What is a field boundary?

[m-mohr]

I think we should start a discussion about what a field boundary is - or there should be a type-like field for boundaries in core.

I just stumbled across a dataset (Thuringia, Germany) that included (in my view) more than just agricultural fields, e.g. forests.

To make sure that we don't mix things up, we should try to define something that guides a user.

[petescarth]

Good discussion. We're into mixed farming (Cibo Labs) and define a "field" as a management unit that's usually inside a "paddock", and the paddocks are mostly (given surveying errors and image errors) within a "cadastral boundary". This, I suspect, is a local definition.

Paddocks can be large—some in our system are ~90,000 ha, so close to the Fiboa spec max size—but they are relatively stable over time, whereas our field bounds can change multiple times/year.

This definition differs from what I commonly see elsewhere, where "field" and "cultivated area" distinguish the fenced/surveyed bounds from the management.

Either way, some way to distinguish these two boundaries with say, a type-like field is essential.

Other definitions use "fields" but include landUse and grazableArea to inform that there are ungrazable areas such as forests, dams etc.

Also, it's worth looking at Datalinker-Org for some thinking on this - they define a few farm data classes.

It would certainly be helpful to have a consistent definition for traceability and conformance that could fit into UN/CEFACT.

[StefanBrand]

Thank you, @m-mohr for pointing me to this discussion. I had added a viewpoint in https://github.com/fiboa/specification/issues/28#issuecomment-2089987957 and here are responses to your questions:

Would it be meaningful to split the MultiPolygons that you get into single Polygons?

Not from the authority's perspective: These parcels are managed by the same farmer and have the same crop (mostly grassland). The farmers receive subsidies for the whole parcel, not for subparts.

From a satellite monitoring perspective it might make sense: The farmer might have their sheep on one side of the hedge in one week and on the other side of the hedge in the other week. Then we are talking about "partial grazing", and splitting the parcel up would result in more fine-grained monitoring results, i.e. subpolygon 1 was grazed during CW 19 and subpolygon 2 was grazed during CW 20.

Who assigns the IDs, i.e. is it possible to have unique IDs per Polygon after the split?

No, the IDs are assigned by the authority based on a farmer's subsidy application. From their perspective, the subparts of the multi-polygon still belong to one entity.

Do they regularly have different crops planted?

I'm mostly talking about grassland, which does not change across years.

---

In response to the line on the main README of the specification:

The Field Boundaries for Agriculture (fiboa) project is focused on making field boundary data openly available in a unified format on a global scale.

Maybe my data exchange use-case with fields that are prepared for satellite monitoring (split by hedges, inner buffer) does not fit with the open data goal of fiboa. In this case I would argue that a field can only be a single polygon.

[m-mohr]

There could be potentially be interesting take aways in https://adaptstandard.org/docs and https://github.com/ADAPT/Standard/issues/97

Also interesting: https://aggateway.org/Portals/1010/WebSite/About%20Us/FIELD%20BOUNDARY%20FLYER%20122123.pdf?ver=2024-01-03-212959-590 :

Field A named and farmer-accepted physical space where production agriculture takes place used to partition and identify data.

Field Boundary A geometry that identifies the geo-spatial coordinates of a field. The boundary can be used to define the area for a particular operation, a particular crop or crops, or for legal purposes. A field can have different boundaries that may vary in geometry based on their specific use but are always either a polygon or multi-polygon

[m-mohr]

We didn't really came to a conclusion today, e.g. what about forests, pastures, orchards, etc. Could we potentially define at least a very high-level categorization that people can filter on if we don't restrict/define field boundaries?

Note from the meeting:

why was a field constructed? => ?
how was a field constructed? => determination_method
=> two different things
=> leads to accuracy for field boundaries (what can you use it for)