With eBird now collecting up to 3 million observations in a single month, future versions of the ERD will continue to grow. The eBird Reference Dataset (ERD) is the driving force behind much of the science conducted with the eBird dataset, including the visually striking occurrence maps and the resultant State of the Birds analysis, which utilized ERD 2.0 data to assess the role of public lands in the protection of birds within the United States and plays a major role in policy recommendations. We look forward to more science and conservation successes in the years to come.

Compared to ERD 2.0, this year's data set is a much richer data resource:

The Dataset

Download the data

Download the data manual

Download the species primer (50MB file - please be patient)

The eBird Reference Dataset (Version 3.0) is free to download and available for use within the usage and copyright restrictions below. The data manual PDF above describes the eBird Reference Dataset and the processing steps taken during creation. We hope these data will be a useful resource for studying avian dynamics and for developing new ecological modeling techniques. This dataset contains count data for bird species observed by novice and experienced bird observers (a.k.a. birders). The data were submitted to eBird, a citizen-science project run by the Cornell Lab of Ornithology and the National Audubon Society. Each record in this dataset is a part of a checklist that a birder has collected, marking the number of birds of each species detected; one checklist is submitted per sampling event (i.e., birding session). Each checklist submitted from the 48 states in the contiguous United States is additionally annotated with hundreds of predictor variables (covariates) that are derived from the location of the sampling event. Additional checklists from Canada, Alaska, the Caribbean, Middle America, and South America are also included but do not have these additional predictor variables.

As with ERD 2.0, this version includes "Incidental"" Observations (not included in ERD 1.0).  Incidental Observations are those submitted to eBird without effort information (i.e., duration, distance, or area covered). New this year is the addition of Fragstat statistics for National Land Cover Dataset (NLCD) source data for the year 2001 and newly released source data year for the year 2006.                                          

Usage and Copyright

The eBird reference data is freely available for all usages. The observational data included in the data set have data access level 5 in the Avian Knowledge Network (AKN) data warehouse and are published here in compliance with the AKN data sharing policy. eBird is run by the National Audobon Society and the Cornell Lab of Ornithology, and the data is copyrighted by both organizations. A primary goal in publishing these data is to provide a common data resource for studying and comparing ecological models; as such, derivative versions of the eBird reference data set must not be distributed without explicit permission from the copyright holders. The data set is a snapshot of submitted observations for years prior to 2010 that were submitted to eBird and reviewed by 8 March, 2011. Observations with an observation date later than December 31, 2010 were not included.

Special thanks to Tim Levatich for database support on this project.

Published results using this data should cite this document as follows:

M. Arthur Munson, Kevin Webb, Daniel Sheldon, Daniel Fink, Wesley M. Hochachka, Marshall
Iliff, Mirek Riedewald, Daria Sorokina, Brian Sullivan, Christopher Wood, and Steve Kelling.
The eBird Reference Dataset, Version 3.0. Cornell Lab of Ornithology and National Audubon
Society, Ithaca, NY, December 2011.

Please direct any questions to:
M. Arthur Munson at mmunson@cs.cornell.edu and Steve Kelling at stk2@cornell.edu

In addition, we would be very pleased to be informed of any publications that use this data. Send us an email at ebird@cornell.edu.