12 15:00")
qobs = nwis.get_streamflow("01646500", date, freq="iv")

We can get the CAMELS dataset as a geopandas.GeoDataFrame that includes geometry and basin-level attributes of 671 natural watersheds within CONUS and their streamflow observations between 1980-2014 as a xarray.Dataset, like so:

attrs, qobs = gh.get_camels()
CAMELS

The WaterQuality has a number of convenience methods to retrieve data from the web service. Since there are many parameter combinations that can be used to retrieve data, a general method is also provided to retrieve data from any of the valid endpoints. You can use get_json to retrieve stations info as a geopandas.GeoDataFrame or get_csv to retrieve stations data as a pandas.DataFrame. You can construct a dictionary of the parameters and pass it to one of these functions. For more information on the parameters, please consult the Water Quality Data documentation. For example, let’s find all the stations within a bounding box that have Caffeine data:

from pynhd import WaterQuality

bbox = (-92.8, 44.2, -88.9, 46.0)
kwds = {"characteristicName": "Caffeine"}
wq = WaterQuality()
stations = wq.station_bybbox(bbox, kwds)

Or the same criterion but within a 30-mile radius of a point:

stations = wq.station_bydistance(-92.8, 44.2, 30, kwds)

Then we can get the data for all these stations the data like this:

sids = stations.MonitoringLocationIdentifier.tolist()
caff = wq.data_bystation(sids, kwds)
Water Quality

Moreover, we can get land use/land cove data using nlcd_bygeom or nlcd_bycoods functions, percentages of land cover types using cover_statistics, and overland roughness using overland_roughness. The nlcd_bycoords function returns a geopandas.GeoDataFrame with the NLCD layers as columns and input coordinates as the geometry column. Moreover, the nlcd_bygeom function accepts both a single geometry or a geopandas.GeoDataFrame as the input.

from pynhd import NLDI

basins = NLDI().get_basins(["01031450", "01318500", "01031510"])
lulc = gh.nlcd_bygeom(basins, 100, years={"cover": [2016, 2019]})
stats = gh.cover_statistics(lulc["01318500"].cover_2016)
roughness = gh.overland_roughness(lulc["01318500"].cover_2019)
Land Use/Land Cover

Next, let’s use ssebopeta_bygeom to get actual ET data for a basin. Note that there’s a ssebopeta_bycoords function that returns an ETA time series for a single coordinate.

geometry = NLDI().get_basins("01315500").geometry[0]
eta = gh.ssebopeta_bygeom(geometry, dates=("2005-10-01", "2005-10-05"))
Actual ET

Additionally, we can pull all the US dams data using NID. Let’s get dams that are within this bounding box and have a maximum storage larger than 200 acre-feet.

nid = NID()
dams = nid.get_bygeom((-65.77, 43.07, -69.31, 45.45), "epsg:4326")
dams = nid.inventory_byid(dams.id.to_list())
dams = dams[dams.maxStorage > 200]

We can get also all dams within CONUS in NID with maximum storage larger than 200 acre-feet:

import geopandas as gpd

world = gpd.read_file(gpd.datasets.get_path("naturalearth_lowres"))
conus = world[world.name == "United States of America"].geometry.iloc[0].geoms[0]

dam_list = nid.get_byfilter([{"maxStorage": ["[200 5000]"]}])
dams = dam_list[0][dam_list[0].is_valid]
dams = dams[dams.within(conus)]
Dams

The WBD class allows us to get Hydrologic Unit (HU) polygon boundaries. Let’s get the two Hudson HUC4s:

from pygeohydro import WBD

wbd = WBD("huc4")
hudson = wbd.byids("huc4", ["0202", "0203"])