แบบจำลองภูมิประเทศเชิงเลขจากระบบอากาศยานไร้คนขับเพื่อการคาดการณ์พื้นที่เสี่ยงต่อการเกิดภัย ดินโคลนถล่ม กรณีศึกษา ตำบลสกาด อำเภอปัว จังหวัดน่าน

Abstract

The study of digital terrain model (DTM) from unmanned aerial vehicle system to predict landslide risk areas, a case study in Skad Subdistrict, Pua District, Nan Province has the objectives of the study are to 1) investigate and evaluate the process of aerial photography using an unmanned aerial vehicle system to produce DTM and 2) develop a model to predict landslide risk areas using DTM data for Skad Sub-district, Pua District, Nan Province. Landslides affect a large number of residents and people who want to cross the area. Help local authorities work with simple methods that are not complicated to work in the local authority area and with tools available in the market at a reasonable price. Investigate and evaluate aerial photography with unmanned aerial vehicle systems to find appropriate techniques and methods for digital terrain modeling of the study area by using small multirotor drones for aerial photography in a dense block group format. The results obtained from grid aerial photography are suitable for developing a DTM. This is because they have higher positional accuracy than conventional aerial photography. The DTM obtained from aerial photographs with camera angles of 90, 75, and 55 degrees at an altitude of 250 – 300 meters above the ground, was compared with the DTM from the ALOS satellite data system using Pearson correlation and analyzed with a confidence level of 95%. The comparison of the DTMs obtained from aerial photography with a camera angle of 90 degrees and the ALOS DTM showed that the highest correlation was 0.996. Then, the correlation was checked with a high-level model of the 1: 4,000 scale DTM from the Land Development Department. The Pearson correlation of the DTM with a camera angle of 90 degrees was found to be 0.997 and the least vertical error was 16.427 m. Therefore, this DTM is used for further model development. The highest factor related to landslides in Sakat village, Sakat sub-district, Pua district, Nan province was the distance from the road, land use, NDVI, elevation, and geology. In using these factors to develop a predictive model using statistical methods including 1) logistic regression model and 2) frequency ratio model. The results of the efficiency comparison of the model were obtained using the area under the curve method. It was found that the landslide risk prediction model developed from the frequency ratio model was effective and best suited for the study area with the equation “LSI = 0.50769 ELEV + 0.33622 ROAD + 0.57149 GEOL + 0.59182 LU +. NDVI”. This model has a prediction accuracy of 82.26% and can predict the risk of landslides in the study area at 72.88%. Based on the results of this model, the landslide risk areas can be divided into 5 levels: the lowest risk area is 289.84 rai (15.19 %), the low-risk area is 957.91 rai (50.20 %), the medium risk area is 343.95 rai (18.03 %), the high-risk area is 194.24 rai (10.18 %) and the highest risk area is 122.07 rai (6.4 %).