I am wondering if it is possible to specify the time range for past data in GFS, NAM, RAP, HRRR, and the NDFD forecasting models! the current option is recalling data for the present daya as stated here
`start = pd.Timestamp(datetime.date.today(), tz=tz)`
end = start + pd.Timedelta(days=7)
The source code for the model.get_data() function indicates yes. The only stipulation is that the start and end arguments must be datetime or timestamp datatypes to be compatible with the netcdf2pandas library
I'm trying to import data from SAS exports posted publicly. I am not using SAS to import these files, but importing them via custom script into another format. There are no specifications that I could find, but it seems pretty self-explanatory if you're used to databases.
And then I found a dictionary file that specifies the last three columns as having lengths of 11.5, 12.5, and 9.5 characters. This makes zero sense to me.
#2682 PATWT 11.5 /* Patient visit weight used for national, regional, and divisional estimates */
#2693 PATWTST 12.5 /* Patient visit weight used for state estimates */
#2705 PHYSWT 9.5 /* Weight used for physician-level estimates based on responding in-scope physicians seeing */
An explanation of this, or being pointed to official SAS specifications for this format would be extremely helpful.
11.5 is an IN-format specification.
I’d like to use pvlib library to calculate irradiance POA data for a single axis tracker system.
From the documentation it appears that this is possible, by creating a pvlib.tracking.SingleAxisTracker class (with the appropriate metadata), and then calling the get_irradiance method.
I've done so as such:
HSAT = SingleAxisTracker(axis_tilt=0,
axis_azimuth=167.5,
max_angle=50,
backtrack=True,
gcr=0.387)
I then use the get_irradiance method of the HSAT instance of the SingleAxisTracker I just created, expecting it to use the metadata that I just entered to calculate POA data for this Horizontal single axis tracker system:
hsat_poa = HSAT.get_irradiance(surface_tilt=0,
surface_azimuth=167.5,
solar_zenith=sz,
solar_azimuth=sa,
dni=dni,
ghi=ghi,
dhi=dhi,
airmass=None,
model='haydavies')
When I go to plot hsat_poa, however, I get what looks like POA data for a fixed tilt system.
When I looked at the source code, I noticed that the SingleAxisTracker.get_irradiance method ultimately calls the location.total_irrad() method, which only returns POA data for a fixed tilt systems.
Do I need to provide my down surface_tilt data from the HSAT system? I had assumed that pvlib models an HSAT system, and would generate the surface_tilt values for me, based on the arguments provided in the SingleAxisTracker class instantiation. But it appears that's not what happens.
So my question is does pvlib require the tracker angle as an input in order to calculate POA data for Single Axis Tracker systems, or can it model the tracker angle itself, based on metadata like axis_tilt, max_angle, and backtrack?
Turns out pvlib.tracking.singleaxis() is the missing link.
This will determine the rotation angle of a single axis tracker system.
tracker_data = pvlib.tracking.singleaxis(solar_position['apparent_zenith'],
solar_position['azimuth'],
axis_tilt=MOUNTING_TILT,
axis_azimuth=MOUNTING_AZIMUTH,
max_angle=MAX_ANGLE,
backtrack=True,
gcr=MOUNTING_GCR)
and then using tracker_data like so:
hsat_poa_model_tracker = HSAT.get_irradiance(surface_tilt=tracker_data['surface_tilt'],
surface_azimuth=tracker_data['surface_azimuth'],
solar_zenith=solar_position['apparent_zenith'],
solar_azimuth=solar_position['azimuth'],
dni=dni,
ghi=ghi,
dhi=dhi,
airmass=None,
model='haydavies')
will calculate POA data for a single axis tracker.
Found the answer in this jupyter notebook:
http://nbviewer.jupyter.org/github/pvlib/pvlib-python/blob/master/docs/tutorials/tracking.ipynb
Can it model the tracker angle itself, based on metadata like
axis_tilt, max_angle, and backtrack?
pvlib's ModelChain will do this. See the PV Power Forecast documentation for an example of using a ModelChain with a SingleAxisTracker.
I am new to machine learning and working on a project in which I collect environment data like temperature, humidity, dust, light intensity , Carbon mono oxide and rain data through sensors and send it to cloud. Now I want to generate alerts on the basis of conditions to occur in coming days. What machine learning algorithms should I use for this? I was working on neural network but can we predict the temperature on the basis of temperature past days ? I tried to do that but I tried to take temperature as X(input) and what should I take "Y"(output matrix)? Can neural network be only used when some X variable is dependent on some OTHER Y variable? Which algorithm would be most suitable for my work? Thanks in advance.
Weather forecasting and alerting requires huge simulations.
I suggest you first study meteorology. Then you buy a supercomputer (some if the largest are used in weather forecasting, for a reason) and run your own climate models.
This may be a niche question but I'm working with ICU to format currency strings. I've bumped into a situation that I don't quite understand.
When using the MesssageFormat class, is it possible to get access to the NumberFormat object it uses to format currency strings. When you create a NumberFormat instance yourself, you can specify attributes like precision and rounding used when creating currency strings.
I have an issue where for the South Korean locale ("ko_KR"), the MessageFormat class seems to create currency strings w/ rounding (100.50 -> ₩100).
In areas where I use NumberFormat directly, I set setMaximumFractionDigits and setMinimumFractionDigits to 2 but I can't seem to set this in the MessageFormat.
Any ideas?
Probably because one Korean Won is worth $0.00088. They don't have pennies.
Same idea with the Swiss Franc, the 1 centime coin stopped being legal tender in 2007.
I've determined that gaining access to the internal formatter used is not possible. I've opened a ticket with the ICU project. http://bugs.icu-project.org/trac/ticket/7571#preview