#randomforest — Public Fediverse posts

Detection Of Old-Growth Forests Using Airborne Laser Scanning [LiDAR] And Other Auxiliary Data Combined With National Forest Inventory Field Plot Data
--
https://doi.org/10.1016/j.ecoinf.2026.104085 <-- shared paper
--
H/T @janne Toivonen | D.Sc., Research Scientist (postdoc) at Natural Resources Institute Finland
“Can remote sensing detect old-growth forests at a national scale? 🌲 🛰️
[The author’s study (link above)] evaluates multi-source remote sensing data (ALS + Sentinel-2 + aerial imagery + auxiliary remote sensing data) to detect old-growth forests across Finland.
💡[Some] HIGHLIGHTS:
1️⃣ Multi-sensor synergy: Combining ALS and Sentinel-2 data significantly improves mapping accuracy compared to using ALS data alone
2️⃣ Structure vs. Continuity: Remote sensing excels at detecting structural maturity, but capturing ecological continuity remains a hurdle
3️⃣ XGBoost - strong performer: Extreme Gradient Boosting consistently delivered top performance across different OGF definitions and model configurations
4️⃣ Regional patterns: Detection accuracy was highest in Northern Finland, whereas severe fragmentation in the south created distinct detection challenges..”
#remotesensing #oldgrowthforest #machinelearning #openscience #GIS #spatial #mapping #earthobservation #AI #oldgrowth #forest #Finland #spatialanalysis #model #modeling #Airborne #LiDAR #laserscanning #Sentinel2 #oldgrowthforest #randomforest #XGBoost #generalisedlinearmixedmodel #vegetation #forest #tree #survey #vegetation