Improving neural network classification of indigenous forest in New Zealand with phenological features
(2022) In Journal of Environmental Management- Abstract
- Accurate and up-to-date land cover maps inform and support effective management and policy decisions. Describing phenological changes in spectral response using time-series data may help to distinguish vegetation types, thereby allowing for more specificity within vegetation classification. In this research, we test this by classifying indigenous forest vegetation in New Zealand, using PlanetScope (PS) and Sentinel-2 (S-2) satellite time-series data. The study was undertaken in a podocarp forest in New Zealand's central north island, which was classified into nine land cover classes. Phenological features, based on S-2 imagery, were extracted, including the enhanced vegetation index (EVI), enhanced vegetation index 2 (EVI2) and normalised... (More)
- Accurate and up-to-date land cover maps inform and support effective management and policy decisions. Describing phenological changes in spectral response using time-series data may help to distinguish vegetation types, thereby allowing for more specificity within vegetation classification. In this research, we test this by classifying indigenous forest vegetation in New Zealand, using PlanetScope (PS) and Sentinel-2 (S-2) satellite time-series data. The study was undertaken in a podocarp forest in New Zealand's central north island, which was classified into nine land cover classes. Phenological features, based on S-2 imagery, were extracted, including the enhanced vegetation index (EVI), enhanced vegetation index 2 (EVI2) and normalised difference vegetation index (NDVI). Google Earth Engine (GEE) harmonic analysis and TIMESAT double logistic fitting function were used to extract phenological features. Pixel-based classifications were performed using a Neural Network on six different scenarios. The accuracy of the classification scenarios was determined and the importance score for each feature was evaluated. Using only the fused PS and S-2 bands, the land cover in the study area was classified with 90.1% accuracy. Adding phenological features increased the classification accuracy to 93.1%. When combined with VIs, texture features, and a digital terrain model, the addition of phenological features increased the classification accuracy to 96.6%. Including GEE-generated phenological features resulted in better classification accuracies than TIMESAT features. In terms of feature importance evaluation, EVI2- and NDVI-generated phenological features all had high scores; the effectiveness of EVI features could potentially have been limited by the quality of the blue band. The results demonstrate that it is possible to produce a more accurate classification of New Zealand's native vegetation by using phenological features. This method offers important cost-savings as the platforms for phenological analysis are free to use. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/6d22b7a3-9cba-4391-bf8c-185177a48312
- author
- Ye, Ning ; Morgenroth, Justin ; Xu, Cong and Cai, Zhanzhang LU
- organization
- publishing date
- 2022-04-23
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Time-series data, Google earth engine, Phenology, Vegetation classification, Machine learning
- in
- Journal of Environmental Management
- article number
- 115134
- publisher
- Elsevier
- external identifiers
-
- pmid:35472842
- scopus:85129064064
- ISSN
- 0301-4797
- DOI
- 10.1016/j.jenvman.2022.115134
- language
- English
- LU publication?
- yes
- id
- 6d22b7a3-9cba-4391-bf8c-185177a48312
- date added to LUP
- 2022-04-28 16:01:56
- date last changed
- 2022-05-24 04:00:16
@article{6d22b7a3-9cba-4391-bf8c-185177a48312,
abstract = {{Accurate and up-to-date land cover maps inform and support effective management and policy decisions. Describing phenological changes in spectral response using time-series data may help to distinguish vegetation types, thereby allowing for more specificity within vegetation classification. In this research, we test this by classifying indigenous forest vegetation in New Zealand, using PlanetScope (PS) and Sentinel-2 (S-2) satellite time-series data. The study was undertaken in a podocarp forest in New Zealand's central north island, which was classified into nine land cover classes. Phenological features, based on S-2 imagery, were extracted, including the enhanced vegetation index (EVI), enhanced vegetation index 2 (EVI2) and normalised difference vegetation index (NDVI). Google Earth Engine (GEE) harmonic analysis and TIMESAT double logistic fitting function were used to extract phenological features. Pixel-based classifications were performed using a Neural Network on six different scenarios. The accuracy of the classification scenarios was determined and the importance score for each feature was evaluated. Using only the fused PS and S-2 bands, the land cover in the study area was classified with 90.1% accuracy. Adding phenological features increased the classification accuracy to 93.1%. When combined with VIs, texture features, and a digital terrain model, the addition of phenological features increased the classification accuracy to 96.6%. Including GEE-generated phenological features resulted in better classification accuracies than TIMESAT features. In terms of feature importance evaluation, EVI2- and NDVI-generated phenological features all had high scores; the effectiveness of EVI features could potentially have been limited by the quality of the blue band. The results demonstrate that it is possible to produce a more accurate classification of New Zealand's native vegetation by using phenological features. This method offers important cost-savings as the platforms for phenological analysis are free to use.}},
author = {{Ye, Ning and Morgenroth, Justin and Xu, Cong and Cai, Zhanzhang}},
issn = {{0301-4797}},
keywords = {{Time-series data; Google earth engine; Phenology; Vegetation classification; Machine learning}},
language = {{eng}},
month = {{04}},
publisher = {{Elsevier}},
series = {{Journal of Environmental Management}},
title = {{Improving neural network classification of indigenous forest in New Zealand with phenological features}},
url = {{http://dx.doi.org/10.1016/j.jenvman.2022.115134}},
doi = {{10.1016/j.jenvman.2022.115134}},
year = {{2022}},
}