Transcript Document
C17 SC for Environmental Applications and Remote Sensing I M S C I A
Soft Computing for Environmental Applications and Remote Sensing
Environmental applications
Fabio Scotti - Manuel Roveri Università degli studi, Milano, Italy
U N I V E R S I T À D E G L I S T U D I D I M I L A N O
C17 SC for Environmental Applications and Remote Sensing I M S C I A
Introduction • This lecture introduces examples of environmental applications of the Remote Sensing technologies • In the first part of the lesson we proposes some examples of classical approaches • In the second part of the lesson we describes some solutions solved using soft-computing techniques: – Oil Spill Detection; – Biomass measurement; – Satellite Cloud Classification.
Fabio Scotti - Manuel Roveri
C17 SC for Environmental Applications and Remote Sensing I M S C I A
Overview of Environmental applications – – – – – – – – Agriculture ( Crop Type Mapping, Crop Monitoring) Forestry ( Clear cut Mapping, Species identification, Burn Mapping) Geology ( Structural Mapping, Geologic Units) Hydrology ( Flood Delineation, Soil Moisture) Sea Ice ( Type & concentration, Ice Motion) Land Cover ( Rural/Urban change, Biomass Mapping) Mapping ( Planimetry, DEMs, Topo Mapping) Oceans & Coastal ( Ocean Features, Ocean Colour, Oil Spill Detection) • Please read the tutorial
L4_Enviro1.pdf
(*) linked in the course page.
(*) Canada Centre for Remote Sensing Fabio Scotti - Manuel Roveri
C17 SC for Environmental Applications and Remote Sensing I M S C I A
Neural Networks for Oil Spill Detection
• We mention here a neural network approach for semi-automatic detection of oil spills.
• The goal of the reading is to understand the application problem, the topology selection of the neural network, the creation of the datasets and the network testing phase.
• Please read carefully the paper
L4_Enviro2.pdf
(*) linked in the course page.
Fabio Scotti - Manuel Roveri
C17 SC for Environmental Applications and Remote Sensing I M S C I A
Neural Networks for biomass measurements
• In this application we have a neural network used to estimate the forest stand biomass.
• Interestingly the NNs has been considered in three situations: – trained on model data to invert model values; – trained on real data; – to invert actual measurements, and trained on simulated data to invert measured data.
• The goal of the reading is to understand the application problem, and the consider the adopted methodology to design the neural networks.
• Please read carefully the paper
L4_Enviro3.pdf
(*) linked in the course page.
Fabio Scotti - Manuel Roveri
C17 SC for Environmental Applications and Remote Sensing I M S C I A
Neural Networks for Satellite Cloud Classification
• In this application we have a neural network used to classify clouds from satellite images.
• A temporal updating approach for probabilistic neural network (PNN) classifiers was developed to account for temporal changes of spectral and temperature features of clouds in the visible and infrared.
• The goal of the reading is to understand the application problem, and the consider the adopted methodology to design the neural networks.
• Please read carefully the paper
L4_Enviro4.pdf
(*) linked in the course page.
Fabio Scotti - Manuel Roveri
C17 SC for Environmental Applications and Remote Sensing I M S C I A
End of the lecture
Fabio Scotti - Manuel Roveri