Results and discussionThe implement

Results and discussion
The implementation of BFAST to MODIS VIs time series resulted in the detection of the time
and direction of the sudden and gradual changes in the study area. The fire caused abrupt
decrease in the values of all VIs that were analyzed. Therefore, the breakpoints identified in the
trend component presented the time and range of these changes caused by the fire event in
August of 2007, leading to the detection and mapping of all fire affected pixels (Figure 1). In
Figure 2, the decomposition of the GEMI time series into trend, seasonal and remainder
components is indicative of the BFAST fitting operations. The produced fire perimeters per VI
were spatially compared with a validated perimeter derived by the resampled DMC image. The
percentage of the common area for every pair compared was found to be quite high (over 90%)
and only minor differences existed among the estimated perimeters. More specifically,
comparison revealed a common area of 95.2%, 94.7%, 93.8% and 91.5% for GEMI, BAI, SAVI and
NDVI perimeters, respectively. However, it should be mentioned that unburned patches within
the estimated perimeters were not detected, unlike the validated perimeter, probably due to
the coarse resolution of the MODIS pixel.

Hasil (Bahasa Indonesia) 1:

Hasil dan diskusiPelaksanaan BFAST MODIS VIs waktu seri mengakibatkan deteksi waktudan arah perubahan mendadak dan bertahap dalam areal studi. Api disebabkan tiba-tibapenurunan nilai-nilai semua Vis yang dianalisis. Oleh karena itu, breakpoints yang diidentifikasi dalamtren komponen disajikan waktu dan berbagai perubahan ini disebabkan oleh acara api diAgustus 2007, menyebabkan Deteksi dan pemetaan semua api terkena piksel (gambar 1). DalamGambar 2, dekomposisi GEMI time series menjadi tren, musiman dan sisanyakomponen indikasi BFAST pas operasi. Strategis diproduksi api per VIspasial dibandingkan dengan perimeter divalidasi berasal oleh gambar DMC resampled. ThePersentase daerah umum untuk setiap pasangan dibandingkan ditemukan untuk menjadi cukup tinggi (lebih dari 90%)dan hanya terdapat sedikit perbedaan antara strategis perkiraan. Lebih khusus lagi,perbandingan mengungkapkan area umum 95.2%, 94.7%, 93.8% dan 91,5% bagi SAVI GEMI, BAI, danStrategis NDVI, masing-masing. Namun, perlu disebutkan bahwa pembakaran perbaikan dalamstrategis diperkirakan tidak terdeteksi, tidak seperti perimeter divalidasi, mungkin karenaResolusi kasar MODIS pixel.