Details

Written by Super User

Category: Geographic Information System

Published: 05 September 2018

Even though there are already online species mapping platforms, such as Map Of Life, the range of environmental variables and spatial resolution within these platforms are limited. On the other hand, Google Earth Engine (GEE) provides terabytes of earth observation data and GBIF provides primary biodiversity data for scientist to analyze. Sadly, conservation biologists are usually having difficulties in collating, analyzing, and visualizing these data. Here, the author propose a workflow and a script in GEE for conservation biologist in order to develop habitat or niche models. The workflow consist of 13 steps that can be customized, that include data preparation, masking, creation of derivatives, training, prediction, and calculation of error matrices. There are four machine learning algorithm used (Maxent, Random Forest, SVM, and Naive Bayes), but the script can be extended to include other algorithms available on GEE. User can also edit the script to select which environmental variables they want to use in their analysis.

Instruction:

1. Go to https://code.earthengine.google.com/1edb615e401ccf866ddaa76977a90828

2. Edit the area of interest (AOI) layer.

3. Edit or input the target species layer based on the data from GBIF. Add absence data if necessary.

4. Edit the temporal filter.

5. Edit to select the environmental variables required for your analysis. You can uncomment the display the environmental variables you want to review.

6. Edit to select the machine learning algorithm required.

7. Edit the export snippet according to your needs.

Result of classification