Mapping of Shorea robusta forest using time series MODIS data-implications for sub-national forest carbon monitoring in Nepal | |
| Author | Ghimire, Bhoj Raj |
| Call Number | AIT Diss. no.RS-18-06 |
| Subject(s) | Forest mapping--Nepal--Remote sensing Vegetation mapping--Nepal--Remote sensing Remote-sensing images--Interactive multimedia |
| 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 | Forest type mapping and keeping track of carbon stock in national as well as subnational level is very challenging due to high cost, time and resources involved in mapping and surveying. Despite available technologies and methods, mapping forest types using remote sensing data is a long-standing challenge because of similar spectral reflectance from different tree species and significant time and resources are required for acquiring and processing the remote sensing data. However, the central government (and eventually local governments) is required to maintain account of carbon stocks and their change over time in order to address the growing concern of carbon foot-print, global warming and comply by various international conventions. Nepal aims to increase carbon stock growth by at least 5% by 2025 as compared to 2015 and decrease mean annual deforestation rate to 0.05. Underdeveloped countries like Nepal not only can maintain ecological balance, but also can get financial incentives by reducing carbon emission or increasing sequestration if due attention is provided and forest are well managed over a longer period. The amount of carbon stock stored by different forest type are different depending upon species distribution, carbon volume and density for each species, and their distribution along ecological and physiological regions. So, to keep proper account of carbon pool at national level, it is necessary to map the forest by its species type on the first hand, and analyze their distribution in sub-national level on the other hand for better administration and management intervention. The purpose of this study was to map Sal forest using earth observation data and to determine the optimum number of remote sensing images to map the Sal forest through the analysis of Vegetation Index (VI) signatures and calculate the carbon stock in sub national level. We analyzed the eight days' composite moderate resolution imaging spectroradiometer Qv10DIS) time series normalized differential vegetation index (NDVI), and enhanced vegetation index (EVI) for the whole year of 2015. Jeffries-Matusita (J-M) distance was used for the separability index. Performance ofEVI and NDVI was tested using random forest (RF) and support vector machine (SVM) classifiers. Boruta algorithm and phenological statistical analysis were performed to identify the optimum set of imageries. We also performed data level five-fold cross validation ofthe model and field level accuracy assessment of the classification map. The finding confirmed that EVI with SVM (F-score of Sal 0.88) performed better than NDVI in terms of classification accuracy either with SVM or RF. The optimum 12 images during growing and post monsoon season significantly decreased processing time (to one-fourth) without much deteriorating accuracy. Accordingly, we were able to map the Sal forest whose area is accounted for about 36% of the forest cover in the study area. Later the proposed method was extended to produce forest map and Sal map of the Terai region and then whole country. Change of forest area in general and Sal forest in particular was calculated between 2005 and 2015 for each physiological region and province in Nepal. Corresponding carbon content were also determined. According to the study, Sal forest is the dominant species in Nepali forest. It was observed that, around 5.1 million hectares of Nepali land was forest in 2015 increasing from 4.2 million hectares in 2005. However, Sal forest has decreased during the same period occupying 43.4 % percentage of total forest in 2005 to only 33.8% in 2015. The subnational level forest and carbon statistics were produced during this study which can be an important asset for the very fresh federal governance system being practiced for the first time ever in Nepal. Estimated carbon stock considering forest category was found higher than that estimating with national average. This can also pave way for policy formation and preparation of action plan for sustainable forest management and intervention strategy and obtaining better financial incentives participating in reduction of emission due to deforestation and forest degradation (REDD) plus programs. |
| Year | 2018 |
| 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) | Nagai, Masahiko;Sasaki, Nophea |
| Examination Committee(s) | Tripathi, Nitin Kumar;Apichon Witayangkurn;Koh, Lian Pin |
| Scholarship Donor(s) | Government of Japan-AIT Fellowship |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2018 |