Implementation of species distribution models in Google Earth Engine. (11th February 2022)

• Record Type: Journal Article
• Authors: Crego, Ramiro D.; Stabach, Jared A.; Connette, Grant
• Editors: Leroy, Boris
• Is Part Of: Diversity & distributions. Volume 28:Number 5(2022)
• Page Start: 904
• Page End: 916
• Publication Date: 2022-02-11
• Subjects: gradient boosting -- habitat suitability models -- JavaScript -- niche models -- occurrence data -- random forest; Biodiversity -- Periodicals; Biodiversity conservation -- Periodicals; 577
• DOI: 10.1111/ddi.13491
• Languages: English
• ISSNs: 1366-9516

Abstract

Abstract: Aim: Google Earth Engine (GEE) is a free Web‐based spatial analysis platform that requires only a web browser and an Internet connection to programmatically access and analyse data from its multi‐petabyte catalog of regularly updated satellite imagery (e.g. MODIS, Landsat, Sentinel) and other geospatial datasets. The high computing capacity of GEE can make computationally demanding analyses more accessible to researchers and practitioners, especially those with limited access to advanced computational resources. Here, we present a workflow in GEE to fit species distribution models, offering direct access to a multi‐petabyte catalog of raster products to obtain estimates of habitat suitability. Innovation: We implemented a workflow for species distribution modelling in GEE that includes importing species occurrence data into the GEE platform, selecting and preparing predictor variables, and performing model fitting with spatial or temporal split‐block cross‐validation techniques. We present three case studies that demonstrate: (i) a baseline SDM workflow that produces informative model predictions, (ii) a workflow that accounts for temporal variability in predictor variables to study changes in habitat suitability over time and (iii) a complex and computationally demanding analysis incorporating thousands of satellite images for modelling habitat suitability at high spatial resolution. Main Conclusions: Our SDM workflow allows users to benefit from the high speed and performance of GEE without the need for significant computing infrastructure. This workflow may be especially beneficial to researchers in countries where computing power is limited, as SDMs frequently require the download, storage and processing of large raster datasets. We also discuss key limitations of implementing SDMs in GEE, such as user memory limits and the lack of high‐level functions. We include a step‐by‐step guide for the general model workflow and for each of the case studies presented to facilitate its implementation.