Editor's note
A better forest plan nobody can read.
Ulvdal's team got at least 2% more expected net present value by letting the optimiser account for uncertain inventory data. Then they showed it to foresters, who could not interpret the result and wanted charts and plainer numbers first. The method is finished. What is missing is the presentation layer.
That is the shape of a lot of what stalls here. We fund the model and treat the interface as something bolted on at the end. A planner who cannot say why the system picked one schedule over another will not sign off on it.
Two percent on a large holding is real money sitting behind a readability problem, and the first vendor to fix that gets the account.
Axel
Ulvdal lets the planning optimiser doubt its own inventory data
A planning system picks harvest schedules decades ahead from one set of inventory figures, and treats those figures as true. Ulvdal's team at SLU, Skogforsk, Laval and Bergvik Skog Öst fed the optimiser many versions of the same figures instead, and let it pick the schedule that holds up across all of them. That schedule returned at least 2% more expected net present value and still ran fast enough to be practical. The foresters in the workshop could not read the output, and asked for charts and plainer numbers first.
Skogforsk carries a digital product passport from stump to finished house
The pilot with Sveaskog and house builder Älvsbyhus followed one log all the way to a delivered home, QR code on the finished product. Skogforsk's conclusion: the sector already holds the data, it just stops at the mill gate where wood from different sources gets mixed. 11,500 homeowners were asked, and they want it.
Twenty-three researchers put the phone ahead of the headset in forestry AR
The review covers every augmented reality tool used in forestry since 2000, and finds the field sitting at technology readiness level 4 to 6 with only a handful of systems in real operation. What works commercially today runs on a phone or tablet: diameter measurement lands within 2 to 3 cm RMSE, and one ARKit-based method hit R² 0.95 across 414 trees. Headsets for tree marking and harvesting support are still pilots, held back by tracking that drifts under canopy.
Leipzig modellers rebuild the vegetation index and more than double the biomass signal
NDVI reads greenness off two wavelengths picked in the 1970s. A team at the UFZ and the Julius Kühn-Institut simulated 900,000 temperate stands and searched the whole 400 to 2,400 nanometre range for the wavelengths that actually track biomass. Their index explains 75% of the variation where NDVI manages 34%, and near-infrared paired with shortwave infrared did most of the work. Choosing the right wavelengths mattered more than making the formula clever. All of it is simulated, with no field data yet.
Allianz Seguros puts a wildfire detection net over 50 million hectares of Spain
The insurer is buying the detection layer, not just paying out after the fact. Satellites on Fire combines geostationary and polar satellite feeds to flag heat signatures, and claims 35 minutes off NASA's alert times, up to 30% more fires found and 40% fewer misses in the early stages. Policyholders with assets near a detected heat source are warned automatically. Spain lost 354,793 hectares last year, and the sector paid out 770 million euros.