Seasonal Variations and Challenges in Estimating Populations and Identifying Species of Korean Ungulates Using Drone-Derived Thermal Orthomosaic Maps

Thermal drones offer significant time and labor savings in estimating wild ungulate populations. Among available methods, Thermal Infrared (TIR) orthomosaic maps have shown potential for animal counting and population estimation. In this study, we assessed the feasibility of counting Korean ungulates in March and June using TIR and RGB orthomosaic maps, aiming to compare population estimates across season and identify the appropriate Ground Sampling Distance (GSD). The average estimated population was 25.3 {+/-} 4.2 individuals (range: 18-29) in March and 28.2 {+/-} 9.1 individuals (range: 14-40) in June. The average counting error was 4.1 (95% CI: 1.3-6.9) across the orthomosaic maps, with no substantial difference between March and June (P = 0.75). Counting errors increased when RGB GSD exceeded 2.5 cm/pix and TIR GSD exceeded 7.5 cm/pix, suggesting that these thresholds may guide future GSD selection for more accurate population estimates. We measured 149 thermal marks for species identification across three ungulate species. MANOVA revealed strong effects of species (P < 0.001) and posture (P < 0.001) on body measurements, but variations in posture limited the effectiveness of thermal mark measurements for species identification. This issue was especially prominent in lying positions, where species overlap occurred. Further data collection on thermal measurements from TIR orthomosaic maps, particularly from a perpendicular angle and across diverse environments and postures, could improve the accuracy of thermal methods for species identification.