Inventering av älg med drönare: Utveckling av metodik för uppskattning av täthet, könskvot och reproduktion

Unmanned aerial systems (UAS) equipped with thermal infrared sensors are increasingly used for
wildlife monitoring, but standardized methods for moose (Alces alces) surveys are still lacking. The
objective of this study was, in a first study, to evaluate drone-based moose surveys as a first step to
develop recommendations for estimating population density, sex ratio, and calf recruitment within
Swedish moose management systems.
Field studies were conducted in three areas in Sweden using both automated grid flights and
manually piloted surveys. Survey performance was evaluated through repeated flights, comparisons
among experienced pilots, and targeted surveys of GPS-collared moose. In addition, simulations
based on complete survey datasets were used to evaluate how survey precision was affected by
sampling intensity and plot size.
Detection probability was estimated to be close to one under favourable winter conditions
characterized by low temperatures and overcast skies. All GPS-collared moose present within
survey plots were detected, and manual and automated survey approaches produced highly similar
results. Drone-based density estimates were consistent with independent estimates from other
monitoring methods in areas where such data were available.
Simulation analyses indicated that plot-based sampling designs can provide reliable density
estimates, but required sampling intensity depends strongly on the spatial aggregation of moose. In
areas with moderate aggregation, surveying approximately 20–30% of the total area may be
sufficient to achieve acceptable precision, whereas up to 50% coverage may be required in highly
aggregated populations. Estimation of sex ratio and calf-to-cow ratio generally requires larger
sample sizes than density estimation.
Sex and age classification was feasible using high-resolution zoom imagery, although reliable
classification required additional flight time, image collection from multiple viewing angles, and
experienced observers.
Major limitations of the method include dependence on favourable weather conditions, limited
battery endurance, and current aviation regulations restricting operations beyond visual line of sight
(BVLOS). Further research should focus on detection probability under varying environmental
conditions, evaluation of AI-assisted detection systems, and development of standardized survey
protocols. Overall, the results demonstrate that drone-based surveys have considerable potential as
a cost-effective, scalable tool that complements existing methods for moose monitoring and
management.