Mapping Burned Areas in Tropical forests using MODIS data

1 This paper presents a new burned area product for the tropical 2 forests in South America and South-east Asia. The product is derived 3 from Moderate Resolution Imaging Spectroradiometer (MODIS) mul4 tispectral surface reflectance data and Active Fire hotspots using a 5 novel rare class detection framework that builds data-adaptive clas6 sification models for different spatial regions and land cover classes. 7 Burned areas are reported for 9 MODIS tiles at a spatial resolution 8 of 500 m in the study period from 2001 to 2014. The total burned 9 area detected in the tropical forests of South America and South-east 10 ∗(mithal,nayak,ankush,kumar)@cs.umn.edu †(rama.nemani,nikunj.c.oza)@nasa.gov 1 Asia during these years is 2,286,385 MODIS pixels (approximately 11 571 K sq. km.), which is more than three times compared to the es12 timates by the state-of-the art MODIS MCD64A1 (742,886 MODIS 13 pixels). We also present validation of this burned area product using 14 (i) manual inspection of Landsat false color composites before and af15 ter burn date, (ii) manual inspection of synchronized changes in vege16 tation index time series around the burn date, and (iii) comprehensive 17 quantitative validation using MODIS-derived differenced Normalized 18 Burn Ratio (dNBR). Our validation results indicate that the events 19 reported in our product are indeed true burn events that are missed 20 by the state-of-art burned area products. 21

[1]  N. Koutsias,et al.  Logistic regression modelling of multitemporal Thematic Mapper data for burned area mapping , 1998 .

[2]  D. Roy.  Multi-temporal active-fire based burn scar detection algorithm , 1999 .

[3]  Hajime Akimoto,et al.  The atmospheric impact of boreal forest fires in far eastern Siberia on the seasonal variation of carbon monoxide: Observations at Rishiri, A northern remote island in Japan , 2000 .

[4]  M. Fulk,et al.  Burned area in Kalimantan, Indonesia mapped with NOAA-AVHRR and Landsat TM imagery , 2001 .

[5]  Alan H. Strahler,et al.  Global land cover mapping from MODIS: algorithms and early results , 2002 .

[6]  M. Schultz,et al.  Atmospheric Chemistry and Physics , 2001 .

[7]  Ruiliang Pu,et al.  Determination of Burnt Scars Using Logistic Regression and Neural Network Techniques from a Single Post-Fire Landsat-7 ETM+ Image , 2004 .

[8]  T. Loboda,et al.  Reconstruction of fire spread within wildland fire events in Northern Eurasia from the MODIS active fire product , 2004 .

[9]  E. Kasischke,et al.  AVHRR-based mapping of fires in Russia: New products for fire management and carbon cycle studies , 2004 .

[10]  J. Randerson,et al.  Global estimation of burned area using MODIS active fire observations , 2005 .

[11]  Ted E. Senator,et al.  Multi-stage classification , 2005, Fifth IEEE International Conference on Data Mining (ICDM'05).

[12]  P. Morgan,et al.  Evaluation of linear spectral unmixing and ΔNBR for predicting post‐fire recovery in a North American ponderosa pine forest , 2007 .

[13]  E. Chuvieco,et al.  Mapping burned areas from Landsat TM/ETM+ data with a two-phase algorithm: Balancing omission and commission errors , 2011 .

[14]  J. Randerson,et al.  Global burned area and biomass burning emissions from small fires , 2012 .

[15]  Vipin Kumar,et al.  RAPT: Rare Class Prediction in Absence of True Labels , 2017, IEEE Transactions on Knowledge and Data Engineering.