Detection of plastic film residues on cropland using remote sensing techniques: from proximal to UAV-based remote sensing

Plastic residues found on cropland span across a wide range of polymers, shapes, sizes, and sources, which are all factors influencing the environmental risks of plastic pollution. A major potential plastic source of cropland soil systems is the intentional use of plastic in agriculture. Despite their role in improving crop production and management, plastic films have been associated with the generation of macro- and micro-plastic residues. Understanding the variables driving plastic film pollution and residue generation is important to quantify the overall input into soil systems and design good management practices. However, plastic residues are usually collected by manual sampling and quantified by ex-situ analyses, which is very time-consuming and limits data acquisition to small areas.

This study aims to analyse the potential of remote sensing data acquired from Unmanned Aerial Vehicles (UAVs) as a fast and cost-effective method for the detection of macroplastic residues on cropland. To understand the factors influencing plastic film detection when moving from pristine films to residues in soil, we collected hyperspectral (i.e., with a spectroradiometer) and multispectral data (i.e., with a multispectral camera) on different plastic films in an experimental outdoor setup. We used two different soils as a background and simulated the following conditions: pristine films, crumpled films, soil-covered films, and crumpled and soil-covered films. In this way, we built spectral libraries and identified absorption peaks. Moreover, the simulated spectral signature of plastic film residues matched the magnitude of changes observed in field samples available for black mulching films. We then assessed major challenges and possible workflows for detecting plastic residues with a multispectral camera mounted on a UAV.