Size Distribution

[vperraud-design]

Hello,
I have this code that extracts and plot 2D size distributions from the Bankhead National Forest Station dataset. It works really well for the BNF site, however, I systematically get an error when I try to apply it to a different sites (for example SGP or EPCAPE). Could you help with sourcing the problem? (Note: I am able to plot different plots from the SGP and EPCAPE site from that same smps file (like a time series), but somehow the time index is not computing for SGP and EPCAPE). I am posting the code below to help troubleshooting.

Code For BNF (works):
import matplotlib.pyplot as plt
import numpy as np
import act

""Read the smps data (I will have uploaded it to my directory)""
smps_ds = act.io.arm.read_arm_netcdf('bnfaossmpsM1.b1.20260101.000459.nc')

""Clean up the data""
smps_ds.clean.cleanup()

""Create a DistributionDisplay object""
display = act.plotting.DistributionDisplay(smps_ds, figsize=(10, 5))

""Create a size distribution plot at a specific time""
t_ind = np.datetime64('2026-01-01T00:00:00')
display.plot_size_distribution('dN_dlogDp', 'diameter_mobility', set_title='Size Distribution (Jan 01, 2026)', time=t_ind, subplot_index=(0,), label='00h00')
display.set_xrng([10,800])
plt.xscale('log')
plt.legend(loc="upper right")

Same code For SGP (but this time, it doesn't work):
import matplotlib.pyplot as plt
import numpy as np
import act

""Read the smps data (I will have uploaded it to my directory)""
smps_ds = act.io.arm.read_arm_netcdf('epcaossmpsM1.b1.20230501.000000.nc')

""Clean up the data""
smps_ds.clean.cleanup()

""Create a DistributionDisplay object""
display = act.plotting.DistributionDisplay(smps_ds, figsize=(10, 5))

""Create a size distribution plot at a specific time""
t_ind = np.datetime64('2023-05-01T00:00:00')
display.plot_size_distribution('dN_dlogDp', 'diameter_mobility', set_title='Size Distribution (May 01, 2023)', time=t_ind, subplot_index=(0,), label='00h00')
display.set_xrng([10,800])
plt.xscale('log')
plt.legend(loc="upper right")

=> Key error: "not all values found in index 'time'"
(time=t_ind seems to be the problem, but it works in the code above)

Thanks!

[zssherman]

Hi @vperraud-design , thanks for sharing the code and files used. I can take a look and see if I can find the issue.

[zssherman]

I was able to reproduce the error, just trying to figure out the cause. I compared the time coordinates etc between datasets as well as the plotting variables, but so far can't find a difference between the working on non working datasets. Will keep digging.

[vperraud-design]

Hi Zach, Thank you so much for taking the time to look at this for us! Let me know if you need any other input. Best, Vero.

…

On Tue, Jan 13, 2026 at 1:43 PM Zach Sherman ***@***.***> wrote: I was able to reproduce the error, just trying to figure out the cause. I compared the time coordinates etc between datasets as well as the plotting variables, but so far can't find a difference between the working on non working datasets. Will keep digging. — Reply to this email directly, view it on GitHub <https://urldefense.com/v3/__https://github.com/ARM-DOE/ACT/discussions/978*discussioncomment-15489828__;Iw!!CzAuKJ42GuquVTTmVmPViYEvSg!LSR1TWzwkt67U7-9dlaldY1Mu8omElJENwiTC9YNrV2llk6uOtUjYFKmua0j6fc3F74HZOrc9Xe1i6H6JJLff1Be$>, or unsubscribe <https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/B4WJNSXFJS4LLJNFJGCR75L4GVRHVAVCNFSM6AAAAACRPSRFVOVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTKNBYHE4DEOA__;!!CzAuKJ42GuquVTTmVmPViYEvSg!LSR1TWzwkt67U7-9dlaldY1Mu8omElJENwiTC9YNrV2llk6uOtUjYFKmua0j6fc3F74HZOrc9Xe1i6H6JHnBmqkx$> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[vperraud-design]

I am not sure if this is helpful or not, but the following code works for both datasets, so I know I can read & plot both the BNF & the EPCAPE datasets, it's only the 2D size distribution code that seems to return an error for the EPCAPE dataset (and other sites as well, eg SGP).

Code:
import matplotlib.pyplot as plt
import numpy as np
import matplotlib.colors as colors
import act

""Read the smps data (I will have uploaded it to my directory)""
smps_ds = act.io.arm.read_arm_netcdf('bnfaossmpsM1.b1.20260101.000459.nc')
smps_ds2 = act.io.arm.read_arm_netcdf('epcaossmpsM1.b1.20230501.000000.nc')

""Clean up the data""
smps_ds.clean.cleanup()

""Plot up size distribution color plot for the two days
display = act.plotting.TimeSeriesDisplay({'SMPS':smps_ds}, figsize=(10, 5))
plt.yscale('log')
display.plot('dN_dlogDp', ylabel='diameter (nm)', set_title='BNF Size Distribution time series (Jan 01, 2026)', norm=colors.LogNorm(vmin=100., vmax=10000.), cbar_label='dN/dlogDp (cm^-3)')
display.set_yrng([10, 1000])
plt.show()

display = act.plotting.TimeSeriesDisplay({'SMPS':smps_ds}, figsize=(10, 5))
plt.yscale('log')
display.plot('dN_dlogDp', ylabel='diameter (nm)', set_title='EPCAPE Size Distribution time series (May 01, 2023)', norm=colors.LogNorm(vmin=100., vmax=10000.), cbar_label='dN/dlogDp (cm^-3)')
display.set_yrng([10, 1000])

[zssherman]

@vperraud-design I think I found the issue, I believe how we were handling times in distributiondisplay, we were removing times if a field value had a nan so when the plotting routine would retrieve the data using the index the times were being removed, however we wouldnt want to remove a time if we had nans in one field but not another. #979 should fix the issue.

[vperraud-design]

Great! I will look for the fix. Thank you for the help on this!