Improvement of Local Maximum Fitting (LMF) for high temporal remote sensing data using meteorological data | |
| Author | Salinthip Kungvalchokechai |
| Call Number | AIT Thesis no.RS-09-03 |
| Subject(s) | Remote sensing Meteorological satellites |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Remote Sensing and Geographic Information Systems |
| Publisher | Asian Institute of Technology |
| Abstract | This study intends to improve Local Maximum Fitting (LMF), for the purpose of removing satellites image noise from cloud or haze. The LMF process, is a technology of combination time series filtering and function-based fitting procedures. The three step of LMF are; interpolation using Min Max Filter, simulate the fluctuation by limited number of Harmonic Waves, and to avoid overfitting by selecting limited number of independent variables based on minimum AIC. This improvement of LMF in order to obtain the reliable NDVI value are in the first step, acquired raw data (in this study, 8-days MODIS image) should be selected meticulously before going on the second step of fitting model. The improved Min Max Filtering method changing window size depends on weather condition, is divided to 2 concepts. Concept 1: window size depends on average monthly rainfall historical statistical data. Concept 2: window size depends on the number of continuous cloud-days for NDVI 8-days data by using cloud mask method to define cloud-days. The advantages are to avoid the case that correct NDVI value may be eliminated by using too large window filter, on the other hand, the noise cannot be eliminated by using too small window filter. Next step of improvement, 'Best Max Value Method', in case of similar pattern and similar condition of same type of crop in each year, select maximum NDVI at same day within analysis period. Before selection, should adjust trend and intersection point of NDVI axis first to be the same base. After the acquisition of the expected correct raw data, the NDVI will be reconstructed in the third step of process, by using fitting the time series model with Fourier formula. The coefficients and fitting functions can be selected by the least square and minimum AIC. |
| Year | 2009 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Remote Sensing (RS) |
| Chairperson(s) | Honda, Kiyoshi; |
| Examination Committee(s) | Kibe, Seishiro ;Vivarad Phonekeo ;Ohira, Wataru; |
| Scholarship Donor(s) | Thailand (HM King); |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 2009 |