I have a GRIB file from NOAA that contains several variables for a specific subregion in Norht-East Atlantic.
I'm using ECMWF grip-api Python library to parse the file. While I'm able to get the values for each {latitude, longitude} pair for each GRIB message with this code:
def iterate_latitude_longitude():
f = open(INPUT)
while 1:
gid = grib_new_from_file(f)
if gid is None: break
iterid = grib_iterator_new(gid,0)
#Get the missingValue for this GRIB message (e.g. NaN)
missingValue = grib_get_double(gid,"missingValue")
i=0
while 1:
#Get the value from the geo iterator obtained previously, value is a tuple with [latitude, longitude, value]
result = grib_iterator_next(iterid)
if not result: break
[lat,lon,value] = result
sys.stdout.write("-Point # %d - lat=%.6f lon=%.6f value=" % (i,lat,lon))
#Check if missing value is present or not
if value == missingValue:
print "missing"
else:
print "%.6f" % value
#increase iterator
i += 1
grib_iterator_delete(iterid)
grib_release(gid)
f.close()
While this is great, this file contains forecast data, it was extracted from this place, more precisely, this file:
multi_1.nww3.t00z.grib2
Which contains world data. There are other files for different times: 00-60-12-18-00, thus, every 6 hours.
What I'm missing is the relation between the value obtained for each latitude,longitude and the time dimension. Does anybody know how to extract this info from a grib file, whether using Python or Java?
I know I can use things like wgrib2 or pygrib to convert to CSV and do the other way around…but it just looks so inefficient
Best Answer
You probably didn't want to just iterate over the whole file, but rather iterate over each of the (interesting) values, for each of the time steps you want.
Here is some code adapted from the sample code
And some sample output:
Note that per your comments on the other answer, the results are in 3 hour blocks (not 6 hour as noted in the question).