Abstract Drones have proven to be versatile tools for scientific studies aimed at assessing the environmental quality of various ecosystems. This study diagnosed the restinga vegetation in the coastal municipality of Pontal do Paraná, in the state of Paraná, Southern Brazil. This vegetation, which extends over 21.72 km of the coastline of the municipality, was grouped into eight sectors to optimize 27 drone flight plans. The images obtained were transformed into orthomosaics and classified into eight classes, following the random forest classifier: herbaceous, shrubby, arboreal, sand, degraded area, irregular construction, water, and other. We observed 289.11 ha of restinga, of which 183.66 ha (63.5%) corresponded to conserved areas and 105.45 ha (36.5%), to a total degraded area with irregular constructions, vegetation suppression, and introduction of exotic species. Sectors 1 and 5 showed the lowest (8.19 ha) and highest (21.00 ha) values of degraded areas, respectively. Sectors 6 and 8 were the most critical: their degraded areas exceeded the conserved areas of restinga by 4.15 and 3.95 ha, respectively. The main causes of the degradation n in this study refer to irregular occupation. Drone images are a valuable tool for environmental management and vegetation analysis of difficult-to-access sites. ecosystems Paraná Brazil 2172 21 72 21.7 2 plans classes classifier herbaceous shrubby arboreal sand construction water other 28911 289 11 289.1 18366 183 66 183.6 63.5% 635 63 (63.5% 10545 105 45 105.4 36.5%, 365 36.5% , 36 (36.5%) constructions suppression species 8.19 819 19 (8.1 21.00 2100 00 (21.0 respectively critical 415 4 15 4.1 395 3 95 3.9 occupation difficulttoaccess difficult access sites 217 7 21. 2891 28 289. 1836 18 183. 63.5 (63.5 1054 10 105. 36.5 (36.5% 8.1 81 (8. 21.0 210 0 (21. 41 4. 39 9 3. 63. (63. 36. (36.5 8. (8 (21 (63 (36. ( (2 (6 (36 (3

Resumo O objetivo deste estudo foi avaliar a qualidade ambiental da restinga da costa atlântica, no sul do Brasil, utilizando imagens de drones nos espectros de vermelho, verde e azul (RGB) obtidas em setembro de 2020 e 2021. A ortoimagem gerada foi classificada em diferentes classes de vegetação como a herbácea, a arbustiva e a arbórea, além de outras feições como áreas degradadas e construídas. A classificação foi realizada utilizando o algoritmo Random Forest, disponível no plugin Dzetsaka para Quantum GIS. Do total de área da restinga avaliada (289,17 ha), em 2020, 61,54% encontrava-se conservada e 36,72% estava degradada. Em 2021 houve diminuição das áreas conservadas para 59,56% e aumento das áreas degradadas para 38,85%. Em relação aos setores avaliados nos dois anos, o setor 6 teve a menor área conservada de restinga (5,40 ha e 8,14 ha em 2020 e 2021, respetivamente), e o setor 2 teve a maior área de preservação, com 44,07 ha em 2020 e 43,55 ha em 2021. Os setores 1, 2, 4 e 5 apresentaram redução nas áreas de restinga conservadas (em média, 6,75%). Aumento nas áreas conservadas foi observado nos setores 6 e 7 (20,4%) e no setor 8 (2%). Esses resultados mostram o potencial do uso de imagens obtidas por drones em análises multitemporais de áreas frágeis como as restingas, auxiliando nas medidas de proteção e conservação desses ecossistemas. atlântica Brasil vermelho RGB (RGB 202 herbácea arbórea construídas Forest GIS 289,17 28917 289 17 (289,1 ha, , ha) 6154 61 54 61,54 encontravase encontrava se 3672 36 72 36,72 degradada 5956 59 56 59,56 3885 38 85 38,85% anos 5,40 540 40 (5,4 814 14 8,1 respetivamente, respetivamente respetivamente) preservação 4407 44 07 44,0 4355 43 55 43,5 1 média 6,75%. 675 6,75% . 75 6,75%) 20,4% 204 20 (20,4% 2%. 2% (2%) restingas ecossistemas 289,1 2891 28 (289, 615 61,5 367 3 36,7 595 59,5 388 38,85 5,4 (5, 81 8, 440 0 44, 435 43, 67 6,75 20,4 (20,4 (2% 289, (289 61, 36, 59, 38,8 5, (5 6,7 20, (20, (2 (28 38, ( 6, (20

Abstract The aim of this study was to evaluate the environmental quality of the Atlantic coast restinga, in south Brazil, using red, green and blue (RGB) drone images obtained in September of 2020 and 2021. The generated orthoimage was classified into different vegetation classes like herbaceous, shrubs and arboreal as well as other classes such as degraded and constructed areas. Classification was performed using the Random Forest algorithm, available from the Dzetsaka plugin for Quantum GIS. Of the total assessed area of restinga (289.17 ha), in 2020, 61.54% were conserved and 36.72% were degraded. In 2021 there was a decrease in conserved areas to 59.56%, and an increase in degraded areas to 38.85%. Regarding the sectors (1–8) evaluated in the two different years, sector 6 had the smallest preserved area of restinga (5.40 and 8.14 ha in 2020 and 2021, respectively), and sector 2 had the largest preservation area, with 44.07 ha in 2020 and 43.55 ha in 2021. Sectors 1,2,4, and 5 showed a reduction in conserved restinga areas (on average, 6.75%). An increase in conserved areas was observed in sectors 6 and 7 (on average, 20.4%) and in sector 8 (2%). These results show the potential of using images obtained by drone, in multi-temporal analyses of fragile areas such as restingas, helping with protection and conservation measures for these ecosystems. Brazil red RGB (RGB 202 herbaceous algorithm GIS 289.17 28917 289 17 (289.1 ha, , ha) 6154 61 54 61.54 3672 36 72 36.72 5956 59 56 59.56% 3885 38 85 38.85% 1–8 18 1 (1–8 years 5.40 540 40 (5.4 814 14 8.1 respectively, respectively respectively) 4407 44 07 44.0 4355 43 55 43.5 124 4 1,2,4 on average 6.75%. 675 6.75% . 75 6.75%) 20.4% 204 20 2%. 2% (2%) multitemporal multi temporal restingas ecosystems 289.1 2891 28 (289. 615 61.5 367 3 36.7 595 59.56 388 38.85 1– (1– 5.4 (5. 81 8. 440 0 44. 435 43. 12 1,2, 67 6.75 20.4 (2% 289. (289 61. 36. 59.5 38.8 (1 5. (5 1,2 6.7 20. (2 (28 59. 38. ( 1, 6.