Improved ground and aerial system technologies enable mapping and monitoring forests and land to mitigate forest fires. UAV plays a role in monitoring by collecting forest area images from the air, which could be processed into 2D and 3D images. They can be analyzed to identify land cover types and objects in forest areas. This image data collection uses the DJI Phantom 4 Pro UAV controlled automatically with a flight plan made with Pix4D Capture, which is then processed using Agisoft. The result of the mapping has an average GSD of 2,03 cm/px. The mapping result shows that the 3D image produced can show objects in various land cover types. Weather related parameters were measured using ground sensors both in forest and plain area. We had successfully gathered forest and plain area images in addition to weather related parameters in Tangkuban Perahu Mountain area.

Indonesia has 15 big potential forest fires locations which destroyed twice the area annually. Moreover, there are almost 20 fire points for each location which increase gradually according to the satellite images of LAPAN (Lembaga Penerbangan dan Antariksa Nasional - National Institute of Aeronautics and Space of Indonesia). Therefore, some monitoring and early warning system for this disaster are needed to minimize the loss and damage of forest ecosystem, smoke exposure to human respiratory system, aviation business, and some forest related industries. Unmanned Aerial Vehicle (UAV) technology could be proposed as the alternative to solve this problem. The forest is monitored by an UAV, in other hand it could be used for image data acquisition to construct 2D and 3D maps for further analysis of the early warning system. The images of sampling location had been taken by an UAV - DJI Phantom 4 Pro which automatically controlled by a flight plan using Pix4D Capture and processed by Pix4D Mapper. The sampling location took place in different areas: near a building, few km squares of campus area consists of building-vegetation combination, and three forest locations in two active volcanoes. The maps with Ground Sample Distance (GSD) under 3 cm/pixel result in under 1% error for X, Y and Z of 2D and 3D constructed maps. The maps also show an important finding where the movement object could be detected, which potential to be applied for fire evolution detection of a forest fire early warning system.