Knowledge- and Data-driven Geospatial Big Data Analytics

Main Analysis and Modelling Techniques We Use

Rapid urban expansion shown by Landsat-7 (Left) / Landsat-8 (Right) imagery between 2000-2015. Location: Shenzhen – Hong Kong border, China.

Advanced Image Processing and Feature Extraction

  • Object- and pixel-based image classification
  • Orthorectification, image fusion, image enhancement, computer vision techniques
  • Land covers change detection and trend analysis
  • Pre- and post-event site profiling and investigation
  • Analysing short-wave infrared (SWIR) bands (e.g. from Sentinel-2, Landsat-8, WorldView-3 imagery)
  • Making large-sized imagery mosaics (e.g. up to multiple terabytes in file sizes)

Suburban buildings and vegetation shown by the USGS LiDAR data captured in 2014. Location: New York City, the U.S.

Terrain and Geomorphological Modelling

  • Digital Elevation Model (DEM), Digital Surface Model (DSM) and Digital Terrain Model (DTM) generation
  • Contour generation
  • Processing large-sized LiDAR Point Clouds (e.g. DEM generation, ground feature extraction)
  • Developing geomorphological metrics for terrain characterisation and quantification
  • Making large-sized elevation and shaded-relief mosaics (e.g. up to multiple terabytes in file sizes)

GIS, Mapping and Visualisation

  • Mainstream GIS geospatial analyses (both raster- and vector-based)
  • Development and customisation of geospatial algorithms
  • Advanced spatial interpolations and geocoding (e.g. adaptive and context-aware)
  • Spatial data management (e.g. through PostGIS/MySQL/Microsoft SQL Server spatial databases, ESRI Geodatabase)
  • 2D/3D cartographic mapping, web mapping (e.g. Google/Bing/ESRI/OSM/MapBox/CartoDB tileset), time-series visualisation

Applied Maths

  • Probability and statistics, extreme value distributions and analysis, curve fitting
  • Operations research, linear programming, multi-criteria evaluation and decision support, optimisation
  • Spatio-temporal geostatistical modelling, spatial uncertainty analysis
  • Wavelet transformation in image analysis and geoscience
  • Numerical methods (e.g. approximation and finite difference methods)

Artificial Intelligence, Machine Learning and Predictive Analytics

  • Applications of both supervised and unsupervised learning methods
  • Applications of artificial neural networks and machine learning (especially the latest deep learning) for image classification and feature extraction
  • Comparison between machine learning methods and conventional statistical methods in time-series prediction
  • Comparison of various clustering techniques and cluster analyses
  • Applications of data mining and knowledge discovery in databases (KDD) techniques

Environmental/Geophysical Modelling and Simulation

  • Modelling environmental degradation and impact (e.g. air/water/soil pollution)
  • Deterministic “what-if” scenario analysis (e.g. for physical and socioeconomic attributes)
  • Monte Carlo stochastic simulation (e.g. for physical environmental attributes)
  • Earth Systems modelling (e.g. for climate change impact, population dynamics, biodiversity, land cover change and sustainability)
  • Model comparisons and critical evaluation
  • Model development, computer programming and software release