Design and Development of Open Source Software Solution for Hyperspectral Data Simulation: Hydas.

Abstract Multispectral remote sensing data is available with broad spectral bands in the wavelength range of Visible-NIR-SWIR, and finds its applications in assessment of Land Use Land Cover classes (LULC). There are plenty of onboard sensors which are providing these datasets regularly. However, for detailed LULC classification there is a requirement of distinguishing spectrally similar land cover features, which is possible through hyperspectral data. These datasets are spectrally rich and are available with narrow spectral bandwidth but the availability of such datasets is limited. This is due to requirement of sophisticated detectors and high storage, which makes the acquisition of these datasets challenging. Therefore, there is a opportunity to utilize the available multispectral data for generating hyperspectral data. Existing spectral reconstruction technique is used for simulating hyperspectral bands from multispectral bands in combination with ground spectra. In this study, an attempt has been made to develop a software solution for automating the simulation of hyperspectral data using multispectral data in an open source platform. The developed software solution (HyDAS) consists of modules for satellite data visualization, spectral library creation, spectral unmixing and hyperspectral data simulation. HyDAS toolbox is successfully tested for simulating hyperspectral data from EO-1 ALI multispectral data. The simulated results were validated using the available EO-1 Hyperion data and results were satisfactory with similarity being 80–90%. This open source solution was also tested for simulating hyperspectral bands from EO-1 ALI and Indian Satellite (LISS III & LISS IV) multispectral datasets, the results obtained were satisfactory for species level classification