Test and validate performance of soil moisture analysis

[fedjo]

• Create input datasets and perform irrigation analysis. You can use a script or an online generator or on line LLM to create them,
• Validate that the analysis is performed correctly. You can use visual validation. For example, plot dataset in a timeline graph and check if the irrigation events have been found correctly.
• Input test datasets can be also created in CSV/XLS format in order to be easily shared

[fedjo]

@vlf-stefan-drobic & @prske I managed to retrieve fine-grained datasets from SIP4. You can use these dataset to refine/tweak the soil moisture analysis.

The API call to get them is

GET http://sic.openagri.psnc.pl:8001/api/data/measurements/?from=2025-04-01T15:00:00Z&to=2025-07-17T15:00:00Z&device_ids=lpit-sensor-1

use can also use device_ids=lpit-sensor-2 or 3 etc

I am adding the dataset for lpit-sensor-1

telemetry_combined-lpit-sensor-1.csv
telemetry-lpit-sensor-1.json

Here are the ploted graphs for

soil_moisture1_percent

soil_moisture2_percent

[fedjo]

@prske please provide some updates regarding your current working issue. You could also give us some insights about the algorithm that you will use for generic analysis.

[prske]

Hi @fedjo,

With new algorithm tested on dataset received output from below.

poland.json

Right now, I'm using some statistics calculation with weights assigned for each depth. In that way, user does not need to have all six depths in his/her dataset.

I think I will finish algorithm next week before I go to vacation.

Regards,
Aleksandar