Metsähallitus will start using assortment level timber yield predictions produced by artificial intelligence. This is the first type of data produced by artificial intelligence that will be utilised in operational forest management planning. 

This autumn, the field experts of Metsähallitus Forestry Ltd will have access to timber assortment yield predictions produced by artificial intelligence (AI). The yield predictions are the first type of data produced by AI that aims to improve operational forest planning and empower the work of the field experts.

– In the commercial forest managed by Metsähallitus, the field experts must consider various objectives at the same time, and the amount of information used in planning has increased. The objective of the new AI-based approach is to reduce manual field measurements to free up more of the field expert’s time for other important tasks such as considering aspects related to biodiversity and stakeholder engagement, says Hannu Lehtonen, Metsähallitus’ Director of Forest Management.

The yield predictions are used for forecasting the volume of different timber assortments gathered from harvests. More accurate timber assortment level data has a significant impact on all processes and operations of Metsähallitus.

– The more accurate yield predictions will streamline the entire supply chain from the planning and harvesting to the delivery of timber to customers, Lehtonen emphasises.

The yield predictions produced by AI have been tested across Finland since last winter. This autumn, Metsähallitus will ramp up the utilisation of the data. The predictions produced by AI will be used by some 100 field experts in almost 3 million hectares of forest land.

– We are expanding the use of the data to gain more user experience and to find out how the AI predictions work under different conditions, says project manager Kari Louhisalmi from Metsähallitus.

Metsähallitus operates within a wide geographical scope and applies many forest management models. Initially, the field experts will use the yield predictions when planning traditional final felling and thinning operations. In the future, predictions will be developed for other forest management models as well.

For the past couple of years, the utilisation of AI in forest management has been studied at Metsähallitus in partnership with CollectiveCrunch Oy. The partner was chosen through an innovation partner model that will enable a new and improved approach to the development of modern methods and solutions.

– We are very pleased that Metsähallitus is now ramping up the utilisation of yield predictions produced by our AI-powered service platform Linda. The digitalisation of the forestry sector is advancing at a tremendous pace, and Metsähallitus is the global forerunner in introducing modern approaches. We couldn’t hope for a better partner for this development, says Jarkko Lipponen, CEO of CollectiveCrunch Oy.

AI can combine information collected from various sources and process huge amounts of data. The data used to train the machine learning models which produce the predictions include remote sensing data from various sources and data from harvesting machines.
– The predictions will continue to improve as we continuously receive more data from the harvesting machines operating on our lands across the country. Feedback from the field experts is also important, says Louhisalmi.

In addition to the timber assortment yield predictions, the collaboration between Metsähallitus and CollectiveCrunch has included developing an AI based approach to identification of silvicultural needs. This development will continue during the coming winter.

*This article was originally published in Finnish on Metsahallitus’ website on 10^th November 2021. (Link)