Abstract Summary
The City of Amsterdam is facing specific problems relating to climate change. One of these problems is the increase in Harmful Algae Blooms (HABs) in the extensive canal network of the city. HABs are the spontaneous and excessive growth of algae capable of producing toxins harmful to humans as well as marine animals and birds. The occurrence of HABs has a detrimental effect on the water quality and biodiversity of the canals and also prevents the recreational use of water bodies affected by them. Currently, many urban phenomena are measured by using stationary sensors spread around the city. However, it has been shown that the use of drive by sensing, a new paradigm in urban sensing, can increase the spatiotemporal coverage of sensor networks. For instance, MIT’s City Scanner project aims to transform everyday vehicles such as buses and taxis into sensing nodes for measuring air quality, showcasing the potential of drive-by sensing in expanding the spatiotemporal coverage of sensor networks. When it comes to the monitoring of algae growth in Amsterdam. The current approach of Waternet (the water authority of Amsterdam) is the use of several fixed sampling locations across the city. While this is a good start at monitoring HABs, the fact that the sensors are fixed makes the spatial coverage quite low. This is a problem because the excessive growth of algae colonies is a hyper-local phenomenon with hyper-local implications. The ambition is to use drive-by sensing similar to that in the City Scanner project to capture high-resolution data of algal colonies. Researchers at the Senseable City Lab have therefore designed a mobile LED-induced fluorescence spectrometer that can be placed on boats owned by the municipality. This new device, called REMORA, can make measurements autonomously, without the need for human interference. The REMORA sensor can be used to detect, quantify, and identify algae species. The sensor consists of an LED light source emitting light at certain wavelengths. This light excites the pigments in the algae which then emit light at a certain wavelength as well. These wavelengths are then measured and collected in the output as an Excitation Emission Matrix (EEM). This high dimensional datatype is used in several different contexts. Recently, researchers have detailed the use of machine learning in analyzing EEM data. This research aims to find the model best suited for the classification of algae species, using EEM data from the REMORA sensor. Therefore making it possible to analyze the real time growth of algae in Amsterdam. The following research question has been devised: ”Could the performance of the REMORA sensor in identifying and quantifying algae species be improved by using machine learning techniques?”
