Hyperspectral Sensor Data Capability for Retrieving Complex Urban Land Cover in Comparison with Multispectral Data

City:

Vernice, Italia

Organization:

1. National Research Council, Institute of Atmospheric Pollution, Via Fosso del Cavaliere, Roma, Italy

2. National Research Council, Institute of Methodologies for Environmental Analysis, C.da S. Loja - Zona Industriale, Tito Scalo (PZ), Italy

Project Start Date:

Unkown

Project End Date:

20 May 2018 (Published)

Reference:

Hyperspectral sensor data capability for retrieving complex urban land cover in comparison with multispectral data, Sensors 2008, pp 3299-3320; 

https://doi.org/10.3390/s8053299

Problem:

Evaluation of the added value of the hyper-spectral sensors in mapping a complex historical urban centre.

Technical Solution:

1. An object-oriented approach  and the ISODATA Clustering procedure for imagery segmentation 
2. Spectral Angle Mapper (SAM) supervised classification method for classifying the land cover

Datasets Used:

For this study, both satellite and airborne remote sensing data were used

• MIVIS airborne data acquired on July 26, 2001 at 10:45 (GMT), using scan rates of 25 scans/s at an altitude of 4000m, corresponding to a 8-m ground-pixel resolution at the instruments IFOV 
• Earth Observer (EO-1) sensors satellite data acquired on June 7, 2001 at 11:56 (GMT)
• Landsat ETM+ satellite data acquired on July 2, 2001
• IKONOS satellite data acquired on July 2001

Outcome:

The results of the comparison between hyperspectral and multispectral remote sensing datasets highlights that 

1- Hyperion hypspectral satellite data are capable of mapping the complex urban surface components of the Venice urban land cover with accuracy similar to the higher spatial resolution MIVIS multispectral airborne data

2-Imagery segmentation leads to an appropriate classification only for the three main urban land cover (i.e vegetation, paving and roofing materials) for all the sensors 

Issues that arose:

In a complex urban context, such as that of the Venice study area, it is desirable at the Hyperion 30m/pixel spatial resolution, to decompose pixels into their components as their sizes are smaller than the pixel size.

Not the entire Venice region was captured due to limitations on the ALI

Status:

Terminated

Entered by:

Sept-19-19: Stephen Izuchukwu Oguegbu, oguegbus@mail.utoronto.ca