Changes in the canopy structure of a fragmented forest drive habitat shifts by regulating understory solar radiation and temperature. Mapping canopy structure therefore improves both predictions of abiotic conditions within fragments and estimates of the effects of fragmentation on the distribution and abundance of ground-dwelling species. Forest fragmentation and its associated changes in canopy structure happen at large scales that are challenging to map using traditional field surveys. Remote sensing from a small un-crewed aerial vehicle (UAV) provides a more efficient, data-rich method to quantify altered canopy structure. Data were collected at the Wog Wog fragmentation experiment in southeastern Australia where native Eucalyptus forest was fragmented in 1987 and fragments were surrounded by an exotic pine tree plantation. We took multispectral images of the experiment from a UAV and built a 3D point cloud of the canopy structure using structure-from-motion photogrammetry. Using the canopy point cloud, we created a physically-based model of solar radiation reaching the understory and forest floor, calibrated and validated by data from ground-based solar radiation measurements. Alongside the radiation sensors, we measured soil temperature, which is important for invertebrate distributions.
Results/Conclusions
We found a high correlation between our solar radiation model and the radiation sensor data. This means that our solar radiation model run on canopy maps generated from UAV data are effective for mapping the solar radiation environment at the ground at Wog Wog, which is a less time- and labor-intensive approach than traditional techniques of setting out radiation sensors or taking hemispherical photographs. Furthermore, this method is scalable and reproducible in time, which allows us to track how both the canopy structure and microhabitat conditions at the forest floor change in time. Our next step will be to link the modeled solar radiation and ground temperature to existing data for ground-dwelling invertebrates to predict past and future changes in the distribution and abundance of species due to fragmentation. These canopy data also establish an important baseline as the experiment will be re-fragmented when the pine trees are harvested for the first time in the near future. Our approach is a first effort toward proving the utility of UAV approaches for modeling understory microhabitat conditions.