5-3 Adding Wind

[grhyne]

When adding wind to the grl graph (5-3-wind-graph_create), I have hit this error regarding NA values only for tags CB596, CB599, and CB616:

> # Add wind
> grl <- graph_add_wind(grl,
+   pressure = pam$pressure,
+   filename = paste0("data/5_wind_graph/", gdl, "/", gdl, "_"),
+   thr_as = gpr$thr_as
+ )
[==========================                        ] 363925 / 696923 | sta = 2/10Error in ncdf4::ncvar_get(nc, "u", start = c(id_lon[1], id_lat[1], id_pres,  : 
  Error, passed a 'start' argument that has NA values: NA NA 2 1

I tried adding on an na.omit() to the graph_add_wind() function, but that did not seem to change anything. Which id_lon[1] and id_lat[1] is the error referring to in grl?

[Rafnuss]

Is the repo up to date? Can a use the data there to reproduce the issue?

[grhyne]

Yes I just committed all my recent updates, go ahead and try

[Rafnuss]

Hi @grhyne ,
I managed to track down this error. This is due to an error in labeling with a flight of a single timestep. Have you performed all checks suggested in 1-pressure.R? One of them should have pick up on this.
You can see that it has three flights (CB616), one with no duration.

> metadata(static_prob[[2]])
$sta_id
[1] 2

$nb_sample
[1] 13

$max_sample
[1] 300

$temporal_extent
[1] "2021-10-17 11:12:05 UTC" "2021-10-17 23:12:05 UTC"

$margin
[1] 30

$flight
$flight$start
[1] "2021-10-17 23:42:05 UTC" "2021-10-18 08:42:05 UTC" "2021-10-18 11:12:05 UTC"

$flight$end
[1] "2021-10-18 06:42:05 UTC" "2021-10-18 08:42:05 UTC" "2021-10-18 11:42:05 UTC"

$flight$sta_id
[1] 2 3 4

This is the currently label file and you can see the 1hours flight.

Here would be my suggestion for this file:

[grhyne]

Thank you for finding this! Ok I see, I think I was going conservative on labelling flights based on pressure spikes, and assumed those individual labels were short (~20-30 min) flight, which would only be detected by one point. I understand how you would label them, and it worked smoothly. It should have been picked up by the tests in 1-pressure, but I believe I had left the debug <- FALSE from the previous tag and thus did not detect the mislabelling.