Area Monitoring — Going Large and Towards Finer Detail — Partnership with Planet and Batch Stat API
Without good data, even fancy models can and will produce noise. Without good technology, even perfect data can prove useless. Best things happen with cooperation.
The progress of our Area Monitoring (AM) activities has been tremendous in the past twelve months. We have run a full scope AM production in Slovenia, informing farmers about preliminary results in August, then making an update every month, as there could be more and more relevant agriculture activities identified, confirming farmers’ compliance. We and the Paying Agency were impressed with the engagement of the farmers and the positive feedback we have received. In parallel to Slovenia, we have run several small- and large-scale pilots, helping various stakeholders understand the benefits of the monitoring system, both for improvement of quality and reduction of costs. There was even a surprise waiting for us — when starting this activity, we were targeting governmental bodies, but then suddenly quite a few commercial actors started to contact us, being interested in some of the covered services. That said, it should not really be a surprise — the benefits of monitoring are just too large to be limited to subsidy compliance checks. And this will only increase in time, with more and more entities requiring a systematic insight in the state of our planet. How else can you manage if you can’t measure?
Some of you might find the previous sentence familiar. It is the motto of the Planet , which brings us to the first important announcement. We’ve been a reseller of Planet data for quite a while and were using their PlanetScope and Planet Fusion data as one of the inputs to our AM services, not just to address the problem of small parcels, but also to avoid follow-up checks for more than 10,000 parcels, for which we were not able to identify relevant activity using Sentinel data, due to clouds, resolution or other reasons. Planet’s data stream proved to be essential, reliable and impossible to replace.
While working with Planet, we realized we can tap into another asset of theirs — the partner network. We have concluded that the opportunity is just too large to be addressed solely by us. Our AM solution has always been designed to allow re-use by 3rd parties — either at the level of signals, markers, visualization services, traffic light system or Expert app. And now, as we have all parts of the system operational, it is time to kick in the higher tier. We are therefore happy to announce that we have entered into partnership with Planet, which allows us to provide the most efficient access possible to Planet data (both PlanetScope and PlanetFusion). On the other side, it provides Planet’s partners better conditions to access our Area Monitoring services, making it easier for them to build on top of them.
The second novelty is of a technical nature. While working on machine learning models for our services, we realized the importance of the signal processing part. And for parcels, statistical time-series prove to be super-efficient as they contain the majority of the information packed in a couple of orders of smaller volume of data, compared to pixels. We’ve had the Statistical API in operation since April this year and it proved to be an extremely valuable service for this purpose. Over the last six months users made more than 100 million requests, processing an equivalent of 2.5 billion sq. km of Sentinel data, large part of this related to Area Monitoring. It is, however, tedious to run requests for each parcel, typically several million of them. One has to take care for optimal load, error checking, retrying, etc. We have realized we can simplify this process further and this is how Batch Statistical API was born. One now simply has to prepare the geopackage with polygons, configure the processing script that fits their requirements (either simple band values, some vegetation indices or, as in the case of our data science team, a one-megabyte script containing raw values, cloud removal and gap filling, even calculation of some markers) and run the process. Batch Statistical API will take care that out of millions of polygons as many as needed will be processed and the outputs delivered to the object storage, in JSON format, so that they can be read easily with post-processing tools. The service is now in beta, ready to be used.
The European Common Agriculture policy is heading towards the major update in 2023, which is why there will probably be a number of member states prototyping the system over the next year, to be ready when the deadline comes. This is, however, larger than EU CAP. Sustainable agriculture has to become a norm worldwide, if we want to reduce carbon emissions. And agriculture is just the first and most obvious step. Soon after, we hope that other governmental bodies and large corporations that are on the path to carbon-neutrality will follow. Just about anyone will benefit from signals, markers and indicators based on Earth Observation data. We are confident that our partnership with Planet, combined with the proven technology, will play an important part.
Want to know more?
Check a live webinar we’ve done with Planet to learn how the Slovenian National Paying Agency has benefited from using Planet Fusion.
Watch it here.