Impact of Image-Processing Routines on Mapping Glacier Surface Facies from Svalbard and the Himalayas Using Pixel-Based Methods

Glacier surface facies are valuable indicators of changes experienced by a glacial system. The interplay accumulation and ablation facies, followed intermixing with dust debris, as well the local climate, all induce observable mappable on supraglacial terrain. In absence or lag continuous field monitoring, remote sensing observations become vital for maintaining constant supply measurable data. However, satellite suffer from atmospheric effects, resolution disparity, use multitude mapping methods. Efficient image-processing routines are, hence, necessary to prepare test derivable data applications. existing literature provides an application-centric view selection image processing schemes. This can create confusion, it is not clear which method correction would be ideal retrieving spectral reflectance, nor effects pansharpening examined facies. Moreover, variety supervised classifiers target detection methods now available, prudent impact variations in schemes resultant thematic classifications. this context, current study set its experimental goals. Using very-high-resolution (VHR) WorldView-2 data, we aimed three common methods, viz. Dark Object Subtraction (DOS), Quick Atmospheric Correction (QUAC), Fast Line-of-Sight Analysis Hypercubes (FLAASH); two Gram–Schmidt (GS) Hyperspherical Color Sharpening (HCS), classification using 12 classifiers. conventional included: Mahalanobis Distance (MHD), Maximum Likelihood (MXL), Minimum Mean (MD), Spectral Angle Mapper (SAM), Winner Takes All (WTA). advanced/target consisted of: Adaptive Coherence Estimator (ACE), Constrained Energy Minimization (CEM), Matched Filtering (MF), Mixture-Tuned (MTMF), Target-Constrained Interference-Minimized Filter (MTTCIMF), Orthogonal Space Projection (OSP), (TCIMF). experiment was performed glaciers at sites, Ny-Ålesund, Svalbard, Norway; Chandra–Bhaga basin, Himalaya, India. overall performance suggested that FLAASH delivered realistic reflectance spectra, while DOS least realistic. Spectra derived HCS sharpened subsets seemed match average trends, whereas GS reduced reflectance. WTA achieved highest accuracy (0.81). MTTCIMF yielded lowest 0.01. consistently provided better (less variable generally accurate) than QUAC, making more reliable hence recommended algorithm. While HCS-pansharpened lower error rate (0.71) comparison (0.76), neither significantly improved efficiency. Ny-Ålesund glacier were best classified MXL (error = 0.49) 0.53), Himalayan MD 0.61) 0.45). final comparative analysis based total across corrections following reliability order: > MHD ACE CEM MF SAM MTMF TCIMF OSP MTTCIMF. findings VHR visible near-infrared (VNIR) correction, may deliver classification. extensive account varying exertions each scheme discussed, could transferable when compared against other VNIR