Notes in English to avoid RStudio conflict with non UTF encoding (PT).

I'd suggest us to work with functions from RSToolbox from Benjamin Leutner, particularly unsuperClass() to start.

https://stackoverflow.com/questions/1595848/configuring-git-over-ssh-to-login-once

We will explore Landsat Surface Reflectance Level-2 Science Data Products. Details here: https://landsat.usgs.gov/landsat-surface-reflectance-data-products

Ideally we'll work with Landsat scenes from Tier 1 collection More about Tier collections here: https://landsat.usgs.gov/what-are-landsat-collection-1-tiers

A Technical Guide for USGS products (focus on COllection 1) https://above.nasa.gov/pdfs/Landsat_Surface_Reflectance_ABoVE_21Apr2017.pdf

The best and last scene capturing a large fraction of exposed sediments is:

Understant product levels, particularly when delivered from bulk download order of obtained directly from ESPA notification

• Typical bulk order (T1 product): LC08_L1TP_204052_20170106_20170312_01_T1_B1.tif
• Typical ESPA download (T1 product): LC08_L1TP_204052_20170106_20170312_01_T1_sr_band1.tif

From LEDAPS Product Guide Manual v8.0 (for L 4 to 7 https://goo.gl/92n5Nb) Changes implemented in 2017-03-10: "Edited for new quality assurance (QA) band information in Collection 1 (cfmask, cfmask_conf replaced by pixel_qa.) L1 quality band (bqa) removed from standard output. Per-pixel sensor/solar angle bands (derived from band 4) now provided with each product. toa_qa replaced by radsat_qa."

also described on LASRC Product Guide Manual v4.1 (for Landsat 8) (https://goo.gl/CxFGSy)

• From LaSRC pixel_qa band
• MNDWI (with potential if combined with hierarchical approach - try CTree)

• Classification of Potential Water Bodies Using Landsat 8 OLI and a Combination of Two Boosted Random Forest Classifiers (https://goo.gl/Xxh7gw)

• Target Detection Method for Water Mapping Using Landsat 8 OLI/TIRS Imagery (https://goo.gl/RkD7Ss)