Journal article 116 views 7 downloads
A Spatial Pattern Extraction and Recognition Toolbox Supporting Machine Learning Applications on Large Hydroclimatic Datasets
Remote Sensing, Volume: 14, Issue: 15, Start page: 3823
Swansea University Authors: Han Wang, Yunqing Xuan
PDF | Version of Record
© 2022 by the authors.This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) licenseDownload (7.68MB)
DOI (Published version): 10.3390/rs14153823
This paper presents the development and applications of a new, open-source toolbox that aims to provide automatic identification and classification of hydroclimatic patterns by their spatial features, i.e., location, size, orientation, and shape, as well as the physical features, i.e., the areal ave...
|Published in:||Remote Sensing|
Check full text
No Tags, Be the first to tag this record!
This paper presents the development and applications of a new, open-source toolbox that aims to provide automatic identification and classification of hydroclimatic patterns by their spatial features, i.e., location, size, orientation, and shape, as well as the physical features, i.e., the areal average, total volume, and spatial distribution. The highlights of this toolbox are: (1) incorporating an efficient algorithm for automatically identifying and classifying the spatial features that are linked to hydroclimatic extremes; (2) use as a frontend for supporting AI-based training in tracking and forecasting extremes; and (3) direct support for short-term nowcasting of extreme rainfall via tracking rainstorm centres and movement. The key design and implementation of the toolbox are discussed alongside three case studies demonstrating the application of the toolbox and its potential in helping build machine learning applications in hydroclimatic sciences. Finally, the availability of the toolbox and its source code is included.
Data Availability Statement: The datasets for testing the toolbox were provided by the Centre ofHydrology and Ecology (CEH) and available in https://doi.org/10.5285/33604ea0-c238-4488-813d-0ad9ab7c51ca (accessed on 4 August 2022), and the details of the catchments can be found inhttps://nrfa.ceh.ac.uk/content/catchment-boundary-and-areas .
spatiotemporal pattern; pattern recognition; cluster analysis; machine learning; pattern identification and classification; rainfall monitoring and nowcasting
Faculty of Science and Engineering
This research was funded by the Academy of Medical Sciences, Grant Ref: GCRFNGR4_1165.
The Guangdong Radar data were provided by the UK-China joint project “Early Warning Systems
for Urban Flooding in Chinese Mega Cities using Advanced Phased Array Radar” supported by the
Royal Academic of Engineering, Grant Ref: UUFRIP-10021.