MistNet : Measuring historical bird migration in the US using archived weather radar data and convolutional neural networks
Lin, Tsung Yu ; Winner, Kevin ; Bernstein, Garrett ; Mittal, Abhay ; Dokter, Adriaan M. ; Horton, Kyle G. ; Nilsson, Cecilia LU
; Van Doren, Benjamin M.
; Farnsworth, Andrew
and La Sorte, Frank A.
, et al.
(2019)
In Methods in Ecology and Evolution
10(11).
p.1908-1922
- Abstract
Large networks of weather radars are comprehensive instruments for studying bird migration. For example, the US WSR-88D network covers the entire continental US and has archived data since the 1990s. The data can quantify both broad and fine-scale bird movements to address a range of migration ecology questions. However, the problem of automatically discriminating precipitation from biology has significantly limited the ability to conduct biological analyses with historical radar data. We develop MistNet, a deep convolutional neural network to discriminate precipitation from biology in radar scans. Unlike prior machine learning approaches, MistNet makes fine-scaled predictions and can collect biological information from radar scans that... (More)
Large networks of weather radars are comprehensive instruments for studying bird migration. For example, the US WSR-88D network covers the entire continental US and has archived data since the 1990s. The data can quantify both broad and fine-scale bird movements to address a range of migration ecology questions. However, the problem of automatically discriminating precipitation from biology has significantly limited the ability to conduct biological analyses with historical radar data. We develop MistNet, a deep convolutional neural network to discriminate precipitation from biology in radar scans. Unlike prior machine learning approaches, MistNet makes fine-scaled predictions and can collect biological information from radar scans that also contain precipitation. MistNet is based on neural networks for images, and includes several architecture components tailored to the unique characteristics of radar data. To avoid a massive human labelling effort, we train MistNet using abundant noisy labels obtained from dual polarization radar data. In historical and contemporary WSR-88D data, MistNet identifies at least 95.9% of all biomass with a false discovery rate of 1.3%. Dual polarization training data and our radar-specific architecture components are effective. By retaining biomass that co-occurs with precipitation in a single radar scan, MistNet retains 15% more biomass than traditional whole-scan approaches to screening. MistNet is fully automated and can be applied to datasets of millions of radar scans to produce fine-grained predictions that enable a range of applications, from continent-scale mapping to local analysis of airspace usage. Radar ornithology is advancing rapidly and leading to significant discoveries about continent-scale patterns of bird movements. General-purpose and empirically validated methods to quantify biological signals in radar data are essential to the future development of this field. MistNet can enable large-scale, long-term, and reproducible measurements of whole migration systems.
(Less)
- author
- Lin, Tsung Yu
; Winner, Kevin
; Bernstein, Garrett
; Mittal, Abhay
; Dokter, Adriaan M.
; Horton, Kyle G.
; Nilsson, Cecilia
LU
; Van Doren, Benjamin M.
; Farnsworth, Andrew
and La Sorte, Frank A.
, et al. (More)
- Lin, Tsung Yu
; Winner, Kevin
; Bernstein, Garrett
; Mittal, Abhay
; Dokter, Adriaan M.
; Horton, Kyle G.
; Nilsson, Cecilia
LU
; Van Doren, Benjamin M.
; Farnsworth, Andrew
; La Sorte, Frank A.
; Maji, Subhransu
and Sheldon, Daniel
(Less)
- publishing date
- 2019-11-01
- type
- Contribution to journal
- publication status
- published
- keywords
- aeroecology, bird migration, convolutional neural networks, deep learning, machine learning, movement ecology, ornithology, weather radar
- in
- Methods in Ecology and Evolution
- volume
- 10
- issue
- 11
- pages
- 15 pages
- publisher
- John Wiley & Sons Inc.
- external identifiers
-
- scopus:85071303029
- ISSN
- 2041-210X
- DOI
- 10.1111/2041-210X.13280
- language
- English
- LU publication?
- no
- additional info
- Publisher Copyright: © 2019 The Authors. Methods in Ecology and Evolution © 2019 British Ecological Society
- id
- 9bae51af-7c62-41e2-b429-0b86fb883560
- date added to LUP
- 2023-08-30 11:47:19
- date last changed
- 2023-11-15 16:45:09
@article{9bae51af-7c62-41e2-b429-0b86fb883560,
abstract = {{<p>Large networks of weather radars are comprehensive instruments for studying bird migration. For example, the US WSR-88D network covers the entire continental US and has archived data since the 1990s. The data can quantify both broad and fine-scale bird movements to address a range of migration ecology questions. However, the problem of automatically discriminating precipitation from biology has significantly limited the ability to conduct biological analyses with historical radar data. We develop MistNet, a deep convolutional neural network to discriminate precipitation from biology in radar scans. Unlike prior machine learning approaches, MistNet makes fine-scaled predictions and can collect biological information from radar scans that also contain precipitation. MistNet is based on neural networks for images, and includes several architecture components tailored to the unique characteristics of radar data. To avoid a massive human labelling effort, we train MistNet using abundant noisy labels obtained from dual polarization radar data. In historical and contemporary WSR-88D data, MistNet identifies at least 95.9% of all biomass with a false discovery rate of 1.3%. Dual polarization training data and our radar-specific architecture components are effective. By retaining biomass that co-occurs with precipitation in a single radar scan, MistNet retains 15% more biomass than traditional whole-scan approaches to screening. MistNet is fully automated and can be applied to datasets of millions of radar scans to produce fine-grained predictions that enable a range of applications, from continent-scale mapping to local analysis of airspace usage. Radar ornithology is advancing rapidly and leading to significant discoveries about continent-scale patterns of bird movements. General-purpose and empirically validated methods to quantify biological signals in radar data are essential to the future development of this field. MistNet can enable large-scale, long-term, and reproducible measurements of whole migration systems.</p>}},
author = {{Lin, Tsung Yu and Winner, Kevin and Bernstein, Garrett and Mittal, Abhay and Dokter, Adriaan M. and Horton, Kyle G. and Nilsson, Cecilia and Van Doren, Benjamin M. and Farnsworth, Andrew and La Sorte, Frank A. and Maji, Subhransu and Sheldon, Daniel}},
issn = {{2041-210X}},
keywords = {{aeroecology; bird migration; convolutional neural networks; deep learning; machine learning; movement ecology; ornithology; weather radar}},
language = {{eng}},
month = {{11}},
number = {{11}},
pages = {{1908--1922}},
publisher = {{John Wiley & Sons Inc.}},
series = {{Methods in Ecology and Evolution}},
title = {{MistNet : Measuring historical bird migration in the US using archived weather radar data and convolutional neural networks}},
url = {{http://dx.doi.org/10.1111/2041-210X.13280}},
doi = {{10.1111/2041-210X.13280}},
volume = {{10}},
year = {{2019}},
}