Hyperspectral reflectance of plastic samples and natural background materials at the shores of Lake Victoria in Uganda measured using an SR-3501 Spectral Evolution (SEV) Spectroradiometer

Plastic samples were collected from Kasenyi Fish Landing Site and Mulungu Beach on the northern shores of Lake Victoria, Uganda, between 21st and 24th April 2025. An SR-3501 Spectral Evolution (SEV) spectroradiometer, which records the full solar-reflected spectrum (280nm-2500nm), was used to measure the reflectance of these samples. Because the experiment was conducted outdoors, a very low-reflectance black material was sourced locally and used as the background for the plastic samples. To enable modelling of mixed-pixel environmental scenarios and to examine background contributions, reflectance data for representative natural-environment background surfaces (water, vegetation and sand) were also measured at one of the sites (Kasenyi Fish Landing Site). Furthermore, to account for differences in water and vegetation types in natural freshwater environments, spectra of both shallow water with visible bottoms and turbid water were recorded. Similarly, spectra of both floating vegetation and vegetation (grass) at the back of the beach were measured. Using the same experimental setup as described in Atuhaire et al. (2026), the reflectance of each sample was measured with the SEV instrument, operated in DARWin SP v1.5 software and configured to record in reflectance mode. The relative reflectance of each sample was automatically obtained by white referencing on a 99% reflectance panel (SG3120 standard Sphere Optics GmbH Zenith Polymer). For each plastic sample category, spectra were obtained by taking multiple readings on each sample to ensure all sides were measured. Representative reflectance spectra for each plastic sample and natural-environment background materials were obtained by averaging multiple measurements recorded for each category using the DARWin SP v1.5 software. Furthermore, the documented water absorption wavelengths, i.e., 1800nm-1960nm, and data beyond 2350nm were masked out during data processing due to high-frequency noise observed in the measurements.