Land cover and soil type mapping from spaceborne PolSAR data at L-band with probabilistic neural network

Oleg Antropov, Yrjö Rauste, Heikki Astola, J Praks, Tuomas Häme, M Hallikainen

Research output: Contribution to journalArticleScientificpeer-review

40 Citations (Scopus)

Abstract

This paper evaluates performance of fully polarimetric SAR (PolSAR) data in several land cover mapping studies in the boreal forest environment, taking advantage of the high canopy penetration capability at L-band. The studies included multiclass land cover mapping, forest-nonforest delineation, and classification of soil type under vegetation. PolSAR data used in the study were collected by the ALOS PALSAR sensor in 2006-2007 over a managed boreal forest site in Finland. A supervised classification approach using selected polarimetric features in the framework of probabilistic neural network (PNN) was adopted in the study. It has no assumptions about statistics of the polarimetric features, using nonparametric estimation of probability distribution functions instead. The PNN-based method improved classification accuracy compared with standard maximum-likelihood approach. The improvement was considerably strong for soil type mapping under vegetation, indicating notable non-Gaussian effects in the PolSAR data even at L-band. The classification performance was strongly dependent on seasonal conditions. The PolSAR feature data set was further modified to include a number of recently proposed polarimetric parameters (surface scattering fraction and scattering diversity), reducing the computational complexity at practically no loss in the classification accuracy. The best obtained accuracies of up to 82.6% in five-class land cover mapping and more than 90% in forest-nonforest mapping in wall-to-wall validation indicate suitability of PolSAR data for wide-area land cover and forest mapping.
Original languageEnglish
Pages (from-to)5256-5270
Number of pages15
JournalIEEE Transactions on Geoscience and Remote Sensing
Volume52
Issue number9
DOIs
Publication statusPublished - 2014
MoE publication typeA1 Journal article-refereed

Fingerprint

soil type
land cover
synthetic aperture radar
Neural networks
Soils
boreal forest
scattering
Surface scattering
PALSAR
ALOS
vegetation
image classification
Probability distributions
Maximum likelihood
Distribution functions
Computational complexity
penetration
canopy
Statistics
Scattering

Keywords

  • Boreal forest
  • classification
  • forestry
  • land cover
  • polarimetry
  • soil type
  • synthetic aperture radar

Cite this

@article{8cedfbef10e7473bb9b3525f99d20e30,
title = "Land cover and soil type mapping from spaceborne PolSAR data at L-band with probabilistic neural network",
abstract = "This paper evaluates performance of fully polarimetric SAR (PolSAR) data in several land cover mapping studies in the boreal forest environment, taking advantage of the high canopy penetration capability at L-band. The studies included multiclass land cover mapping, forest-nonforest delineation, and classification of soil type under vegetation. PolSAR data used in the study were collected by the ALOS PALSAR sensor in 2006-2007 over a managed boreal forest site in Finland. A supervised classification approach using selected polarimetric features in the framework of probabilistic neural network (PNN) was adopted in the study. It has no assumptions about statistics of the polarimetric features, using nonparametric estimation of probability distribution functions instead. The PNN-based method improved classification accuracy compared with standard maximum-likelihood approach. The improvement was considerably strong for soil type mapping under vegetation, indicating notable non-Gaussian effects in the PolSAR data even at L-band. The classification performance was strongly dependent on seasonal conditions. The PolSAR feature data set was further modified to include a number of recently proposed polarimetric parameters (surface scattering fraction and scattering diversity), reducing the computational complexity at practically no loss in the classification accuracy. The best obtained accuracies of up to 82.6{\%} in five-class land cover mapping and more than 90{\%} in forest-nonforest mapping in wall-to-wall validation indicate suitability of PolSAR data for wide-area land cover and forest mapping.",
keywords = "Boreal forest, classification, forestry, land cover, polarimetry, soil type, synthetic aperture radar",
author = "Oleg Antropov and Yrj{\"o} Rauste and Heikki Astola and J Praks and Tuomas H{\"a}me and M Hallikainen",
note = "Project code: 71273",
year = "2014",
doi = "10.1109/TGRS.2013.2287712",
language = "English",
volume = "52",
pages = "5256--5270",
journal = "IEEE Transactions on Geoscience and Remote Sensing",
issn = "0196-2892",
publisher = "Institute of Electrical and Electronic Engineers IEEE",
number = "9",

}

Land cover and soil type mapping from spaceborne PolSAR data at L-band with probabilistic neural network. / Antropov, Oleg; Rauste, Yrjö; Astola, Heikki; Praks, J; Häme, Tuomas; Hallikainen, M.

