Tools for Housing Condition Mapping (HOCOM) were developed, which are useful for identifying buildings footprints, height, regularity, and density from VHR satellite images.
Further HOCOM contributions include the use of multi-temporal change detection by taking advantage of exchanged expertise from RMA during secondments. Specifically, we are using state-of-the-art techniques on multi-temporal change detection to monitor subsoil movement by using PS-InSAR - Persistent Scattering Interferometry SAR. This on-going work is part of Master joint-advisorship between University of Brussels, RMA and UFAL. This contribution is planned to be applied to Brazilian relevant problems such as monitoring landslides, dams, and the current problem of the Pinheiro neighborhood in Maceió city.
Regarding disease spread proxies, areas most exposed to risks of viruses carried by the Aedes Aegypti mosquito specie include Latin America, Central Africa and South-East Asia, with major disease spread outbreaks already recorded in Brazil, Argentina Colombia and Venezuela. The disease incidences recorded have been shown to correlate with mosquito vector population density. Major environmental conditions that have already proven to influence the life cycle and population density of mosquitoes include air temperature, precipitation, moisture, and vegetation condition – all reasonably measurable from freely available EO data products.
We have developed a procedure for modelling Aedes Aegypti mosquito specie based on time-series environmental variables obtainable from NASA’s MODIS EO data products and geolocated mosquito population data. To advance the state-of-the-art in this domain, our method uses non-linear relationship measures to extract the most informative features from a larger set of data representing hypothesized environmental conditions that affect the life and abundance of mosquito vectors. By this, our procedure reduces the computational redundancy and model complexity diseases spread models, and thus is application to big data. Furthermore, our procedure creates a process for empirical explainability of the local conditions that mostly affect the temporal variance of Aedes Aegypti vectors, thereby supporting principal investigators in the field of epidemiology in making informed decisions.
We are currently applying our procedure to the city of Vila Velha in Brazil and working with Universidade Federal de Alagoas (UFAL) to extend the functionality of the core algorithms implemented within our processing chain.
Regarding mapping physical proxies to security threats, PHYMAP tools aim at providing an easy way to map and monitor materials that relate to security threats against public health or safety. These proxies include landcover (roads and water bodies) landuse (settlements, crop fields and forests) and population density. Generating these proxies with high quality requires the integration of data from multisource or multitemporal Earth observation data at various spatial resolutions and quality together with open sourced data and contextual information.
PHYMAP toolbox facilitates the mapping of proxies by providing and integrating processing scheme and algorithms into a user friendly interface.
Considering the huge data sets and the required computing resources, we created PHYMAP in a cloud computing environment. The platform consists of components that support both real-time and batch processing and is ingested using SQL for optimization and distribution.
