wrj0 Posted December 24, 2023 Posted December 24, 2023 Recently, I noticed my freeze warning programs were triggering off and on around the freezing point, even though I have 2 degrees of hysteresis in the programs. So, I created a webpage log to track the temperatures (CLITEMP) reported by OpenWeather (to log each reported temperature change), and was surprised to see how much the values change over a short time period - sample of the log below. 2023/12/24 00:05:27 46.35 2023/12/24 00:55:27 42.53 2023/12/24 01:05:27 45.19 2023/12/24 01:15:27 44.92 2023/12/24 01:25:27 45.19 2023/12/24 01:45:27 44.92 2023/12/24 02:05:27 40.73 2023/12/24 02:35:27 45.23 2023/12/24 02:45:28 40.73 2023/12/24 02:55:27 46.08 2023/12/24 03:05:12 42.48 Should I conclude that the local reporting station has some problems accurately reporting current temperature? And maybe move to another service that pulls data from a different source? Or is there something I can do with OpenWeatherMap?
bpwwer Posted December 24, 2023 Posted December 24, 2023 I agree that it seems like a lot of variance, but I don't know what to do about it. You could play around with how you list your location, that may cause it to use a different station. You could also compare with another service using the other service's website for the data instead of getting another plug-in just to test with.
DennisC Posted December 25, 2023 Posted December 25, 2023 You could try adding a slight delay to your program, say wait 2 or 3 minutes?
wrj0 Posted December 25, 2023 Author Posted December 25, 2023 2 hours ago, DennisC said: You could try adding a slight delay to your program, say wait 2 or 3 minutes? I appreciate the suggestion, but if the reporting station is really reporting this much variation in temperature, how would a delay in my program help? Meanwhile, I'm looking at other weather nodeserver/plug-in options.
DennisC Posted December 25, 2023 Posted December 25, 2023 23 hours ago, wrj0 said: I noticed my freeze warning programs were triggering off and on 3 minutes ago, wrj0 said: I appreciate the suggestion, but if the reporting station is really reporting this much variation in temperature, how would a delay in my program help? Looking at the data you posted, you can see the temperature dip and then within 2-3 minutes an increase in temperature. Adding a wait for 2-3 minutes, would prevent the program from triggering until the temperature dip stayed below your trigger temperature for the duration of the wait period. If the temperature increase during the wait period to a point above your trigger point, the If statement will re-evaluate as false, thereby, preventing a notification from being sent. After the wait period, if the temperature is still below your trigger temperature, the program continues to run and your notification will be sent. It's a way of minimizing the freeze warnings until the temperature is below your trigger point for x minutes. While in my example, I used 2-3 minutes, the wait can be any time period you choose. If you waited 5 - 10 minutes, or even 15 minutes, what would be impacted by having a temperature below freezing for a short period of time?
wrj0 Posted December 25, 2023 Author Posted December 25, 2023 (edited) Thanks again for your input. My polling is set to 600, so no temperature changes are provided to my programs in less than 10 minutes, as shown in my sample above. I'd really like to know the reason for the sudden changes in the reported temperature before trying to mask a possible underling problem with the accuracy of the temperature data. I'll spend some time to look more closely at the data from source station. OpenWeatherMap indicates they use a numerical weather prediction (NWP) model to do some manipulation of data from several sources to improve quality and accuracy. Curious if this may be the cause of the variations in temperature data that is reported. Edited December 25, 2023 by wrj0 typo
Ross Posted January 5 Posted January 5 @wrj0 There are many ways to capture temperature at your location. You may want to move away from a service like OpenWeatherMap. Yolink has a great temp sensor (you also need a Yolink hub) and there is a good Plug-in for it. And the batteries last for ages. There are other cheaper and much more expensive options (including having your own weather station). I still use OpenWeatherMap for forecasts and rely on it to determine whether to run my sprinklers or not. Ross
wrj0 Posted January 5 Author Posted January 5 Thanks for the suggestion, @Ross. I'm still looking at options, including Yolink. Still, it seems curious to me that the weather data from my local station (a Marine Corps Auxiliary Field) varies so widely. I've loaded two other weather plug-ins and am looking at weather data from other sources to see what may be the most consistent for my location.
Recommended Posts