Estimating Area of Deforestation and Degradation using AREA2 and CODED

_images/area2.png

By Eric Bullock (bullocke@bu.edu)

This tutorial will demonstrate how the Accuracy and Area Estimation Toolbox (AREA2) can be used to estimate map accuracy and disturbance area using a dataset created with the Continuous Degradation Detection (CODED)1 methodology on the Google Earth Engine (GEE)2. The equations used in this exercise are presented in Cochran 19773 and described in a remote sensing context in Olofsson 20144.

I refer users to the AREA2 documentation for background information on the rational of using statistical inference for assessing map accuracy and area estimation. Definitions for select statistical terminology is provided for words in italics and a more detailed description of terms can be found in the AREA2 documentation.

Disclaimers:

  1. The data presented is here for the purpose of demonstration. Therefore, it should not be considered reflective of the actual land change dynamics of the region.
  2. All calculations should be validated using the original calculations, and it is the user’s responsibility to ensure the calculations produced using AREA2 are correct.

Objectives and Tutorial Parts:

  1. Part 1: CODED Create a spatial dataset of deforestation, degradation, forest, and non-forest in the Brazilian Amazon using CODED.
  2. Part 2: Sample Design and Interpretation Create a sample under stratified random sampling and assign reference labels using the AREA2 toolbox.
  3. Part 3: Estimation of Activity Data and Accuracy Use AREA2 to estimate areas of activity data and map accuracies.