Characterizing forest succession in Khao Yai National Park using LiDAR and satellite image | |
| Author | Siriruk Pimmasarn |
| Call Number | AIT Diss. no.RS-20-07 |
| Subject(s) | Plant succession--Thailand--Khao Yai National Park Optical radar--Thailand--Khao Yai National Park Remote-sensing images--Thailand--Khao Yai National Park |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Remote Sensing and Geographic Information Systems |
| Publisher | Asian Institute of Technology |
| Abstract | With regards to the development of the tropical forest succession classification and to understand their recovery process, this study explored how the forest succession in Khao Yai National Park, Thailand recovery through the change of Plant Area Index (PAl) within the first 42 years. The results indicate, PAl accumulation increased non-linearly (pseudo-R? of 0.56) during 42 years since forest re-establishment. Compared to neighboring old growth area of200 years, the pattern of PAl slightly increase from 42 years without the effect from topographic factors as elevation, slope and wetness. For characterization of forest successional stages, the results reveal that the combination of 10 meters resolution with 11x11 moving window size of three LiDAR metrics as meanCHM75th, cvCHM10 and B75 are the suitable combination for forest successional stage classification in this study. The percentage of accuracy along with three successional stages are 94%, 94% and 100% of OGS, SES and SIS respectively. Based on this study, we proposed a simple methodology with minimum number of variables for forest successional stage classification to easily applied in other landscape area. To upscale the forest successional classification to large landscape level, the combination of spectral bands, vegetation indices and texture features from the sentinel-2 image were investigated using the Random Forest algorithm. The results showed that using the combination of spectral band and vegetation indices, yield the lower accuracy than the combination of three types variables which was indicated as 59.21% and 61.44% respectively. When considering the number of variables for classification accuracy improvement, we found that using 14 variables can yield the classification accuracy higher than using 11 variables with the same condition which was only 0.33%. On the basis of these results, it can be suggested that the number of variables along with their effectiveness towards classification accuracy should be considered because high number of variables can cause the overfitting model and high processing time when we employed the random forest algorithm. In summary, this study clearly demonstrates the effect of LiDAR data on forest dynamics and characterizing forest successional stages with the low dimensions related to ecological perspective. The results from this study can greatly help for better understanding the forest recovery through successional processes in tropical moist forest which is important for forest management planning for conservation and forest management. |
| Year | 2020 |
| Type | Dissertation |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Remote Sensing (RS) |
| Chairperson(s) | Tripathi, Nitin Kumar |
| Examination Committee(s) | Sarawut Ninsawat;Sasaki, Nophea |
| Scholarship Donor(s) | Royal Thai Government;Asian Institute of Technology Fellowship |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2020 |