High population densities and infrastructure in cities make them extremely vulnerable to disasters that challenge their ability to mitigate and adapt and enhance their resilience. A data-centric approach to map complex relationships between the city’s social, economic, political and administrative components is critical to manage disasters and plan mitigation and adaptation efforts. A multi-resolution approach and data integration is essential to plan and model the city’s ability to mitigate and adapt to extreme weather events. At the Center we use a range of spectral and spatial resolutions [VIS-NIR-SWIR] facilitated by remotely sensed data sets to model the City processes at all scales.

Urban areas are characterized by unplanned growth, sprawl, frequent land cover and land use changes, and population concentrations. Planning urban areas requires accurate information over space and time. A wide array of remotely sensed data acquired from spaceborne and airborne platforms offer great promise in modeling urban environments. However spectral confusion caused by mixed signals makes feature extraction from urban satellite datasets challenging and inaccurate. We have developed models to extract urban features by using a combination of per-pixel and object-based approaches to extract features based on their spectral and spatial attributes. The extracted features may be used for various applications ranging from human health and environment to climate change and disaster management.

Global solutions demand the analyses of BIG DATA acquired from multiple sensors. The volume of data is disproportionately matched by the quantum of analyses performed. The availability of large volumes of data and their global spatio-temporal coverage may be harnessed by innovative methods of classification and a multi-modal approach consisting of machine learning, deep learning and customized neural network architectures. We build foundation models by using combination of unsupervised, supervised, per-pixel, object-based machine learning, deep learning, NN techniques and by testing the model on global datasets use them to design and build solutions for a wide range of issues [Climate Change, Urban planning, disaster management, policy decisions]

Urban heat islands demand rapid solutions and nature-based solutions for their mitigation and adaptation. At the Center we work with a range of spaceborne imagery to derive Land Surface Temperature [LST] to map concentrations of heat islands in designated disadvantaged zone of New York City boroughs. By fine tuning the satellite-based retrievals with fine resolution field data acquired from portable and static temperature | weather monitoring stations and using AI and statistical techniques we derive mitigation and adaptation policies and informed decision making processes at borough and city scales.