Spectrally resolved insect flashes by sunlight

Abstract

Insects are vital for the environment and biodiversity. Optics-based remote techniques can quickly and effectively identify and track insects in their natural habitat. This has the potential to speed up and improve the time-consuming process of manually monitoring insect diversity. This thesis presents a method to remotely sense insects by analyzing the sunlight reflected by free-flying insects. The study proposes a novel approach by simultaneously capturing the spectral and temporal signals of reflected sunlight, containing new information about the spectral properties of insects in the visible wavelength range. For this project, a spectrometer is designed using raytracing, CAD-design, and 3D-printing. The instrument is capable of analyzing light in the visible to near-infrared range (420-840 nm) with 32 spectral bands. The instrument was calibrated and successfully deployed in the field to capture sunlight flashes from free-flying insects, with a sampling rate of 20 kHz. The collected data shows that the instrument successfully captured signals from insects. This data is used to determine if spectrally resolved modulation provides additional information for species differentiation. To achieve this, a hierarchical clustering algorithm is applied to both the spectrally resolved and non-spectrally resolved signals. The number of clusters found when using both spectral and temporal data was larger than the number of clusters found when using only temporal data. The conclusion is drawn that there is additional information for species identification in the spectral data.

Popular Abstract

Insects are an incredibly important part of our ecosystem. They serve as pollinators for many of the fruits and vegetables that we eat. Due to climate change, the number of insects and the amount of insect species is declining at an alarmingly high rate, which can have catastrophic consequences for all other life on Earth. It is important to closely monitor this decline so that we can combat it effectively. However, it is difficult to find out how and where this decline is taking place. Current methods for insect monitoring rely on manual counting which is labor intensive. Automatic monitoring of insects could give better insight into this issue, which would allow us to combat it more effectively.

This thesis presents an automatic method to monitor insects by capturing and analyzing the spectrum of sunlight that is reflected by free-flying insects. This method is also referred to as Dark-Field Spectroscopy. A large advantage over manual counting of insects is that this does not harm the insects. Additionally, insects use sunlight to see each other. Therefore, these measurements can also provide clues about the mechanism behind how insects see each other. The research question that forms the basis of this thesis is whether the spectral analysis of the reflected sunlight contains extra information for identifying the species.

The sunlight reflected by insects is captured by a large telescope that is placed in a field. A big black box is placed 100 meters away from the telescope. When there is nothing in the field of view of the telescope, it only sees the black box. When an insect flies through the field of view, it will reflect sunlight into the telescope. The light captured by the telescope is then sent into a spectral analyzer which can determine the different wavelengths that were reflected by the insect, which is also referred to as the spectrum. The spectrometer takes 20,000 measurements every second. The fastest wing beat is from mosquitoes, who beat their wings 1,000 times per second. This means that their spectrum is recorded 20 times during one wing beat.

During three days of recording in the field, 24.000 insects were observed flying through the field of view. The collected data is rich in information. The wing beat frequency is determined by light intensity fluctuations. Opaque wings exhibit thin-film interference, causing spectral fringes from which the wing thickness can be derived. The entire dataset is used to estimate the number of species through clustering, which means grouping similar observations to determine the number of species present.

The data clustering was done with and without the spectral information. Clustering with spectral information resulted in more groups compared to clustering without spectral information. The main conclusion of this project is that the spectral information helps to identify more species. Additionally, the data showed that sunlight reflected by insects has up to eight spectral components. This could be an explanation for why, for example, a dragonfly has five effective spectral bands.

• author: Hendriks, Isabel Eline
• supervisor: Mikkel Brydegaard; MEng Li; Hampus Månefjord
• organization: Department of Physics; Combustion Physics
• course: FYSM33 20232
• year: 2024
• type: H2 - Master's Degree (Two Years)
• subject: Physics and Astronomy
• keywords: Spectroscopy, Passive Lidar, Dark Field Spectroscopy, Remote sensing, Insect monitoring
• language: English
• id: 9148814
• date added to LUP: 2024-02-23 13:07:23
• date last changed: 2024-02-23 13:07:23