Identification of a burning area using hotspot data and aerosol properties | |
| Author | Chanakan Wuthisakkaroon |
| Call Number | AIT Thesis no.RS-17-16 |
| Subject(s) | Aerosols Area--Identification |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Remote Sensing and Geographic Information System |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis;no. RS-17-16 |
| Abstract | The main purpose of this study focus ed on burning verification of daily MODIS hotspot and categorize d type of burned hotspot into forest fir e and crop residue burning using C5.0 decision tree algorithm. The classification model was developed using MODIS hotspot and aerosol properties parameter. Firstly, the spatial and temporal pattern analysis was applied to 2006 - 2015 hotspot dataset. The outcome of the first step presents that the highly frequency of 10 years hotspot found in summer season, start in February to April. About 64.84% of hotspots occurred in March. The output of spatial analysis shows that t he most density area was found in Chareom Phrakiat district with approximately three hotspots per square kilometer. The sample dataset which correlated to spatial and temporal characteristic was selected for building burned/unburned classification model. T he training dataset of model development includes parameter from hotspot data such as brightness temperature and fire radiative power and parameters from MODIS aerosol product. Aerosol optical depth at 0.55 micron (AOD 550 ) was analyzed the change of aerosol in ± 7 days period from event day. The AOD 550 and differential of AOD 550 in ± 7 days period was utilized to develop the burned/unburned classification approach. The performance of burned/unburned model was evaluated using confusion matrix and accuracy of model was 82.7% with 0.62 kappa. To achieve the goal of this study, the intensity and duration of burned hotspot was analyzed to extract particularity pattern between forest fire class and agriculture fire class. The biomass burning classifier was advanced using 10 years burned dataset and 59.92% was the accuracy of classification model. In order to verify the model, 2016 MODIS dataset in Nan province was classified using burning classifier and biomass burning classifier. The output shows the accuracy of burning classifier with 72.65% and accuracy of biomass burning hotspot with 56.25% |
| Year | 2017 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis;no. RS-17-16 |
| Type | Thesis |
| School | School of Engineering and Technology |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Remote Sensing (RS) |
| Chairperson(s) | Sarawut Ninsawat |
| Examination Committee(s) | Shrestha, Rajendra Prasad;Ochi, Shiro |
| Scholarship Donor(s) | Royal Thai Government Fellowship |
| Degree | Thesis (M.Sc.) - Asian Institute of Technology, 2017 |