Terrain mapping from unmanned aerial vehicles

• Main Authors: Yazid, A.S.M., Wahid, R. A., Nazrin, K. M., Ahmad, A., Nasruddin, A. S., Dollah, R., Hamzah, M. A., Razak, M. Y. A.
• Format: Article
• Published: 2019
• Subjects: HD1394-1394.5 Real estate management
• Online Access: http://eprints.utm.my/id/eprint/91141/; https://jamt.utem.edu.my/jamt/article/view/5228
• Institution: Universiti Teknologi Malaysia

In land surveying, digital terrain model (DTM) and digital surface model (DSM) have long been benefitted in many applications related to terrain mapping. Conventional methods of generating DTM and DSM have limitations in terms of practicality, time consumption and costing. The problems are much more serious for tropical regions where clouds are persistence and tend to affect the accuracy of most of these devices. This study aims to propose a novel way of generating DTM and DSM by utilising unmanned aerial vehicle (UAV) for different land covers including forest, plantation and developed areas in the tropical region of Malaysia. The aerial images obtained from non-matrix digital compact camera payload on UAV were processed photogrammetrically to produce terrain mapping products including DTM, DSM and orthophoto. A detailed survey is also conducted at these areas to produce contour map as benchmark data in which is less being practiced by UAV mappers. To determine the accuracy, quantitative and qualitative analysis were carried out by means of root mean square error (RMSE) and visual inspection. The results show that the RMSE of DTM for forest, plantation and developed area are ± 1.806 m, ± 0.938 m and ± 0.549 m, respectively while for DSM are ± 3.143 m, ± 0.637 m and ± 0.276 m respectively. This study has determined that, the development area gives the highest accuracy compared to the plantation and forested area in which for developed and plantation area the DSM is better than DTM while vice-versa for forested area. It can be concluded that the complexity if terrain is found to be one of the key factors that influences the accuracy of the generated DSM and DTM.