Reactions to : Introducing RainCheck

Here in Rep of Panama (https://tempestwx.com/station/19591/) I am finding that the rain part of the Tempest flat tops at about .62in/hr in its ability to ready heavy rain. It is not uncommon here in the tropics to exceed 2in/hr rain fall. I have submitted a repost to WeatherFlow on this issue. Attached is a link to a spreadsheet of heavy rain period from the last two days for inspection. You can easily see the issue in the chart.

DataSpreadSheet

comments or thoughts appreciated. Craig

@cmowings you might be looking at the wrong graph. From the startup screen, click the icon in the top right corner. You get to a blue page with lots of real time date. There is one that shows the rain. Click on that and you will get a graph. Now in the top left of that graph is a drop down menu. Select rain rate from that.

Good afternoon Sunny,

I am looking at the correct graph as you explain. However it is easier to see the issue with tabulated data. The problem is the calibration of the Tempest rain sensor. At a rain rate of approximately .4in/hr the rain sensor starts under reading the rate and ergo the accumulation. Look at the attached spreadsheet that has the data from today’s midday rain. Also on the sheet is a graphic of the Tempest direct reading graph.

So the issue becomes, how do we correct the the hepatic rain sensor to read a a rain rate of 0 to 3 or 4 inches per hour of rain intensity.

thanks,
Craig Owings

DataGraph.pdf (65.9 KB)

i’m confused. If I look at your stations data at 1:45 for today I see a value of 1.24 in/hour in the graph, not what is in your table at 0.55 in.
edit: sorry I was zoomed out a bit. but at 1:43 pm today I see
a rate of 1.54 in/hour. That is well above the problem rate you were talking about

Sunny,

I see what you are talking about. I went back a looked again. It appears that the difference is between what Tempest is showing in the realtime graph (https://tempestwx.com/station/19591/graph/66551/rain/2) and what is being reported to Weather Underground. Relook at the newly attached sheet. I have added and highlighted the differences in two data points.

Additionally the other large difference is in the accumulated rain when compared to the WeatherHawk station near me. That station uses a dump bucket accumulation method. Again this is showing up in the Weather Underground charting of the rain. It appears that the date is not being transferred correctly in addition to the actual difference in the rain measurements by two close stations. I also have a manual rain gauge on order for actual verification at my site. Though I doubt that there is about a 100% difference in the readings from the two stations caused my actual rain fall.

Please take a look at this, and verify my findings.

Thanks,
Craig

DataGraph2.pdf (1.68 MB)

I just went to bed and the pdf won’t open on my phone. What I just guessed, but verified on the Web, the rain rate is never send to WU. WU calculates the rates based on the accumulated data (which it does receive) and the time between the data.
WeatherFlow gives you a measured rainrate every minute.
One question you have is why your data differs from a not too far away station. That’s a hard one. Better compare it with your normal gauge.

i looked at your table, but i don’t know how to read it. There are a whole bunch of times in the 4th column ending in 4 or 8 minutes (so they are separated by 4 or 6 minutes. In the 5th column you have some numbers. Where did you get them from? Then there are some times and numbers highlighted in blue, but taken at a different time interval. Where did you get those from. I can’t find a match between the date in your pdf and the graph of your station. But I’m sure I’m not understanding what I’m looking at.

Please note that looking at 1 minute reports and writing down every 5th number is different than reporting rain rate every 5 minutes. Rain rate is defined as amount of rain per hour. You can report it every minute, but then you have to take the amount of rain during that minute and multiply by 60

Good morning Sunny,

Interesting how WU collects and calculates. So this make the accumulation data of Tempest critical to the displayed data on WU. Is there a way that I can see/record the raw data from the Tempest, rather than in a graph form? I will then try to figure out why WU is showing an under counted accumulation.

Sorry about catching you at bedtime last night. Have a great Sunday.

Take care and stay well,

Craig

the hub broadcasts the value. There are third party applications that catch that data and some could probably store the data for you. search for them in this forum.

There is a real easy method. On your phone, install a utility that listens to UDP broadcasts. I use this:

It’s quick and simple. You can see the data from the Hub and save it for further use.

Let me know if you have questions.

1 Like

Look at the Third Party Integrations section for many such pieces of software. Weewx is my go-to solution, but there are quite a few others.

Be aware, however, that you can’t presume that WU not showing data means that the Tempest is wrong. WU is notorious for missing readings (and has basically zero customer support).

I’m not certain that the Tempests have gotten all their continuous learning calibrations yet. That would be a question to ask @WFstaff

2 Likes

Thanks for the info/ideas Gary and Vince. I did a connection to a Google Sheet, and it now records the Tempest output every minute. Gives me the info I need to see if the rain fall amount is accurate. Will be comparing to a manual rain gauge at my station and to a nearby very good gauge. The issue that I am seeing is that the hepatic sensor cannot handle the heavy rain we have here in Panama. 2-4in/hr is not uncommon for some intervals. I am attaching a PDF of two stations near each other. One with an expensive, calibrated bucket rain gauge, reporting 5min intervals and the Tempest reporting 5min intervals (data from WU). In the red outlined area, you can see as the rain intensity increases, in the heavy rain, the Tempest in not correctly recording the rain accumulation. I understand that the precip rate calculated from the precip amount over time, that is amount/time=rate.

Now that I have the Tempest data recording to a Google Sheet and to WU, I can compare the numbers for transmission accuracy.

I do believe that the Tempests will have to have rain recalibrated to heavier rainfalls that is typical in the USA or Europe to function correctly in tropical downpours.

Craig

DataGraph2.pdf (1.68 MB)

I think this is a very known issue…

Here’s my graphs from yesterday.

  • VP2 = 1.10"
  • CoCoRAHS manual gauge = 1.18"
  • Tempest = 2.01" before rain check
  • Tempest = 0.78" after rain check (no idea how they decide that adjustment)

I wrote one of the 3rd party integrations (WeatherFlow UDP listener and MQTT/InfluxDB publisher utility) and run mine feeding influxdb. I then did a Grafana dashboard to display that data vs. my weewx VP2 data (which takes a somewhat different path to get into influxdb).

One more voice on this. My Sky recorded 3.26" from yesterday’s storm (as of midnight, before RainCheck). My physical rain gauge collected slightly over 1.5", and the official total for my town was 1.78". RainCheck is showing 0.76". I’m finding the Sky pretty consistently reports about twice the actual amount, then RainCheck corrects it to something close to reality (not sure what’s up with the current 0.76 measurement). My understanding was that RainCheck is used to help the system self-calibrate during a rain event. If it’s just replacing what my Sky reports with values from elsewhere, then there’s no point in showing amounts during a storm.

1 Like