Marine Forecast

[lswald]

Have marine forecast been implemented yet? thanks

[scadergit]

Sorry, no. It was a priority for a hot minute then buried under others. The current efforts will make that easier to implement in the future, but there is no eta that I'm aware of.

[xt0fer]

are there any old APIs that do marine forecasts?

[IanTBlack]

This is something I am interested in too. I would like to use the NWS API to specify a point to get the forecast zone and then to get the forecast for that zone. I am working on developing a one-stop-shop for weather information for science crew aboard a research vessel and having access to marine forecasts via the API would help tremendously.

While I would like to get forecast data as a table or json, I have hacked together a work around for getting what I need through forecast text. Please keep in mind that there are probably better ways to do this.

In the following Python code (tested in 3.9) I query https://tgftp.nws.noaa.gov/data/forecasts/marine repeatedly to get a complete list of forecast text file urls for the US. I then use the NWS API to obtain a forecast zone ID based on a lat/lon that I feed it (in this case, the location of the OOI CE02 mooring). I then parse the list of forecast text urls by zone ID and then pull in the text for that forecast zone.

import re
import requests
import time
from typing import NamedTuple

def main():
    OOI_CE02_location = Point(lat=44.6393, lon=-124.304)
    nws = NWS()
    zone_text_urls = nws.list_marine_zone_text_urls(ignore_weather_statements=True)
    forecast = nws.get_marine_zone_text_forecast(OOI_CE02_location, zone_text_urls)
    print(forecast)

class Point(NamedTuple):
    """Takes lat and lon as keywords in any order and orders them for later
    use."""
    lat: float
    lon: float

class NWS():
    def __init__(self):
        self._text_base = 'https://tgftp.nws.noaa.gov/data/forecasts/marine'
        self._nws_api = 'https://api.weather.gov'

    def list_marine_zone_text_urls(self,
                                   ignore_weather_statements = True) -> list:
        zone_text_urls = []  #Holder list for urls.
        main_html = requests.get(self._text_base).text #Parse the main page.
        geotypes = re.findall('">(.*?)\/<\/a>',main_html)
        if ignore_weather_statements is True:
            geotypes = [geotype for geotype in geotypes
                        if 'weather_stmt' not in geotype]
        for geotype in geotypes:
            geotype_html = requests.get('/'.join((self._text_base,
                                                  geotype))).text
            geozones = re.findall('">(.*?)\/<\/a>',geotype_html)
            for geozone in geozones:
                zone_files_html = requests.get('/'.join((self._text_base,
                                                         geotype,geozone))).text
                zone_files = re.findall('">(.*?\.txt)<\/a>',zone_files_html)
                complete_urls = ['/'.join((self._text_base,
                                           geotype,
                                           geozone,
                                           file)) for file in zone_files]
                zone_text_urls = zone_text_urls + complete_urls
                time.sleep(0.250) # Wait 250 milliseconds to be polite.
                print(f"Obtained forecast text urls for {geotype}/{geozone}.")
        return zone_text_urls

    def get_marine_zone_text_forecast(self,Point,zone_text_urls) -> (list,str):
        params = {'point':','.join(map(str,tuple(Point)))}
        data = requests.get('/'.join((self._nws_api,'zones')),
                            params = params ).json()
        features = data['features']
        if len(features) > 1:
            raise ValueError('Untested condition...')
        else:
            [zone_feature] = features
            id = zone_feature['properties']['id'].lower()
        txts = []
        urls = [u for u in zone_text_urls if id in u]
        for url in urls:
            txt = requests.get(url).text
            txts.append(txt)
        if len(txts) == 1:
            [txts] = txts
        return txts

if __name__ == "__main__":
    main()

The return forecast looked like this.

Expires:202205272300;;640662
FZUS56 KPQR 270903
CWFPQR
Coastal Waters Forecast
National Weather Service Portland OR
203 AM PDT Fri May 27 2022
Coastal waters from Cape Shoalwater Washington to Florence Oregon
and westward 60 NM
PZZ255-272300-
Coastal waters from Cascade Head to Florence OR out 10 NM-
203 AM PDT Fri May 27 2022
GALE WATCH IN EFFECT FROM LATE TONIGHT THROUGH SATURDAY
EVENING...
.TODAY...SW wind 5 to 10 kt with gusts to 15 kt. Wind waves SW
2 ft at 4 seconds. W swell 5 ft at 8 seconds. Chance of showers
before sunrise. 
.TONIGHT...S wind 10 to 15 kt, rising to 15 to 20 kt with gusts
to 25 kt after midnight. Wind waves S 4 ft at 5 seconds. W
swell 6 ft at 9 seconds. Chance of rain after midnight. 
.SAT...S wind 25 to 30 kt with gusts to 35 kt. Combined seas
12 ft with a dominant period of 8 seconds. Rain in the morning,
then showers and a slight chance of tstms in the afternoon. 
.SAT NIGHT...NW wind 20 to 25 kt with gusts to 30 kt, easing to
15 to 20 kt with gusts to 25 kt after midnight. Combined seas
12 ft with a dominant period of 10 seconds. Showers and a
slight chance of tstms. 
.SUN...W wind 10 to 15 kt. Gusts to 20 kt in the morning. Wind
waves W 3 ft at 5 seconds. W swell 8 ft at 9 seconds. Showers
likely. 
.SUN NIGHT...NW wind 10 kt. Gusts to 15 kt in the evening. Wind
waves NW 2 ft at 4 seconds. W swell 8 ft at 10 seconds,
subsiding to 5 ft at 10 seconds after midnight. Chance of
showers. 
.MON...N wind 5 to 10 kt. Wind waves 2 ft. W swell 4 ft. 
.TUE...N wind 10 to 15 kt. Wind waves 3 ft. W swell 2 ft.  
$$

[sullynole]

One current alternative for marine data is NDFD: https://vlab.noaa.gov/web/mdl/ndfd

[srstsavage]

Checking in on the status of marine forecasts in api.weather.gov. We're currently parsing HTML from the NDBC NWS pages, (example https://www.ndbc.noaa.gov/data/Forecasts/FZAK51.PAJK.html) and it's...awful 😂.

[sullynole]

If just looking for the CWF text product, that is available in api.weather.gov. Not sure if it's any easier to ingest, but at least more accessible: https://api.weather.gov/products/types/CWF/locations/AJK

The gridded forecast elements are not yet available though.

[srstsavage]

Ah, that's hugely helpful, thanks so much! It would of course be nice to have a JSON product with the forecasts split out into separate fields etc but this format is much easier to deal with than HTML. Thanks again!

[srstsavage]

Put together a simple wrapper API to parse the text forecasts to json fields here: https://github.com/axiom-data-science/nws-marine-forecast-api