ESA Fire_cci Burned Area data from coarse resolution sensors
-----------------------------------------------

The AVHRR-LTDR Fire_cci Burned Area (BA) product v1.1 (also called FireCCILT11 for short) was obtained based on spectral information from the AVHRR Land Long Term Data Record (LTDR) product version 5 dataset (https://ltdr.modaps.eosdis.nasa.gov/cgi-bin/ltdr/ltdrPage.cgi)

FireCCILT11 is available for years 1982 to 2018 at pixel resolution (0.05 degrees) and grid resolution (0.25 degrees). Year 1994 is not included in the dataset because there was not enough input data from the LTDR Archive to process the dataset for that year.

This dataset is available at two different resolutions: pixel (0.05 degrees) and grid (0.25 degrees).

A previous version of this product (FireCCILT10) was a BETA product developed using the same main input data as v.1.0, but with an earlier version of the burned area algorithm.
This BETA product has been deprecated, and should not be used, but it it maintained in the archive for transparency and consistency reasons.
  
For further information, please refer to the Product User Guides available at https://climate.esa.int/en/projects/fire/key-documents/

------------------------------------------------
VERSION HISTORY

Version 1.0 (deprecated):
Was released on March 2019 as a BETA product. 
The most significant improvements compared to the previous version are:

   - The pixel product is now delivered as a standard product.
   - The improvement of the BA detection algorithm has allowed to obtain a consistent product with coherent trends and more burned area compared to FireCCILT10.
     o FireCCI51 product is used for training in the algorithm instead of MCD64A1 because the former has more global and regional accuracy and detects more BA and small fires.
     o New No-Cloud Mask is generated to select free-cloud pixels.
     o Land Cover is annual instead an only using a single Land Cover in the time series.
     o Yearly profiles to train random forest (RF) are regressive and progressive to contiguous years instead of months within a single year.
     o Some RF parameters have been modified to obtain more variability of data.
     o Two RF models (global and boreal) per month were run to keep regional variabilities.
     o Thresholding methods have been improved per month and region, adapting better to each image and correcting the solar angle.
     o BA proportions have been improved.
   - The land cover information has been updated, and now uses ESA’s CCI Land Cover v2.0.7 product and Copernicus Climate Change Service (C3S) Global Land Cover products v2.1.1
   - One more year of data (2018) have been processed.