(Note: These are our internal graphs used by data scientists to debug data science bugs. M_1, M_2, M3 refer to “soil moisture at depth 1, 2, 3”)
Device 3343U3: Soil Moisture showing wild swings from very wet to dry within a day.
- Thanks for catching a bug for us!
- After some investigation, your device triggered a very unusual bug in our AI model internal parameter. You’re correct the soil moisture should only be changing from “very wet” to “wet” from day to day. I’ll provide an ETA soon on when this bug would be fixed!
.
Device 337J9T: Your device seems to be working as intended. Your soil was quite loose before your first deep water calibration, and looks like the soil is getting more compact over time.
- The surface level is quite loose though, and AI may potentially complain down-the-road that the surface soil may be too loose for it to accurate track moisture.
- But if your primary use is to monitor mid-depth or deep soil moisture, then the installation looks good. Please give it another few days for gravity to help compact the soil further.
.
Device 333XYP: The device is actually looking really good, very stable readings, and GeoDrops AI is working properly for your device.
- I’m actually curious, are you sure the soil location that this device is installed in is actually wet?
- I have double checked the Raw hardware readings. It’s very stable. No sign of loose soil at all.
- If you can help confirm that the soil is actually wet, and that this device is giving a false reading, then I would love to troubleshoot this with you further.
- In very rare cases, this could be defective hardware (Although based on all the device signals I’m seeing, it looks like the hardware is quite healthy as well).
- Please let me know, thanks!
.
Device 335LDS: Inverted reading between soil depths. Your device needs a bit more time to settle.
- Yes, you’re correct that the reading is wrong. We’re still adding more capabilities in GeoDrops AI to better make decisions and help guide people regarding soil installation.
- What you’re seeing is that the entire middle sections of the sensor probe are actually in poor contact with the soil. Can you following the installation tips in my first post? We forgot to mention in the installation procedure to NOT install into the same soil spot repeatedly
- The problem with loose middle-depth soil is most commonly seen when people install the device into the same spot more than once.
- In extremely rare cases, there could be foreign objects in the middle area of the probe. Although your device looks a lot more like poor/loose install and not foreign object.
.
Device 334KN9: Loose soil installation, likely due to repeated installs at the same location.
- This poor reading is typically caused by inserting the device into the same soil location multiple times.
- For your soil readings, I’d suggest moving the device to a new location in your soil (at least 5 inch or 15cm away from the current installation location).
- Pick a new undisturbed soil location, and try to install the device in a single attempt as much as you can.
- Lastly, when performing deep watering calibration to let GeoDrops AI detect max soil moisture level, it’s best to either turn on your irrigation system as you would normally do, or if you water by hand or via a water hose, use a shower head. Reason for this is that it looks like watering has ended up washing more of the already loose soil away, so now the probe has even worse soil contact.
Just remember that capacitive soil sensors work by having a solid contact between the soil and the sensor probes. The better the initial installation (by not disturbing the soil, keeping it as stiff and tight as possible when inserting the sensor into the ground), the faster the reading will stabilize.