search

CAPLTER / knb-lter-cap.636

dataset publishing: 10-m towers at LDP and DBG
0 stars 2 forks source link

rain gauge at LDP #2

Open srearl opened 6 years ago

srearl commented 6 years ago

Roy: "I forgot to mention that the rain gauge at LDP was vandalized (cover stolen, cable cut). There is an anomalous data point in the file on 4/20/18 (1.524 at 03:30:00) that I’m guessing was caused by the vandals, as I doubt there was any rain there last month, especially that much. So that point and those thereafter are not valid data IMO. Not sure how long the gauge will be out, have yet to hear from Sally who’s been in the field."

Because we do not know when this happened, I changed all precip. data since when the data were last downloaded (2017-12-19 11:30:00) to NA. We likely lost some good data with that approach but that is preferable to housing any erroneous data.

newdata <- newdata %>% 
  mutate(rain_mm_tot = replace(rain_mm_tot, timestamp >= '2017-12-19 11:40:00', NA))
srearl commented 4 years ago

This issue was revisited with version knb-lter-cap.636.6.

The rain gauge signal is lost at LDP on 2017-12-19 11:30:00, which is consistent with the cable being cut. The signal resumes at 2018-05-02 10:00:00, which coincides with a point of other maintenance activities at the site. However, the log book indicates that the rain gauge was not replaced until 2018-07-03. It is not clear why then the signal resumed on May 2nd if it was replaced on 3rd of July. Perhaps a stand-in gauge was added, or the data download or other maintenance activities prompted the Campbell to resume recording rain gauge data even though a bucket may not have been attached. There is no rain recorded at the site between May 2nd and July 3rd so it is hard to tell if the gauge was working or not. A cursory analysis of MCFCD gauges in the area for that time indicates that some sites received a very small amount of rain (6/16/2018, 0.03937 in., Bulldog Canyon) but not others (e.g., Wolverine Pass), which is not particularly helpful. A future version of this data set and this database should address proper storage of these data in a normalized data base form at which time a quality flag should be added and these samples marked accordingly. This issue will remain open until that time.