Data for: A comprehensive characterization of MODIS daily burned area mapping accuracy across fire sizes in tropical savannas

Published: 23 April 2020| Version 1 | DOI: 10.17632/kc89crcwgy.1
Manuel Campagnolo


Burned area maps for the Brazilian Cerrado and fire season 2015 computed by the PT algorithm (Campagnolo et al 2019) There are three products based on MODIS surface reflectance at 250 m (PT-GQ-NIR) , at 500m (PT-GA-BSVI) and combined 250-500 m (PT-NIR-BSVI) Each file covers the rectangle of longitudes between 68°W and 40°W and latitudes between 26°S and 2°N which contains the Brazilian Cerrado, from Abril 30, 2015 to December 31, 2015. Each row represents one BA observation and columns are the following. (1) lon (decimal degrees), (2) lat (decimal degrees), (3) doy (estimated julian date of burn), (4) doy_pre (earliest julian date of burn), (5) doy_post (latest julian date of burn), (6) aream2 (observation burned area in m2), (7) pixelsize (pixel size of original surface reflectance observation, in meters), (8) mscores (MaxEnt score), (9) thr (threshold for mscore), (10) croplands (1 if cropland pixel, 0 otherwise), (11) dg (degree in graph of events), (12) medianscores (median score of the patch where the observation lies), (13) nAFpatch (number of active fires associated with the patch where the observation lies), and (14) IDXpatch (index of patch where the observation lies). Missing values are indicated by -1. Reference: Campagnolo, M. L., Oom, D., Padilla, M., & Pereira, J. M. C. (2019). A patch-based algorithm for global and daily burned area mapping. Remote Sensing of Environment, 232, 111288.



Universidade de Lisboa Centro de Estudos Florestais, Universidade de Lisboa


Remote Sensing, Savannah Ecosystem, Change Detection, Fire