SKYLAB’s Post

Seedling Survival in Loblolly Pine in different age classes! We are delighted to say that our US customer base has expanded considerably in the past year or so, and we have been able to train a series of AI models to successfully detect and count loblolly pine seedlings in all age classes, from 1 year through to full maturity. We also measure the crown diameter of every single seedling.

Tree Height Modelling with Stereo Satellite Imagery! In the past, many of our customers have asked us whether it’s possible to measure tree heights using satellite imagery and, until recently, we have had to disapppoint them by telling them the technology just isn’t quite there yet. However, since mid 2023 we have been modelling tree heights successfully with 30cm-resolution stereo satellite imagery, with an accuracy of approximately +/-1m. There are a few caveats though! - Firstly, the data has to be ‘tasked’ (a satellite has to be sent out specially on command) and there are minimum order sizes and prices for that, so it's only worth it on a larger scale. - Secondly, this works well in sparser forests and/or flatter terrain. In steep terrain and dense forest, an existing LiDAR Digital Terrain Model may be necessary to achieve good results. Feel free to contact us for more info! 

Check out our new website! Simpler products, better #forestAi and more customisation to meet your #precisionforestry needs. www.skylabglobal.com

It was a privilege to be able to present at the The University of Georgia GIS conference last week in the US on the subject of "What CAN and CANNOT be done accurately with remote-sensing in forestry"! Everyone was really welcoming and enthusiastic and it was great to meet some of our US clients in person. The trip was EXTREMELY short, but well worth it!

Oh it's November already? Time for #ForestTECH! Albrecht von Ruffer will be share the latest on how #forestai helps to see and understand #forest change and #climatechange.

SKYLAB #forestai seedling counter can be trained to detect many different species. Our latest example: #mangrove #restoration. This helps to track planting performance as well as monitor survival rates to maximise the success of #reforestation.

Successful counting of 1-yr Loblolly pine in Southern US plantations! We have been running trials with more customers in the US South recently, and have successfully trained a good AI model for detecting this species and age class based on drone data, so that seedlings can be counted even in dense competing vegetation! We have also learned that the best time to do seedling survival surveys with drone technology is in the Fall or Winter of the year following planting. The very abundant growth during the summer season makes the pines larger and more visible, while competing vegetation is also less dense during the Winter. Want to know more? book your free consultation here: https://lnkd.in/esB_ftCN #precisionforestry #satellite #ai #forestAI #precisionforestry #remotesensing #datascience #imagery #forestry #stateoftheart #seedlings #trees #counting #skylabcloud

Cashew counting using hi-res satellite imagery! We used Pléiades satellite data from Airbus, of between 30cm and 50 cm resolution, to successfully count individual cashew trees in South-East Asia. We then created stocking density grid maps, with the exact number of trees per hectare in each grid cell, so that the customer could see the density variation and was better able to predict the future yield of his 5,000ha plantation. We found the count was equally as accurate with the 30cm data as with the 50cm data. Want to know more? book your free consultation here: https://lnkd.in/esB_ftCN #precisionforestry #satellite #ai #forestAI #precisionforestry #remotesensing #datascience #imagery #forestry #stateoftheart #changemonitor #skylabcloud

Determining species distribution with Sentinel satellite data. Spruce is more valuable than pine. Using reference data to determine the spectral signatures of both those species in this Scottish plantation, we were able to locate and map the distribution of each across 2,000ha. We used freely-available Sentinel 10m-resolution satellite data and achieved an accuracy of around 85%. If greater accuracy is required, then this could be achieved with satellite data of a higher resolution. We would also add the caveat that this works on stands with two, or maximum three, species in the mix. A greater species mix would result in overlapping spectral signatures and therefore a lower accuracy. Want to know more? book your free consultation here: https://lnkd.in/esB_ftCN #precisionforestry #satellite #ai #forestAI #precisionforestry #remotesensing #datascience #imagery #forestry #stateoftheart #changemonitor #skylabcloud