This repository organizes the pasture mapping codes developed by Laboratório de Processamento de Imagens e Geoprocessamento (LAPIG/UFG). The methology used by LAPIG team is avaliable in the paper of PARENTE et al. (2017)

Requisites:

• Python 3.7 or above

• Gdal python package and Gdal Binaries

• Scipy python package

• Earth Engine python library

• An folder synchronization with Google Drive (For Windows | For Unix)

Recommendations for Windows:

• Install Miniconda - Python 3.7 or above and the Gdal package. For Windows users, we need to add some system variables like:

  • PATH = C:\ProgramData\Miniconda3; C:\ProgramData\Miniconda3\Library\bin; C:\ProgramData\Miniconda3\Scripts;
  • GDAL_DATA = C:\ProgramData\Miniconda3\Library\share\gdal

• Install and configure the Earth Engine python library. Click here to see how to install and configure with Miniconda.

Recommendations for Unix:

• Install Python-Gdal and Gdal Binaries (sudo apt-get install -y python-gdal; sudo apt-get install -y gdal)

• Install Earth Engine python library. Click here to see how to install and configure with Python PIP.

You have 2 options for make your classification:

Before executing the first Python code, change the parameters below as desired:

 # Build an file with Landsat WRS scenes you wanna use or you can uncomment the tiles used
 # to classify cultivated pasture (default) or natural pasture 
 * line 13 - tiles = open('tiles.in','r').read() #Cultivated Pasture tiles
 * line 14 - #tiles = open('tiles_natural.in','r').read() #Native Pasture tiles
 
 #Choose the classification mode between 0 (natural), cultivated (1) or any (None). P.S.: "Any" considers all the pasture samples as one. 
 * line 22 - pasture_type = 1
 
 #Choose the folder in Google Drive where the data will be saved
 * line 24 - folder = 'mapbiomas_col5_pasture'

Open the console terminal (Prompt on Windows) and run the these line: python3 1_Pasture_classification_mapbiomas_COL_5.py

• Access this link and, if desired, change the parameters of year, landsatWRSPath, landsatWRSRow, my_folder. After that you can click in Run and export your result in Task.

Also, you can change the training dataset (cultivated and natural) by changing the variable TRAIN_DATA (line 9).

Merge the classifications files by year using the binaries gdalbuildvrt and *gdal_translate. E.g.:

• gdalbuildvrt lapig_pasture_map_.vrt .tif
• gdal_translate lapig_pasture_map_.vrt lapig_pasture_map_.tif -co COMPRESS=LZW -co BIGTIFF=YES

In addition, if you want to view a file in a GIS like QGIS, just add a pyramid to your data using:

• gdaladdo -ro lapig_pasture_map_.tif 2 4 8 --config COMPRESS_OVERVIEW LZW --config USE_RRD YES

Before start this processs, be sure of the amount of memory available on your machine and define these variables on the script 2_Multidimensional_median_filter:

 * line 149 - inputDir = '/data/PASTAGEM/mapbiomas_col5_pasture/INPUT'
 * line 150 - outputDir = '/data/PASTAGEM/mapbiomas_col5_pasture/OUTPUT'
 * line 141 - files =  ['Y1985_planted.tif','Y1986_planted.tif','Y1987_planted.tif','Y1988_planted.tif','Y1989_planted.tif','Y1990_planted.tif','Y1991_planted.tif','Y1992_planted.tif','Y1993_planted.t if','Y1994_planted.tif','Y1995_planted.tif','Y1996_planted.tif','Y1997_planted.tif','Y1998_planted.tif','Y1999_planted.tif','Y2000_planted.tif','Y2001_planted.tif','Y2002_plante d.tif','Y2003_planted.tif','Y2004_planted.tif','Y2005_planted.tif','Y2006_planted.tif','Y2007_planted.tif','Y2008_planted.tif','Y2009_planted.tif','Y2010_planted.tif','Y2011_pla nted.tif','Y2012_planted.tif','Y2013_planted.tif','Y2014_planted.tif','Y2015_planted.tif','Y2016_planted.tif','Y2017_planted.tif','Y2018_planted.tif','Y2019_planted.tif']

This code need 2 arguments to run, the start row and the window size (e.g. python3 2_Multidimensional_median_filter 0 2048).

Because of the high memory consumption, this code works in blocks, so you probably will create an bash file (or batch - Windows) to make the process more autonomous. See the code examples below showing how to apply the filter by blocks.

for X in {0..145446..2048} do python3 2_Multidimensional_median_filter.py $X 2048

FOR /l %%X in (0,2048,145446) DO python3 2_Multidimensional_median_filter.py %%X 2048

P.S.: It's possible to obtain the image size using the binary gdalinfo.

Like in the section 2, we will use the gdalbuildvrt and gdal_translate to merge the result files by year.