How is Crop Evapotranspiration Calculated when a PWS is reporting Solar Radiation and when the PWS is not reporting Solar Radiation?
I would assume Rachio uses algorithms to estimate ET. It is very accurate. Hopefully Rachio will respond.
If we do not receive solar radiation we derive it from temperature data, elevation, latitude, and time of year.
Our ET algorithm is based on this paper.
@franz How do we know if our PWS is reporting solar radiation? I’m looking at my local station’s pwsweather.com site and I don’t see anything on there listed as solar radiation. Does it have a different name on the site?
Currently pwsweather.com stations do not report solar radiation (Weather Underground doesn’t either).
Aeris (our weather provider) is working with pwsweather on this.
We investigated this further. Currently the PWSweather ingest API support passing solar radiation & UVI, but PWSweather.com isnt currently utilizing this data.
That said, we are working with the team to gain access to the data so that we can begin ingesting it into the Aeris API. We hope to begin this next week.
I will leave this ticket open and provide updates.
CWOP stations (start with MID_) currently transmit solar radiation data (if the PWS has the data).
I submitted my station to CWOP and now get solar radiation data.
We are rebuilding our entire platform (mobile/web/server) and for the new moisture graphs I want to show all the data used for ET (temp, wind, humidity, dew, solar, etc.)
@franz Is it the elevation/latitude of the weather station or the elevation/latitude of the home the Rachio is installed in?
We don’t have elevation of the weather station, so use latitude/elevation from the controller.
@franz Thank you for the responses! It appears the only way to get better data is to upgrade the PWS
The Penman-Monteith equation uses the following to calculate ET: solar radiation, wind speed, temperature and relative humidity. Rainfall, latitude
Longitude or elevation are not part of the calculation, nor is rainfall.
If you don’t have solar radiation as an input you need to derive it from latitude, elevation, day of year.
The last 6 ‘Crop Evapotranspiration’ numbers, for one of my zones, as reported from an MID weather station about 0.9 mi away from me, are:
0.15 0.13 0.14 0.13 0.12 0.12
Today, as a forecast value from Rachio’s weather forecasting system for my lat and long, it’s
And the next 7 day forecast values, again from Rachio, are
0.13 0.17 0.15 0.17 0.15 0.15 0.18
This illustrates the problem with Rachio I’ve observed all year long. Rachio’s forecasting of Crop Evapotranspiration is higher than what it ends up being from the MID weather station 0.9 miles away.
This causes my schedules to many times appear to have water days coming up, only to find at the 45 minute in advance of actual watering (at 12:15 AM) that the Current Moisture Balance has been recomputed using the actual for the day.
In the heat of the summer, switching from a forecasted to an actual Crop Evapotranspiration was a significant change for me every night.
I look forward to when the forecasted Crop Evapotranspiration numbers for my location are not so aggressively computed.
That makes sense; however if you have all the variables in the Penman-Monteith ET equation than of course elevation is not needed.
I have noticed this too I have my own PWS weather station and my raw data coming off my station is reporting evapotranspiration at .13 in per day but the Rachio controller is seeing .17 inches of evapotranspiration. I am set up through PWS weather with my own station but they don’t report the evapotranspiration so is my Raccio just estimating the .17 my stations actually reading.13 I do have a solar radiation sensor on my station
Also weather underground Now reports Solar radiation my station sensor shows up on Weather Underground
I’m adding on here. I see the same problem. Often forecast ET was around .18 ~ .20 through out the summer, but the observed ET ended up in the 0.12~0.14, and kept pushing out the day the sprinklers would run. Frustrating in my experience as my grass continued to dry out and brown.
I tried adjusting the Crop Coefficient and got to the point where it expected to water every other day. But with the discrepancy between forecasted and observed ET, it ended up watering every 3-5 days. I gave up and went back to Flex Monthly for much more stable watering schedule.
Thanks for sharing, thoughts on how to work around this issue? I’ve also seen a gap between forecasts and observations, but that’s the challenge when dealing with forecasted weather versus the actual observation.
Can you share the URL that is reporting the ET values? Would like to do some comparisons.
It’s an interesting problem. How much historical data do you keep? I see two potential solutions
You could calculate the bias for a particular station if you have history
– or –
Look at a running average, Exponential or Simple.
Although the most simple might be just using a persistence forecast.
Today I switched back to Flex Daily just to see what is going on for my yard. Currently my running average over the last 6 days is about 0.065. The forecast for today is 0.11. I suspect that when the algorithm runs, September 6th will drop to that 0.06~0.07 range. This is based on what I saw over the summer. Persistence forecast is pretty accurate for 1 day forecasts.
After I sent this, I realized it doesn’t quite solve my problem, although it might help.
My problem is that my yard consistently runs under the forecast for ET, meaning the days between watering get stretched out. I setup a schedule where it says it plans on watering every 3-4 days, but because the actual ET is always lower than the forecast, it stretches out to 5-7 days or more. My grass needed more water, especially during the hot summer weeks. If the ET were more accurate, perhaps I could tweak the values to get back to the 3-4 day range where it needed to be.
Now that I have completely hijacked the thread.
I believe I have figured the problem out. Here’s my thought. The Penman-Monteith equation is sensitive to the wind speed differences. Most PWS are installed in people’s backyards which have more obstructions and will report lower wind speeds than what a weather model would forecast for the area. With lower wind speeds the ET values are significantly reduced from the forecast. This seems to matter more than temperature differences in my experiments.
This results in the problem I see where day after day the actual ET value is much less than the forecast ( and much less than a true observation ). This results in less watering than is actually needed, and browning grass.
I use a PWS station (1 mile) closer to me because the closest official station (3 miles) does not report precipitation. Looking at wind speeds this week. I see the PWS is reporting wind speeds around 1/3 of what are reported at the official station.
Using multiple stations to get an average value or using an analysis grid such as the NOAA Real Time Mesoscale Analysis (RTMA) grid would probably help improve the calculation.