Achieving Brazil s deforestation target will deliver air quality and public health benefits

Edward W. Butt1*, Luke Conibear1, Callum Smith1, Jessica C. A. Baker1, Richard Rigby1, Christoph Knote2, and Dominick V. Spracklen1
1School of Earth and Environment, University of Leeds, Leeds, UK
2Model-based Environmental Exposure Science, Faculty of Medicine, University of Augsburg, Augsburg, Germany
*Correspondence to Edward W. Butt: e.butt@leeds.ac.uk 

This document describes the data used to support the findings of this study.
The data can be download from:
https://doi.org/10.5518/1152

The data comprises 16 columns:

1)	'FID':
Unique id for each data point or grid cell.

2)	'geometry': 
Geopandas polygon for each grid cell.

3)	'centre':
Geopandas centroid for each grid cell.

4)	'area_km':
Area of each grid cell.

5)	'states':
Brazilian state that grid cell falls within.
 
6)	'year':
Year of data measurements.

7)	 'month':
Month of data measurements.

8)	'fire_number':
MODIS fire counts within grid cell.

9)	'defor':
PRODES annual deforestation within grid cell. 

10)	 't_month': 
Monthly mean MODIS temperature within grid cell (Degrees Celsius)

11)	'p_6_month':
CHIRPS monthly mean precipitation within grid cell (mm)

12)	 'p_month':
CHIRPS total precipitation in the preceding 6 months within grid cell (mm)

13)	'pasture_frac':
Mapbiomas pasture fraction within grid cell (%).

14)	 'crop_frac':
Mapbiomas cropland fraction within grid cell (%).

15)	 'savanna_frac':
Mapbiomas savannah/grassland fraction within grid cell (%)

16)	'lai_12_month_mean':
Mean MODIS LAI in the preceding 12 months (area/area)


Features used in machine learning models are those described in Table 1 of the main paper.