169 
NEW APPROACH TO THE ESTIMATION OF SNOWPACK ESTENSION 
VIA REMOTE SENSING 
Alessandra Colombo. Giovanmaria Leoni 
Politecnico di Milano. Dip. U.A.R. sez. Rilevamento 
P.za Leonardo da Vinci 32. 20133 Milano (ITALY) 
Commission VI, Working Group 3 
KEYWORD: Remote sensing, photointerpretation, hard classification, soft classification, accuracy 
ABSTRACT: 
it is very important to be able to foresee the water volume coming from snow melting, infect during winter large mass 
of water is stored in the snowpack on mountain and its melting gives a big contribution to the hydrological balance. 
Prediction of water volume can be done combing information regarding snowpack extension, thickness and density. 
Since 70s remote sensing techniques are successfully applied to measure the snow-covered area in a catchment 
basin. In this work we describe the history of remote sensing contribution to the estimation of snow covered area and 
we propose a new approach. At the beginning photointerpretation techniques were applied in order to identify snow- 
covered area. Nowadays automatic classifications are performed since snow has a typical spectral behaviour; but the 
traditional hard approaches are not able to solve mixture problem and the estimation accuracy is low. So we have 
tested our new approach base on soft classification that solves the mixture problem and describes in right way partial 
snowy conditions too. 
1. INTRODUCTION 
Remote sensing techniques are rapid growing and sc 
their applications are changing and the obtained results 
better. In particular we decribe how snowpack monitoring 
via remote sensing techniques is improving by new 
techniques and new approaches. Snow monitoring 
requires combining satellite Images and point 
observations. Remote sensing information is elaborated 
to evaluate snow covered area and the point information 
are then added to evaluate snowpack water equivalent. 
2. WORK AIM 
The aim of this work is to analyse and describe the 
effects of new approach and new methods to a problem 
solution. We analyse an example case, snow monitoring 
via remote sensing, in order to underline how it is very 
important a continuos updating of working systems in 
order to improve the obtained results. This requires 
carrying out continuously tests in order to verify and 
define adjustments to traditional approach or define new 
work approach 
3. SNOW COVERED AREA EXTIMATiON BY 
PHOTOINTERPRETATION 
accuracy of the obtained result is not known. Following 
picture number 1 shows a b/w photo of Bernina 
Mountain. 
Picture 1: b/w photo of Bernina Mountain 
Next picture 2 shows a zoom into the result of a 
photointerpretation: white colour marks snow covered 
areas. 
First step to forecast snowpack water equivalent requires 
to evaluate snow covered area. 
Till 80s snow covered areas 'were localised by 
photo interpretati on on aeria! photos. In this way human 
experts have to recognise snow-covered areas, to mark 
their borderlines and then empirically evaluate the 
snowpack extension This is a very long work and the