Should timestamps be interpreted as the midpoint of the time interval, beginning, or end?
For some background, I have a Tempest mounted ~10 ft above my roof line in a suburban neighborhood and I’m pairing my weather measurements with a few barometric pressure sensors from Gulf Coast Data Concepts to analyze the internal pressure in my house relative to atmospheric in a variety of weather and opening configurations (garage door open vs closed, one window, two windows, etc). I’m logging the internal pressure data at 5 Hz. I’m using IFTTT to log Tempest data in 1-minute intervals, then using a regression model to adjust station pressure to an expected internal pressure to account for altitude differences, minor calibration errors, etc. When I go to calculate a gage pressure, I’m not sure how I should be interepreting the timestamp to compare against the higher resolution measurements. Further, if I were to use WU or Datascope to grab the weather data archive rather than IFTTT, the data interval is even larger and its still not clear whether the timestamp represents the midpoint, first point, or last point.
Happy to have any thoughts or suggestions on this approach. I’ve demoed it for a couple small storms so far and it seems to be producing reasonable data. I realize it’s not perfect, but I’m interested in demoing a framework for using a relatively inexpensive sensor setup to do some worthwhile science.
According to this Data archive buckets explained pressure is always the pressure at that exact point in time, except when watching a graph that is completely zoomed out, then it is an average (I assume the average from the 3 hour measurements or from all the 1 minute samples, but that isn’t specified…yet).
For your analysis, if the pressure is sampled every minute, it would be fair to assume that it would be the best representation for the actual pressure during the interval from half a minute before a specific time, to half a minute after that specific time.
As the time the data is sent from the Hub. If you are using the data from the WeatherFlow servers the time is rounded to the nearest minute. That is why you will occasionally see two readings in the same minute.
I have been analysing my wind readings compared to weather underground because wu reports different values to weatherflow. If I look at the wu graphs they are different from the wu table. So i dug further and i suspect while wu displays 5min values it does not use the full 5 min of weatherflow data to calculate its values. Wu contains 6 different wind values for the same period and when i try to match them up to the values reported by weatherflow i have not been able to correlate them.
But i have not examined the differences with pressure.
I have not reported my findings of differences to wu anywhere to try to fix it.
All I have done is learned not to believe my weather underground page for accurate data.
Cheers Ian
