355
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986
Remote sensing in the evaluation of natural resources:
Forestry in Italy
Eraldo Amadesi & Rodolfo Zecchi
Geography Department, Bologna University, Italy
Stefano Bizzi & Roberto Medri
Consortium CISET-Advanced Computer Systems, Rome, Italy
Gilmo Vianello
CSSAS-Experimental Centre of Land Research and Analysis, Bologna University, Italy
ABSTRACT: Presented here is a methodology adopted for the process recognition and the surface evaluation of
this type of forest in Italy. This methodology is based on satellite imageries and aerial photos, previous the
construction of a "ground point net", the definition of the "interpretation keys" and the "spectral response
patterns" related to the forestry legend classes.
RESUME: On présente une méthodologie d'étude pour le levé et l'évaluation de la surface de forêt et des bois en
Italie. La méthodologie développée, se basant sur l'etude des images et l'analyse des spectrales signatures de
satellite Landsat TM comparées avec 1'interpretation photographique et les contrôles sur le terrain, amène à la
classification et à la delimitation de forêt et des bois.
NTROOUCT10N
The necessity of constructing, in a short period of time, a map showing the
wooded or forest areas of the entire country on a scale of 1:100,000 was met
using the methods of photointerpretation and teledetection.
The determination of the legend for this map was done based on the
information available from the interpretation of aerial photographs and analysis
of images obtained via satellite (Table 1). For this purpose, aerial photographs o
Italian territory taken after 1982 were used.
Since recent aerial photographs were not always available, it was felt
worthwhile, for some Regions, to undertake experiments involving the
interpretation of suitably elaborated satellite produced images, in particular
those coming from Landsat-5 "Thematic Mapper" sensor. The experimental work
has shown that, if suitably processed, the satellite images can give satisfactory
results for the mapping of wooded or forest areas. The procedure followed for the
study is reported, schematically, in Table 2.
METHXCLCGY
l phase: calibration of the inteoretative keys
Once the area under study had been delineated on territorial map of the
"Istituto Geografico Militare" (I.G.M.) on a scale of 1:100,000, "sample windows'
were then established, reasonably distributed throughout the area under study
such that all the different situations occurring in that area were taken into
consideration.
This initial verification is possible, obviously, by consulting pre-existing
aerial photographs and thematic maps.
When this initial analysis was completed, the next step was to proceed
with extensive photointerpretation of each of the “sample windows", using more
recent aerial photographs.
The experiments carried out allowed verification of how a correct analysis
of the satellite images from Landsat 5 (resolution at ground level around 30 x 3C
meters) requires prior calibration via photointerpretation of sufficiently
extensive, non puntiform areas or of individual pixels.
Optimum results using these calibrations are obtained with "windows" on
the earth in a double photographic "models" on a scale of around 1:30,000 and
corresponding approximately to a surface area of 48 Km 2 .
This new method of study and territorial analysis differs substantially
from preeciding methods based more or less on puntiform calibrations or
calibrations limited to small areas.
In our case, where it was necessary to locate and homogenize a large
number of wooded and forest formations covering the entire national territory
from north to south, including the islands, with considerable differences in
latitude, elevation and climate, only an extensive calibration such as that
proposed could give qualitatively valid results.
The size of the “windows", therefore, was conditioned by and related to the
necessity of studying an area large enough to include the greatest possible
number of wooded or forest formations, taking into consideration the time of day
of satellite images acquisition over Italian territory (9:30 a.m. GMT), the time ol
year in which the images were taken and the specific local environmental
conditions (elevation, exposition, slope of the hills, etc.).
After the photointerpretation work, specific field studies must be made,
with the two-fold objective of controlling the correspondence on the ground ol
the interpretations made and determining precise reference points for successive
elaboration. The control on the ground, for sample areas, satisfies the
requirement of testing the accuracy of the maps drawn.
FORMATION FORMAZIONE
GCVEFNAMBTTGCVERNO
CADUCEOUS (prevalent beech)
CADUCIFOGLIE (a prevalenza di faggio)
mature wood
fustaia
coupse
ceduo
CADUCEOUS TREES (prevalent oak,
chestnut, hornbeam)
CADUCIFOGLIE (a prevalenza quercie,
castagno, cerpino)
mature wood
fustaia
coupse
ceduo
EVERGREEN MEDITERRANEE (ilex, cork-oak)
LATIFOGLIE SEMPREVERDI MEDITERRANEE
(leccio, sughero)
mature wood
fustaia
coupse
ceduo
CONIFERS CONIFERE
REAFFORESTATION
RMBOSCH1MENT1E POPOLAMENTI
ARTIFICIALI
BUSHY LAND I f
CESPUGLIETI, ARBUSTETI, MACCHIE 1
WOOD PLANTATIONS (popoiar-wood,
eucalyptus-wood)
PIANTAGIONI ARBOREE DA LEGNO | P
(pioppeti, eucalipteti)
Mixt wood formations are indicated with the double marks: the first
one is referred to the prevalent wood formation.
Le formazioni miste vengono indicate con la doppia siglatura:
la prima lettera indica la formazione prevalente.
TABLE 1 - Legend of the forestry map (Scale 1:100,000)
