Flex schedule after unforecast rain


#1

We had thunderstorms come through last night which the nearest NWS reported we received .54" of rain from. Originally this was forecast for yesterday (2nd), but didn’t end up coming through until early this am. My question is, does the Iro no longer look back at actual precip before running a flex schedule like it did with fixed schedules? I understand that since it happened today, the moisture graph, and I’m guessing the Iro as well, are still in forecast mode, but it seems like if enough rain falls between midnight and the next start that it should take that into account. I got the notification 45 min before start time that my scheduled zones would run, even though half an inch of rain would have been more than enough to fill those zones. My rain sensor canceled the watering, but I was just wondering how this works with flex schedules…

As a follow up, now my moisture graphs show that those zones watered today since it was the rain sensor that skipped them. Will that update tomorrow or will I need to use the upcoming fill zone feature to fix this?


#2

it will look, it will also look at the future weather. It should take that into account.


#3

@Brg468 Same issue here. You may want to check this thread to make sure your Iro is getting the correct weather info:

http://community.rachio.com/t/selected-weather-station-shows-precipitation-received-rachio-says-0/1769/24?source_topic_id=1892

#4

I had the same issue with my fixed schedules this morning. We received about 0.75 in of rain late last night and early morning, but two different schedules ran this AM. I placed a support ticket this morning so hopefully this can be looked into.


#5

We started out using an observation for today but that since most schedules run early morning, we were missing any forecasted precipitation. The observation data changes throughout the day for today. It’s a difficult call because it can go both ways. If we only observe 1/16in precip since schedule running early and a few hours of observation, do you just add the forecast, which could be significant? Definitely looking for enhancements. It is currently using observations from yesterday on, and forecast for today. There is opportunity for a hybrid approach, just don’t know what that looks like.

If these were skipped, there will be no official recorded irrigation. That must be forecasted irrigation and tomorrow it will assume no water was applied, and will read the observation from your weather station to apply precipitation.

Hope this helps.

:cheers:


#6

Just thinking, but is there a way to look at the past 12 hour history and upcoming 12 hour forecast (or something to that nature) to determine if watering is needed?


#7

Observations and forecasts don’t really work at the hourly level, unless the observation is today you by default know how long it is for. So, you are left to some hybrid approach which has pros/cons. Any hybrid approach feels like it would help some, but hurt others. If anyone has some creative ideas let us know :wink:

:cheers:


#8

Thanks for the response. I just wish there was a way to capture the real time data between midnight and the scheduled run time. This is the second time that I awoke to have my sprinkler system running while raining.

One of the main reasons for purchasing the IRO was to get away from an external rain sensor, though it looks like it may not work properly without one.


#9

I have multiple Flex schedules, each with its own designated start time. How often does the system keep weather updated? Also believe the best method for having the correct schedule created, would be to have the system have an update performed just prior any scheduled event. This would allow the system take into the account any weather history that is pertinent to the scheduled event.

Watering based off of forecasted weather can be a risk proposition in certain parts of the country where rainfall can be quite different from 1 location to another within a very small area. Once a watering event is missed or stopped you can’t get it back. I’d rather make the savings up on the back end.


#10

We do have that observed data if it has been reported to our weather provider by the time the schedule is going to run.

We will continue to figure out how to incorporate observed data with forecasted data before the schedule runs.

What if the observed precipitation was > forecasted precipitation, that might be an easy way to defer one over the other?

:cheers:


#11

As frequent as our weather provider, which usually can get down to hourly.

We currently look at all available data one hour before schedules run.

Agreed, but if your schedule starts @ 4am, we only have maybe 2 hours of observed data for that day, and if the forecast calls for significant rain we would always miss it. There has been arguments on both sides for sure. It’s a tough balance from overwatering to underwatering. Still trying to find the right balance, and open to suggestions.

:cheers:


#12

One hour before scheduled event start is great! The reason for needing the rain sensor as a crucial part of your system is for just what you mentioned. If the rain actually falls the event is interrupted. The rain total and run event precipitation is add to the numbers and watering will resume once enough ETo has been lost. One needs to remember not all rainfall is beneficial, there are a lot of variables. c Very difficult to predict, at least in some parts of the country, In other forecasting can work great. Agreed it’s a difficult balance!


#13

Agreed, a $20 rain sensor is cheap insurance for sure.


#14

I think this should work for fixed and flex schedules, by just using observed data if it has a precipitation level for today greater than the forecasted data, no matter how small the observed window has been. Can probably get this in over the next day or so…

:cheers:


#15

I would agree. Statistically, I would think the accuracy of observed data should always be used to override forecasted, if possible. However, one catch could be in the accuracy of the observed data. Junk in = junk out.

A possible solution to this issue would be to allow the user to utilize multiple sources of observed data (rather than just one) in some way to leverage the law of averages to obtain better data.

No matter, I think you are moving in a smart direction.


#16

This may be a viable solution to this particular problem. Is it fairly easy to implement?

The problem with a rain sensor is that it interrupts the schedule no matter how much or how little actual precipitation is actually received. It my interrupt a schedule though the amount of rain may be far less than your set threshold. To me this negates the whole purpose of an ‘intelligent’ controller.


#17

What I like about the rain sensor is that it turns watering off if the system is already watering once enough precipitation has occurred to trigger the sensor.

What I believe these does is, it gives the ‘intelligent’ controller time to recalculate, react and implement new schedules based on actual data. The system might decide not enough precipitation had fallen so irrigation continues at the next closest time slot or enough rainfall occurred to push watering down the road a few cycles. (Still Intelligent watering, Just basing decisions off actual observed data).


#18

Several external rain sensors won’t trigger unless a certain amount of precipitation has fallen. This can usually be set with a physical switch on the rain sensor (1/8 inch, 1/4 inch, 1/2 inch of precipitation, etc.… )


#19

Yes, will have fixed schedules finished tonight, flex in the next day or so.

:cheers:


#20

I really like this idea, I think it should fix exactly what happened last night. Well done guys.