Evaluation of Sentinel-1 Interferometric SAR Coherence efficiency for Land Cover Mapping
Authors
Abstract:
In this study, the capabilities of Interferometric Synthetic Aperture Radar (InSAR) time series data and machine learning have been evaluated for land cover mapping in Iran. In this way, a time series of Sentinel-1 SAR data (including 16 SLC images with approximately 24 days time interval) from 2018 to 2020 were used for a region of Ahvaz County located in Khuzestan province. Using InSAR processing, 25 coherence images were obtained based on different SAR pairs. Five dominant land cover classes in the region including built-up lands, agricultural lands, water bodies, bare soil, and dense natural vegetation cover were identified and considered. Through Google Earth's high-resolution imagery, a total of 4,930 ground truth samples with appropriate spatial distribution were acquired for all classes. The obtained multi-temporal coherence images were used as input variables to the support vector machine (SVM) classifier. The training and validation process of different SVM kernels was performed using 80% and 20% of the ground truth samples, respectively. Overall accuracy in different kernels including linear, 2th-degree polynomial, 4th-degree polynomial, 6th-degree polynomial, radial base function (RBF), and sigmoid were computed 60.7, 64.7, 67.7, 69.9, 66.3, and 59.5%, respectively. Likewise, Kappa coefficients were reported 50. 8, 55.87, 59.62, 62.38, 57.87, and 49.38%, respectively. Accordingly, the highest and lowest overall accuracy and Kappa coefficient were belong to the 6th-degree polynomial and sigmoid kernels, respectively. Based on the user and producer accuracy assessments in all kernels, the built-up lands has the highest accuracy (93%–up to 98.5%), and in opposite the dense vegetation has the lowest accuracy (11%–up to 56.25%). Generally, the results emphasize the high potential of Sentinel-1 InSAR coherence data in land cover mapping. Meanwhile, the contribution of the classifier to the efficiency of data is also important.
similar resources
Application of Multitemporal Interferometric SAR Data for Land Cover Mapping in Mongolia
The aim of this study is to use the original and other ancillary derived features extracted from multitemporal spaceborne interferometric synthetic aperture radar (InSAR) data sets for land cover mapping. For the actual land cover mapping, supervised (statistical maximum likelihood classification) and unsupervised (isodata clustering) classifications have been performed using a) the original In...
full textLand-Cover Mapping Using Airborne Polarimetric SAR
The potential of synthetic aperture radar (SAR) for land-cover mapping applications has since long time been recognized. The main advantage of the SAR is its all-weather mapping capability, which secures the mapping independent of cloud cover that may prevent acquisition by optical and infrared sensors. Furthermore, the backscattering of the radar waves is sensitive to the dielectric properties...
full textLand cover and forest mapping in boreal zone using polarimetric and interferometric SAR data
Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi Author Oleg Antropov Name of the doctoral dissertation Land cover and forest mapping in boreal zone using polarimetric and interferometric SAR data Publisher School of Electrical Engineering Unit Department of Radio Science and Engineering Series Aalto University publication series DOCTORAL DISSERTATIONS 3/2014 Field of research Spac...
full textEvaluation of Multi-Frequency SAR Images for Tropical Land Cover Mapping
Earth Observation (EO) data plays a major role in supporting surveying compliance of several multilateral environmental treaties, such as UN-REDD+ (United Nations Reducing Emissions from Deforestation and Degradation). In this context, land cover maps of remote sensing data are the most commonly used EO products and development of adequate classification strategies is an ongoing research topic....
full textUrban Land-cover Mapping with High-resolution Spaceborne SAR Data
.................................................................................................. I Acknowledgements ............................................................................. III List of Appended Papers ...................................................................... VI
full textMy Resources
Journal title
volume 9 issue None
pages 85- 107
publication date 2021-12
By following a journal you will be notified via email when a new issue of this journal is published.
No Keywords
Hosted on Doprax cloud platform doprax.com
copyright © 2015-2023