Cavazzini, Armando 
  
HYPERSPECTRAL MIVIS SCANNER DATA INTEGRATED INTO A GIS FOR AN 
INDUSTRIAL AREA 
A. CAVAZZINI (*), E. MICCADEI (**), F. SILVESTRI (***) A. FERRARINI (****) N. ZACCARELLI (5 
(*) CGR (Compagnia Generale Ripreseaeree), Italy 
(**) University of Chieti, Italy 
(****) ISPESL, Italy 
(****) CISIG (Consorzio per l'Innovazione dei Sistemi Informativi Geografici dei grandi bacini fluviali), Italy 
KEY WORDS: MIVIS, hyperspectral scanner, classification, risk assessment, industrial area 
ABSTRACT 
ISPESL is the National Institute for the control and monitoring of risks connected with the presence of industrial plants and 
activities. A project between ISPESL Institute and CISIG (Consortium among National Research Council of Italy, 
University of Parma and Consorzio Compagnie Aereonautiche) has been developed aiming to test the integration of ground 
data on the principal industry in the area of the Pescara river basin and remote sensing data devived from both aerial 
photogrammetry and MIVIS scanner data flights. The general philosophy of the project is to integrate orthophotos (that 
allow a precise georeferentiation of different data sets and the possibility to precisely compare observations made in 
different periods) with data directly observed, in situ measurements of principal parameters that describe the industrial risks 
of the area and data recorded by the hyperspectral MIVIS scanner. The MIVIS scanner, property of National Research 
Council of Italy, is a 102 channels scanner covering visible and near infrared (0.43-0.83 um), middle infrared (1.15-1.55 um 
and 1.98-2.50 um) and thermal infrared (8.21-12.70 um) regions of the electromagnetic spectrum, providing a wealth 
qualitative information of the project area. The association of precisely georeferenced orthophotos, ground data and 
qualitative remote sensing images (MIVIS) is the state of the art technique to build GIS systems using remote sensing 
informations. MIVIS data in the middle and thermal infrared were found to be principally applicable for the study of the 
health status of vegetation, soil conditions and water turbidity. These informations are entered in a GIS for modelling the 
impacts resulting from geologic risks. 
1. INTRODUCTION 
Remote sensing and geographical information systems (GIS) have much to benefit from each other (Wilkinson, 1993). 
Remote sensing offers the potential for the regular provision of large spatial data sets on landscape properties required in a 
wide range of applications. On the other hand, the integration of remotely sensed data with other spatial properties may 
improve the analysis and the extraction of information from multispectral and hyperspectral data. Land cover maps are used 
for many purposes. In environmental engineering projects and in environmental and hydrological studies, accurate and up to 
date information is often required. Knowledge of changes in land cover is becoming increasingly important from both the 
ecological and economic point of view. Land cover information is obtained by classifying remote sensed data and the 
resulting thematic maps can be integrated into a GIS for later use with other data sets. The crucial aspect of this operation is 
the quality of the remotely sensed data. Poor quality data sets are useless for landcover mapping investigations even if 
ancillary data informations are added. 
Starting from 1995 the MIVIS scanner, property of the National Research Council of Italy, started to fly on a CASA 212/c 
aircraft of Compagnia Generale Ripreseaeree, making new data sets available to the scientific community for geological, 
environmental and thermal analyses. This new and very powerful instrument makes possible very detailed land 
classifications to be integrated in GIS systems and in the connected databases. 
The MIVIS scanner is a "whisk broom" scanner and, according it, has a geometrically correct scan line that, due to the 
movement of the aircraft, is displaced with the roll, pitch yaw, and with changes in velocity and direction. The operational 
flight heights of the scanner can range from 1500 meters up to 5000 meters above ground; at this height, the nadiral pixel 
dimension ranges from 3 meters up to 10 meters (the IFOV is 2 mrad wide). Due to these fact, the data acquisition of the 
scanner is affected by a lot of geometrical distortions. The cross-track scan line dimension is geometrically good in the 
central region, being affected laterally by a geometrical distortion; in fact, the large FOV (72°) of the scanner, makes the 
  
236 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 
"| ^ € um "m a ^ (ef 
A M d d ry Ay Say ES