LiDAR for enhanced remote sensing of zoonotic insect vectors in the UK

Traditional methods for measuring and sampling insects are inadequate to answer the most pressing questions concerning global biodiversity losses and declines, tracking the movement of insect biomass, or the emergence of new threats from vector-borne diseases. Remote-sensing technologies are emerging, however, that will drastically improve insect ecology and monitoring. In this project, we will leverage recent advances in optical remote sensing and entomological lidar (light ranging and detection) to remotely detect and identify the most important endemic and invasive zoonotic mosquito species in the UK. Our vision is to establish lidar as a biomonitoring tool that will enhance entomological monitoring in the UK and support environmental and public health insect surveillance.
Entomological Scheimpflug lidar (ESL) (developed by project partners at Lund University) can detect hundreds of thousands of insects per day at ground-level, using spectral information gathered from backscattered light as insects traverse an infrared laser hundreds of metres long. Insects are identified based on wing-beat frequency, size and colour. The instrumentation is non-invasive, can be operated continuously, and has observed a diversity of insects ranging from bees to moths to mosquitoes. Until recently, wider use has been constrained by the specialist expertise needed to build and run both the hardware and software.
Here, we will use a new compact and 3D-printed version of the ESL designed specifically for ecologists to detect and identify important native and non-native species of UK mosquito. This will be the first time lidar has been tested for mosquito monitoring in the UK. Stimulated by the climate crisis, the UK will need to prepare and respond to mosquito-borne disease outbreaks in the coming decades, making mosquito surveillance an essential environmental and public health priority. Current monitoring remains reliant on sampling techniques that are laborious, inefficient and limited in the temporal scope of data.
In this application, we will collect ESL (spectral) signals from target mosquito vector species in the insectaries of LSTM. We will then build and train AI-models (based on convolutional neural networks) on our respective mosquito ESL datasets, before testing the ESL in a field site (based on Wirral, Cheshire) with known mosquito species diversity. Following initial testing, we will collect 3-months of ESL observations in a UK wetland to estimate emerging mosquito population sizes and microclimatic drivers with unprecedented accuracy. Finally, we will demonstrate the feasibility of the ESL as a biomonitoring tool, deploying the instrument in different habitats across the UK, and in priority surveillance areas (i.e. threat of invasive species at seaports).
Importantly, while our primary focus is zoonotic mosquito vectors, we will collect ESL signal data from across UK insect fauna. The datasets we generate can therefore be re-used to target other insects of interest, and we see this project as a springboard for developing and integrating lidar-based remote insect detection across UK entomological programmes, from biodiversity assessments to pest control.