Edit on GitHub

Tutorial

This tutorial shows how to:

  1. Download monthly nighttime light data
  2. Load the data cubes of radiance and the number of cloud-free observations.
  3. Clean the data
  4. Generate zonal statistics

1. Downloading the data

  1. Create a folder called radiance and a folder called cfobs. We will use these folders to the radiance and the cloud-free observations images respectively.
  2. Go to the VIIRS Nighttime Lights download page at Earth Observation Group
  3. Click on 2017, the on 201704. This means April 2017.
  4. Click on vcmcfg (we will use this version of the dataset)
  5. Click on SVDNB_npp_20170401-20170430_75N060E_vcmcfg_v10_c201705011300.tgz. Note that the filename has 75N060E in it. This means TILE3, where Mumbai is.
  6. Extract the tar files. Put SVDNB_npp_20170401-20170430_75N060E_vcmcfg_v10_c201705011300.avg_rade9h.tif in the radiance folder. Put SVDNB_npp_20170401-20170430_75N060E_vcmcfg_v10_c201705011300.cf_cvg.tif in the cfobs folder.
  7. Repeat the process for all the months. If you are interested only in a certain set of months. Download only those.

2. Loading the data

Create data cubes for radiance and cloud-free observations.

using NighttimeLights
using Rasters
using Dates
dates = collect(Date(2012,4):Month(1):Date(2022, 06))
timestamps = NighttimeLights.yearmon.(dates)
filelist = readdir("~/Downloads/radiance")
radiances = [Raster(i, lazy = true) for i in filelist]
timestamps = collect(1:length(radiances))
series = RasterSeries(radiances, Ti(timestamps))
radiance_datacube = Rasters.combine(series, Ti)
filelist = readdir("~/Downloads/cfobs")
radiances = [Raster(i, lazy = true) for i in filelist]
timestamps = collect(1:length(radiances))
series = RasterSeries(radiances, Ti(timestamps))
ncfobs_datacube = Rasters.combine(series, Ti)

If you haven't downloaded the data, and you want to run the remaining code, you can use load_example() to load the datacubes of radiance and the number of cloud-free observations around Mumbai

3. Cleaning Data

A single function, clean_complete, can be used on radiance_datacube and ncfobs_datacube to generate a cleaned datacube of radiance. 

cleaned_datacube = clean_complete(radiance_datacube, ncfobs_datacube)