In: IEEE Transactions on Geoscience and Remote Sensing, Vol. 52, No. 9, 2014, p. 5256-5270.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Land cover and soil type mapping from spaceborne PolSAR data at L-band with probabilistic neural network

AU - Antropov, Oleg

AU - Rauste, Yrjö

AU - Astola, Heikki

AU - Praks, J

AU - Häme, Tuomas

AU - Hallikainen, M

N1 - Project code: 71273

PY - 2014

Y1 - 2014

N2 - This paper evaluates performance of fully polarimetric SAR (PolSAR) data in several land cover mapping studies in the boreal forest environment, taking advantage of the high canopy penetration capability at L-band. The studies included multiclass land cover mapping, forest-nonforest delineation, and classification of soil type under vegetation. PolSAR data used in the study were collected by the ALOS PALSAR sensor in 2006-2007 over a managed boreal forest site in Finland. A supervised classification approach using selected polarimetric features in the framework of probabilistic neural network (PNN) was adopted in the study. It has no assumptions about statistics of the polarimetric features, using nonparametric estimation of probability distribution functions instead. The PNN-based method improved classification accuracy compared with standard maximum-likelihood approach. The improvement was considerably strong for soil type mapping under vegetation, indicating notable non-Gaussian effects in the PolSAR data even at L-band. The classification performance was strongly dependent on seasonal conditions. The PolSAR feature data set was further modified to include a number of recently proposed polarimetric parameters (surface scattering fraction and scattering diversity), reducing the computational complexity at practically no loss in the classification accuracy. The best obtained accuracies of up to 82.6% in five-class land cover mapping and more than 90% in forest-nonforest mapping in wall-to-wall validation indicate suitability of PolSAR data for wide-area land cover and forest mapping.

AB - This paper evaluates performance of fully polarimetric SAR (PolSAR) data in several land cover mapping studies in the boreal forest environment, taking advantage of the high canopy penetration capability at L-band. The studies included multiclass land cover mapping, forest-nonforest delineation, and classification of soil type under vegetation. PolSAR data used in the study were collected by the ALOS PALSAR sensor in 2006-2007 over a managed boreal forest site in Finland. A supervised classification approach using selected polarimetric features in the framework of probabilistic neural network (PNN) was adopted in the study. It has no assumptions about statistics of the polarimetric features, using nonparametric estimation of probability distribution functions instead. The PNN-based method improved classification accuracy compared with standard maximum-likelihood approach. The improvement was considerably strong for soil type mapping under vegetation, indicating notable non-Gaussian effects in the PolSAR data even at L-band. The classification performance was strongly dependent on seasonal conditions. The PolSAR feature data set was further modified to include a number of recently proposed polarimetric parameters (surface scattering fraction and scattering diversity), reducing the computational complexity at practically no loss in the classification accuracy. The best obtained accuracies of up to 82.6% in five-class land cover mapping and more than 90% in forest-nonforest mapping in wall-to-wall validation indicate suitability of PolSAR data for wide-area land cover and forest mapping.

KW - Boreal forest

KW - classification

KW - forestry

KW - land cover

KW - polarimetry

KW - soil type

KW - synthetic aperture radar

U2 - 10.1109/TGRS.2013.2287712

DO - 10.1109/TGRS.2013.2287712

M3 - Article

VL - 52

SP - 5256

EP - 5270

JO - IEEE Transactions on Geoscience and Remote Sensing

JF - IEEE Transactions on Geoscience and Remote Sensing

SN - 0196-2892

IS - 9

ER -