Bird-SDPS: A Migratory Birds' Spatial Distribution Prediction System

Abstract

Species distribution modeling is an important ecological research task that has received a great deal of interest. There are several single model packages and applications available for species distribution analysis. This paper introduces Bird-SDPS, a Prediction System for Migratory Birds' Spatial Distribution, which is an extensible system for birds' spatial distribution prediction. The Bird-SDPS uses birds' GPS tracking data and remote sensing data as input to build multiple distribution models, which are implemented by different programming languages. And the system provides online access and visualization functions. In order to store large dataset of remote sensing data, we design a hybrid storage structure based on HBase. We extensively evaluate our system using a real-world GPS dataset collected from 90 wild birds over 3 years. We show that the system can conduct birds' distribution prediction based on multiple models, and our hybrid data storage modes can outperform the traditional storage modes of files.