Change Detection#

{admonition} Tutorial does not work on EO4PH Cluster :class: error

This tutorial is specific to the GEOAnalytics cluster deployed to the Google Cloud environment. This tutorial does not work on the EO4PH cluster which is deployed to Azure. 

Change detection is the process of measuring how the attributes in a particular region have changed over time. Change detection often involves comparing aerial photographs or satellite imagery of the area taken at different times.

In this notebook we will attempt to identify Deforestation within Canadian forests in Central Ontario, using a naive approach by computing the Normalized Difference Vegetation Index (NDVI). Deforestation is an important issue which results in reduced biodiversity, degrading of soil and water quality, negative wildlife habitat impacts and climate change. Canada covers 9% of all the world’s forests with forests occupying only 38% of Canadian land. Since 1990, thankfully, Canada’s annual deforestation rate has decreased swiftly.

We will be using NDVI (Normalized Difference Vegetation Index) to detect clearcutting in the forests. NDVI values range from -1 to 1, where the higher the value is then the higher amount of vegetation in the area. If there is a drastic decrease over the time period in locations with high vegetation, we can infer it as a deforested area.

There are other ways to identify clearcutting, such as using machine learning algorithms (supervised classification). However, using NDVI takes less computation time while also providing stable results.

The main steps in this tutorial are as follows:

1. Retrieve the satellites images to be used.

2. Compute NDVI using a Dask cluster.

3. Plot the NDVI over time.

4. Create a mask showing the deforested areas within the raster image.

Let’s get started by importing the necessary modules and creating our Dask cluster with 30 workers.

[1]:

# First initialize the modules
import os
import glob
import re
import datetime
import requests

import rasterio