SATELLITE REMOTE SENSING FOR SEAGRASS MAPPING IN THE LAGOON OF VENICE
Braga F.
1
, Salviato S.
1
, Alberotanza L.
1
*, Silvestri S.
2
, Ramieri E.
3
, Berton A.
3
, Curiel D.
4
, Rismondo A.
4
1
ISMAR - CNR, Venezia, Italy - federica.braga@ismar.cnr.it, stefania.salviato@ismar.cnr.it, * luigi.alberotanza@ismar.cnr.it
2
MAV - Servizio Informativo, Venezia. Italy - sonia.silvestri@magisacque.it
3
Thetis S.p.A., Venezia. Italy - ramieri.e@thetis.it, berton.a@thetis.it
4
Selc, Marghera (VE). Italy - curiel@selc.it, rismondo@selc.it
Abstract
Seagrass play an important role in preserving lagoon morphology and providing an environment suitable for marine organisms. Seagrass
mapping is therefore particularly relevant for a correct management of the lagoon system. Remote sensing allows the estimation of
seagrass extension and distribution, providing high temporal and spatial resolution maps. Moreover, costs of satellite data are lower than
those currently possible with in situsurveys. In order to investigate the spectral properties of submerged plants, a series of in situradiance
measurements were collected. This spectral library has been used to calibrate a survey system based on the elaboration of remote sensing
data.
Keywords: Seagrass, Venice lagoon, Remote sensing
Rapp. Comm. int. Mer Médit., 37,2004
321
Introduction
In 2002 a research program was started in the framework of a
monitoring project carried out by Consorzio Venezia Nuova, on behalf
of Magistrato alle Acque, Venice. Part of the program consisted in
mapping the distribution of seagrass through remote sensing.
For this purpose, the following activities were planned:
1. definition of a methodology for radiometric data acquisition on field;
2. construction of a spectral library for the different species of seagrass
and macroalgae of the Venice lagoon. This is supposed to be
representative of the spatial and temporal variability of the targets’
spectral behaviour, due to changes in environmental factors as well as in
phenological and growing conditions of plants;
3. calibration and validation of remotely sensed data with collected in
situmeasurements;
4. development of a procedure for mapping seagrass in the Venice
lagoon based on the elaboration of remote sensing images.
Field survey
Experimental activities were carried out during six and five monthly
field campaigns respectively held in spring-summer 2002 and 2003.
During the first year data were measured on six 100 m
2
plots:
• 3 “pure” plots with 100% coverage of Zostera noltii, Zostera marina
and Cymodocea nodosarespectively;
• 1 plot with mixed coverage of Cymodocea nodosa and Zostera marina;
•2 “pure” plots with discontinuous coverage (50%) of Cymodocea
nodosaand Zostera marina.
The three 100% coverage – single species plots - were also sampled
in 2003. In this period, radiance data were collected for complementary
targets (macroalgae, bare substratum with different granulometric
composition, canal) too.
Each monthly campaign lasted six days; i. e. one plot or target per
day. 8 hourly measurements of spectral signatures and ancillary
parameters (turbidity, depth, Chl-a ?uorescence, chemical-physical
parameters, water current intensity and direction, suspended matter,
PAR) were collected in a fixed representative point in order to inves-
tigate the in?uence of the tide effect on the spectral behaviour of targets.
Additional signatures were randomly collected in other points to
enlarge the spectral collection.
Radiance data were measured with a portable radiometer PR-650 and
were taken above the water surface, following an update Sea Viewing
Wide Field of View Sensor (Sea WiFS) protocol (1-3,5).
Monthly measures of phenological and growth data were also
collected in the pure and mixed seagrass plots, in order to investigate
their in?uence on seagrass spectral response.
