Reconstruction of River Boundaries at Sub-Pixel Resolution: Estimation and Spatial Allocation of Water Fractions

Milad Niroumand-Jadidi,Alfonso Vitti

doi:10.3390/ijgi6120383

Abstract

Boundary pixels of rivers are subject to a spectral mixture that limits the accuracy of river areas extraction using conventional hard classifiers. To address this problem, unmixing and super-resolution mapping (SRM) are conducted in two steps, respectively, for estimation and then spatial allocation of water fractions within the mixed pixels. Optimal band analysis for the normalized difference water index (OBA-NDWI) is proposed for identifying the pair of bands for which the NDWI values yield the highest correlation with water fractions. The OBA-NDWI then incorporates the optimal NDWI as predictor of water fractions through a regression model. Water fractions obtained from the OBA-NDWI method are benchmarked against the results of simplex projection unmixing (SPU) algorithm. The pixel swapping (PS) algorithm and interpolation-based algorithms are also applied on water fractions for SRM. In addition, a simple modified binary PS (MBPS) algorithm is proposed to reduce the computational time of the original PS method. Water fractions obtained from the proposed OBA-NDWI method are demonstrated to be in good agreement with those of SPU algorithm (R2 = 0.9, RMSE = 7% for eight-band WorldView-3 (WV-3) image and R2 = 0.87, RMSE = 9% for GeoEye image). The spectral bands of WV-3 provide a wealth of choices through the proposed OBA-NDWI to estimate water fractions. The interpolation-based and MBPS methods lead to sub-pixel maps comparable with those obtained using the PS algorithm, while they are computationally more effective. SRM algorithms improve user/producer accuracies of river areas by about 10% with respect to conventional hard classification.

Highlights

The extraction of river areas is the primary task required for a wide range of remote sensing applications in fluvial systems spanning from hydrological, ecological, and morphological studies to mapping habitat suitability for different aquatic species [1,2,3,4,5,6]
Two different testing approaches are considered to survey the efficiency of Super Resolution Mapping (SRM) algorithms, while accounting for the absence and presence of uncertainty in the input data
Semi-simulated fractions are used as a contrived input with known fractions which provide a unique means of assessing the performance of the spatial allocation of sub-pixels

Summary

Introduction

The extraction of river areas is the primary task required for a wide range of remote sensing applications in fluvial systems spanning from hydrological, ecological, and morphological studies to mapping habitat suitability for different aquatic species [1,2,3,4,5,6]. To address the problem of spectral mixture, a wide variety of unmixing and soft classification algorithms, including physics-based and data-driven techniques, are developed [13,14,15,16] These techniques estimate the fraction of each class within the pixels and are represented on a set of grayscale images. This research aims at the estimation of water fractions within the mixed pixels (i.e., unmixing) and the spatial allocation of corresponding sub-pixels (i.e., SRM) to map river boundaries at the sub-pixel level. Both semi-simulated and the fractions derived from real imagery are used for evaluation of SRM techniques The first of these provides the possibility of accuracy assessment of the sole spatial allocation of sub-pixels task, while the latter considers the uncertainties involved in estimation of water fractions.

Study Area and Imagery Set

Findings

Conclusions and Suggestions