Overview

This repository stores functions written to download MODIS landcover data for coordinates across a range of years. Downloaded raw data can be quickly summarised into a .csv file so that raw ASCII files can be deleted, saving memory (and precious workspace!).

Instructions for use

1. If need be, install MODISTools

Please note that older versions of MODISTools may have problems with broken links! It is therefore better to remove old versions first and install the most up-to-date version, i.e.

remove.packages('MODISTools') 

followed by

install.packages("https://modis.ornl.gov/files/modissoapservice/MODISTools2.tar.gz",repos=NULL,type="source")

or manually downlaod MODISTools2.tar.gz here (https://modis.ornl.gov/data/modis_webservice.html) followed by

install.packages("yourfilepath/MODISTools2.tar.gz",repos=NULL,type="source")

2. Import coordinates data (individual sites as rows with 'lat' and 'long' as colnames)

test.dat <- data.frame(lat = c(-27.4678, -27.5636, -27.5598), 
                       long = c(153.0067, 152.2800, 151.9507))

3. Source functions needed for MODIS downloading and processing from GitHub

source("https://raw.githubusercontent.com/nicholasjclark/LandcoverMODIS/master/R/Landcover_functions.R")

4. Download data for the specified year (or range of years)

!! but don't include 2010-2012, see below !!

This stores the raw downloaded data in a new 'LandCover' folder

download_landcover(years_gather = c(2006:2008), coordinates = test.dat)

4. Once ALL of the necessary files are downloaded, summarise them

!! Do not summarise before all necessary raw files are downloaded (raw files get deleted) !!

This writes a .csv summary file in the 'LandCover' folder and deletes the raw files

summarise_landcover()

Some quirky aspects of downloading that need to be acknowledged

1. Certain years (2010 - 2012) are missing from the Land_Cover_Type_1 band

I'm not sure why this is (maintenance on the satellite??), but they never seem to result in appropriate downloads. This should produce an error message:

download_landcover(years_gather = c(2008:2012), coordinates = test.dat)

For now, these years should just be omitted

2. Landcover downloads struggle if we access more than ~100 points at a time

I've included an error to remind that downloading for multiple subsets is safer. This should produce the error message:

error.dat <- data.frame(lat = rnorm(200, mean = 50),long = rnorm(200, mean = 50))

download_landcover(years_gather = c(2006:2008), coordinates = error.dat)

It is recommended to use subsets for sequential downloads i.e.

download_landcover(years_gather = c(2006:2008), 
                   coordinates = test.dat[1:2,])
download_landcover(years_gather = c(2006:2008), 
                   coordinates = test.dat[3:4,])

etc. to get all of the necessary raw data files. Once ALL of the subsets are downloaded, THEN use the summarise_landcover() function

summarise_landcover()

These directions are also supplied in the Landcover_blueprint_test.R file in the R folder in case you would like to download the script and/or share it around. Please feel free to let me know if you run into issues!