An end-to-end pipeline for earth observation-data-model synthesis for Antarctic decision support
Abstract
Assessing biodiversity changes in remote regions of the planet requires scalable and efficient approaches to ecological monitoring. Using a novel approach to estimate Adélie penguin (Pygoscelis adeliae) abundance from the Landsat 4, 5, 7, and 8 imagery archive, we refit a recently published Bayesian population dynamics model to all known Adélie penguin abundance estimates over the entire continent of Antarctica from 1982-2015. By explicitly incorporating both process and observation error, our time series model for population change at each breeding colony naturally accommodates data from in situ ground surveys, UAV-based surveys, and estimates from sub-meter commercial satellite imagery. Our newly developed algorithm to statistically-downscale 30 m resolution Landsat imagery substantially increases the number of population estimates available, particularly in the decades prior to regular commercial imagery tasking in the region. By comparing models with and without Landsat-derived abundance estimates, we illustrate how incorporating this new survey modality reduces our uncertainty of population trends at scales relevant for management by the Convention for the Conservation of Antarctic Marine Living Resources. Through automated algorithms for earth observation data and a model that dynamically incorporates new data from both in situ and satellite based data streams, we have created a penguin monitoring system that meets the needs of Antarctic stakeholders and can be easily adapted for use within the GEO BON community.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.B41L2889L
- Keywords:
-
- 0410 Biodiversity;
- BIOGEOSCIENCESDE: 0434 Data sets;
- BIOGEOSCIENCESDE: 0439 Ecosystems;
- structure and dynamics;
- BIOGEOSCIENCESDE: 0480 Remote sensing;
- BIOGEOSCIENCES