Satellite data analyses
Satellite data were collected and used to derive distribution maps of
the different seagrass species living in the shallow waters of the Venice
lagoon. Four Landsat 7 ETM+ scenes were acquired during the 2002
field campaigns, and in particular the 18
th
May scene was used for the
analyses. Data were radiometrically calibrated, atmospherically
corrected - using ATCOR2 based on MODTRAN 4.2 - and, finally,
georeferenced. Only the VIS and NIR bands were used for the analyses.
The analysis of spectral signatures of un-vegetated lagoon ?oor
allowed calculating an exponential relation between re?ectance and
water depth. The re?ectance of an imaginary lagoon without submerged
vegetation was simulated using the exponential relation in the green
portion of the electromagnetic spectrum and considering the water
heights calculated by an hydrodynamic model. Difference between
simulated re?ectances and those obtained with Landsat ETM+ enabled
to distinguish between vegetated and bare lagoon.
An ISODATA clustering was then applied on vegetated areas, using
the re?ectance standard deviations calculated on spectra collected in
field (4). Matching cluster spectra with those obtained in the field,
clusters were grouped into four vegetation classes. A good agreement
between these results and a seagrass - macroalgae map obtained from
detailed surveys undertaken on the whole lagoon during spring –
summer 2002 was noticed.
Distinction between Zostera marinaand Ulva rigidaappeared to be
complex, probably due both to the similarity of their spectral responses
and to the fact that they usually grow on the same areas. Zostera noltii
was not found, probably because it was at the beginning of its growing
season. Cymodocea nodosawas well recognized in the central lagoon,
while some disagreement with field survey was found in limited areas of
the southern basin (Valle Millecampi and Chioggia).
Thanks to the 2003 field campaigns, a larger spectral signatures
database will be available for the recognition of vegetation species, and
hence a better distinction among species will be possible.
Results
Experimental activities enabled to collect a considerable quantity of
spectral data of three seagrass species and complementary targets in the
lagoon of Venice, in different environmental situations as well as
phenological and growing conditions of plants. This led to create an
original and rich spectral database.
The preliminary study of re?ectance data pointed out the close
correlation between spectral measurements of different targets,
underlining the relevant in?uence that some environmental factors have
on spectral responses. In fact, this can partially or totally mask the
re?ectance of submerged vegetation. Factors that seem to have the
highest in?uence are:
• depth and thickness of the water column above the submerged
vegetation;
• water turbidity;
• bottom characteristics.
Preliminary elaboration of satellite data confirmed that the created
spectral collection represents a considerable experimental basis for the
development and calibration of a remote sensing system for seagrass
mapping based on satellite imagery.
Acknowledgment.A special acknowledgment to Consorzio Venezia
Nuova, Servizio Ambiente, in particular to Arch. A. G. Bernstein and
Ing. L. Montobbio.
References
1-Hooker S.B., Lazin G., Zibordi G., McLean S., 2002. An evaluation of
above- and in-water methods for determining water-leaving radiances.
Journal of Atmospheric and Oceanic Technology, 19(4): 486–515.
2-Mobley Curtis D., 1999. Estimation of the remote-sensing re?ectance
from above-surface measurements. Applied Optics, 38(36): 7442-7455.
3-Mueller J.L., Austin R.W., 1995. Ocean Optics Protocols for Sea WiFS
Validation, Revision 1. NASA Tech. Memo. 104566, Vol. 25, S.B. Hooker,
E.R. Firestone, J.G. Acker (eds.); NASA Goddard Space Flight Center,
Greenbelt, Maryland, 67 p.
4-Silvestri, S., M. Marani, J. Settle, F. Benvenuto, A. Marani, 2002. Salt
Marsh Vegetation Radiometry: Data Analysis and Scaling, Remote Sensing
of Environment, 80: 473-482.
5-Zibordi G., Hooker S.B., Berthon J.F., D’Alimonte D., 2002.
Autonomous Above-Water Radiance Measurements from an Offshore
Platform: A Field Assessment Experiment. Journal of Atmospheric and
Oceanic Technology, 19(5): 808-819.
