Spatially Adaptive Modelling via a Local Smoothing Algorithm (SALSA) and Some Recent Applications in Environmental Impact Assessment (EIA)


  • Dr. Monique Mackenzie – Centre for Resarch into Ecological and Environmental Modelling (CREEM) – University of St. Andrews – UK
  • FCUL (DEIO) – Campo Grande – Bloco C6 Piso 4 – Sala 64.30 – 15:00 horas
  • Quarta-feira, 8 de Fevereiro de 2012
  • Referência Projeto: PEst-OE/MAT/UI0006/2011

Most off-the-shelf smoothing methods are “globally smooth” and the flexibility permitted in the model (via the single smoothing parameter) is constant across the entire surface/x-range. This is largely unrealistic and more flexibility is typically required in some areas of the surface/x-range than in other areas. Recent methods we’ve been developing are more realistic and permit the flexibility to change across the surface as dictated by objective fit criteria. Specifically, a spatially adaptive local smoothing algorithm (SALSA) has been developed for use in spline-based models in one and two dimensions with good results. SALSA will be described and illustrated using marine mammal and Environmental Impact Assessment (EIA) data including an example from one of the world’s largest off-shore wind farms.