Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zones
<p>ECOCLIMAP-Second Generation (ECO-SG) is the land-cover map used in the HARMONIE-AROME configuration of the shared ALADIN-HIRLAM Numerical Weather Prediction system used for short-range operational weather forecasting for Ireland. The ECO-SG urban classification implicitly includes building...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Copernicus Publications
2022-05-01
|
Series: | Advances in Science and Research |
Online Access: | https://asr.copernicus.org/articles/19/13/2022/asr-19-13-2022.pdf |
_version_ | 1811291069940236288 |
---|---|
author | E. Keany G. Bessardon E. Gleeson |
author_facet | E. Keany G. Bessardon E. Gleeson |
author_sort | E. Keany |
collection | DOAJ |
description | <p>ECOCLIMAP-Second Generation (ECO-SG) is the land-cover map used in the HARMONIE-AROME configuration of the shared ALADIN-HIRLAM Numerical Weather Prediction system used for short-range operational weather forecasting for Ireland. The ECO-SG urban classification implicitly includes building heights. The work presented in this paper involved the production of the first open-access building height map for the island of Ireland which complements the Ulmas-Walsh land cover map, a map which has improved the horizontal extent of urban areas over Ireland. The resulting building height map will potentially enable upgrades to ECO-SG urban information for future implementation in HARMONIE-AROME.</p>
<p>This study not only produced the first open-access building height map of Ireland at 10 m <span class="inline-formula">×</span> 10 m resolution, but assessed various types of regression models trained using pre-existing building height information for Dublin City and selected 64 important spatio-temporal features, engineered from both the Sentinel-1A/B and Sentinel-2A/B satellites. The performance metrics revealed that a Convolutional Neural Network is superior in all aspects except the computational time required to create the map. Despite the superior accuracy of the Convolutional Neural Network, the final building height map created results from the ridge regression model which provided the best blend of realistic output and low computational complexity.</p>
<p>The method relies solely on freely available satellite imagery, is cost-effective, can be updated regularly, and can be applied to other regions depending on the availability of representative regional building height sample data.</p> |
first_indexed | 2024-04-13T04:22:55Z |
format | Article |
id | doaj.art-3315f02a787846c4b012f9fd0f603cd1 |
institution | Directory Open Access Journal |
issn | 1992-0628 1992-0636 |
language | English |
last_indexed | 2024-04-13T04:22:55Z |
publishDate | 2022-05-01 |
publisher | Copernicus Publications |
record_format | Article |
series | Advances in Science and Research |
spelling | doaj.art-3315f02a787846c4b012f9fd0f603cd12022-12-22T03:02:39ZengCopernicus PublicationsAdvances in Science and Research1992-06281992-06362022-05-0119132710.5194/asr-19-13-2022Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zonesE. KeanyG. BessardonE. Gleeson<p>ECOCLIMAP-Second Generation (ECO-SG) is the land-cover map used in the HARMONIE-AROME configuration of the shared ALADIN-HIRLAM Numerical Weather Prediction system used for short-range operational weather forecasting for Ireland. The ECO-SG urban classification implicitly includes building heights. The work presented in this paper involved the production of the first open-access building height map for the island of Ireland which complements the Ulmas-Walsh land cover map, a map which has improved the horizontal extent of urban areas over Ireland. The resulting building height map will potentially enable upgrades to ECO-SG urban information for future implementation in HARMONIE-AROME.</p> <p>This study not only produced the first open-access building height map of Ireland at 10 m <span class="inline-formula">×</span> 10 m resolution, but assessed various types of regression models trained using pre-existing building height information for Dublin City and selected 64 important spatio-temporal features, engineered from both the Sentinel-1A/B and Sentinel-2A/B satellites. The performance metrics revealed that a Convolutional Neural Network is superior in all aspects except the computational time required to create the map. Despite the superior accuracy of the Convolutional Neural Network, the final building height map created results from the ridge regression model which provided the best blend of realistic output and low computational complexity.</p> <p>The method relies solely on freely available satellite imagery, is cost-effective, can be updated regularly, and can be applied to other regions depending on the availability of representative regional building height sample data.</p>https://asr.copernicus.org/articles/19/13/2022/asr-19-13-2022.pdf |
spellingShingle | E. Keany G. Bessardon E. Gleeson Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zones Advances in Science and Research |
title | Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zones |
title_full | Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zones |
title_fullStr | Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zones |
title_full_unstemmed | Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zones |
title_short | Using machine learning to produce a cost-effective national building height map of Ireland to categorise local climate zones |
title_sort | using machine learning to produce a cost effective national building height map of ireland to categorise local climate zones |
url | https://asr.copernicus.org/articles/19/13/2022/asr-19-13-2022.pdf |
work_keys_str_mv | AT ekeany usingmachinelearningtoproduceacosteffectivenationalbuildingheightmapofirelandtocategoriselocalclimatezones AT gbessardon usingmachinelearningtoproduceacosteffectivenationalbuildingheightmapofirelandtocategoriselocalclimatezones AT egleeson usingmachinelearningtoproduceacosteffectivenationalbuildingheightmapofirelandtocategoriselocalclimatezones |