Scent Newsletter #1 | The technologies behind Scent

This article originally appeared in the first Scent Newsletter (July 2017). To receive all of the articles and features in one place, make sure to subscribe to the Scent newsletter here.

Scent System Architecture
The Scent project will enable citizens to become the ‘eyes’ of policy makers by monitoring land-cover/use changes in their everyday activities. This is achieved through the following integrated set of smart collaborative technologies:

• Low-cost / portable data collection tools

• An innovative crowd-sourcing platform

• Serious gaming applications for a large-scale image collection and semantic annotation

• A powerful machine-learning based intelligence engine for image and text classification

• An authoring tool for easy customisation by policy makers

• Numerical models for mapping land-cover changes to quantifiable impact on flood risks

• A harmonisation platform consolidating data and adding it to GEOSS as OGC-based observations

Infographic of the movement of Environmental Data from different sources to be processed by the Scent Intelligence Engine (SIE)

The architectural design of the system followed a well-established methodology that led to a comprehensive architecture description, which serves the following goals:

• Complement existing in-situ infrastructure through a network of low-cost portable sensors and smart collaborative technologies

• Generate an unprecedented and low-maintenance pool of near real-time observations that will monitor land cover/ use changes/environmental phenomena

• Offer data via web services into existing repositories, such as the GEOSS portal

• Allow policy-makers to obtain a dynamic view of land-use changes while directly involving citizens in the decision-making process

• Go beyond a research prototype level by clearly fitting and extending evolving market needs through its operational sustainability and through complementing existing initiatives (e.g. the Geo-wiki)

Scent Innovative Sensors
The main objective of the work package on innovative sensors is to identify suitable low-cost, portable, market-ready sensors that measure soil moisture, water / air temperature and other relevant parameters.

In addition, digital cameras integrated into drones based on open source hardware platforms for obtaining spatially continuous data with an increased accuracy will be designed and implemented. The main activities to date are summarised below:
• A detailed list of identified existing sensorial solutions that meet the defined user needs was compiled

• These solutions were evaluated based on the Scent System Requirements

• Existing drone platforms and related software components were evaluated

• Based on the aforementioned steps, a set of portable and flying sensing devices was proposed

Diagram of the water sensors that are in place at the pilot study sites – the Danube Delta river in Romania and the Kifisos river basin in Greece

A drone deploying water sensors for data

Flooding Patterns
The work package focusing on flooding patterns started with the classical approach of building the models that will represent the flooding behaviour in the two pilot sites of Scent: the Sontea Fortuna area in the Danube Delta in Romania and the Kifisos catchment around Athens in Greece.

The entire Danube Delta was modelled as a one-dimensional hydrodynamic model for its three main branches. Based on water level and discharges obtained in this model, boundary conditions were set for three detailed models of the pilot in the Danube Delta – the Sontea Fortuna area. The three detailed hydrodynamic models are:

• A one-dimensional model of the two main Danube branches in the pilot are together with the canals in the area

• The previous model extended with the lakes that are located in the pilot area

• A 1D / 2D hydrodynamic model of the whole pilot area

Modelling of the second pilot area just started and a synthetic rainfall event is now considered for the hydrological model of the Kifisos catchment. The next steps will include testing the model with real rainfall.
All three models were calibrated for discharge measurements during dry periods and results identified the time and location of stagnant water in some of the canals in the pilot area. Currently the modelling work focuses on validation and flooding patterns for the wet periods.