wmbelk/weather

Baysian Update to environmental data

Weather

Baysian Update to environmental data

weather's Issues

lat and long prior too tight

update to atleast .25
Lat_effect_temp = pm.Normal('Lat_effect_temp', mu=0, sigma=.25)
Long_effect_temp = pm.Normal('Long_effect_temp', mu=0, sigma=.25)

don't downsample ballon

predict from a reload

need to save off trace and model for reload and predict

Trajectory thru photruth

Need to kinds.

• One where it goes through the same space multiple times – confirm get same environment
• One where it traverses down in altitude the changing altitude diminishes

Time effect

• Get truth detection good for lat and long
• Add time effect to temperature and pressure
• add seasonality to model

Xarray variable values extraction

What is the best way to get variable values out of x-ray for an array of ordered pair coordinates?

• .sel with a dictionary?
• A query function?
• Merge?
• Something else?

Where are the coordinate values are not exact

• Is there a close to or about function?

plot order - fix them

quadrent plots:
for i, ((N_S_label,N_S_idx), (E_W_label,E_W_idx)) in enumerate([[i,j]
for i in [('North',idx_north),('South',idx_south)]
for j in [('West',idx_west), ('East',idx_east)]]):
print(N_S_label,E_W_label)
ax[i].set_title(f'Lat: {N_S_label} Long: {E_W_label}')
idx = np.intersect1d(N_S_idx,E_W_idx)
az.plot_forest(idata2.sel(alt_lat_long_time=[x for x in idx_sortby_alt if np.any(np.isin(idx,x))]),
var_names=['mu_t'],
kind='ridgeplot',
combined=True, ax= ax[i],
labeller=labeller
)

all forest:
az.plot_forest(idata2.sel( alt_lat_long_time=idx_sortby_alt), var_names=['mu_t'],kind='ridgeplot', combined=True,labeller=labeller)
az.plot_forest(idata2.sel( alt_lat_long_time=idx_sortby_alt), var_names=['mu_p'],kind='ridgeplot', combined=True,labeller=labeller)

comparison:
az.plot_dist_comparison(idata2, kind='observed', labeller=labeller)

Lat is Lat; Long is Long

Check that lat (and long) dimension is carried correctly through our code.

get rid of unused libraries

Measurement error on temperature recording

Xarray groupby construct

Can you make new dimensions with group y using construct

Truth detector

• Make sure get final truths
• Save off truths better
• Report truths against predictions

PREDICT temp and pres at location

Does the prediction interval contain 'actual' temp and pressure

check out real temp data from NASA

https://ncar.github.io/esds/posts/2022/xarray-groupby-vs-geocat-climatology/

more:
https://ncar.github.io/esds/

ds = xr.open_dataset('603321.sst.sst.mnmean.nc')
ds = ds.sst # Pull out the sea surface temperature data
ds = ds.isel(
time=range(1, 49)
) # Remove the first data point so that we have an equal number of data points from each month

plot interactive and explore

try out plotly bokeh and holoviz