POST-TSUNAMI ASSESSMENT OF MANGROVE ECOSYSTEMS IN PHRA THONG ISLAND, THAILAND,
USING REMOTE SENSING, FIELD SURVEYS, AND GIS
Project duration (in months)* 6
Countries covered by the project*
Project Objectives (300 words max) *
Intact mangrove forests can provide important ecosystem services, such as reducing coastal erosion
and protecting low-lying areas from flooding. On 26 December 2004, an earthquake in the Indian
Ocean triggered massive waves that caused vast destruction in many coastal areas in the region.
A disturbance of this magnitude presents an important opportunity to evaluate the role of
mangroves in protecting shorelines, to determine whether mangroves were impacted differently
than other habitats, and to quantify the loss of mangroves resulting from the tsunami. We will
utilize high-resolution satellite imagery such as SPOT 5 multispectral data to detect and quantify the
impact of the disturbance on Koh Phra Thong, a 100 km2 island in the Andaman Sea in Thailand.
We will apply and test change-detection methodologies to assess tsunami impacts on land cover and
vegetation typologies. Our objectives are 1) to determine the location and extent of tsunami-related
damage to the island’s ecosystems, both immediate (one week after the event) and secondary
(several months later), 2) to distinguish between primary (mechanical) and secondary (physico-
chemical) effects of the surge on vegetation. Our cartographic products can be used by NGOs to
guide ecosystem management.
Many NGOs, including the United Nations Development Programme, are funding and
implementing mangrove restoration in tsunami-impacted areas (Stone, 2006). These actions require
knowledge of the location and condition of damaged ecosystems. Our remote sensing approach can
provide an objective analysis, with quantitative and spatially explicit results that can help to
prioritize and optimize restoration efforts. We will focus our efforts on Koh Phra Thong to support
ongoing restoration activities on the island, including those carried out by Naucrates
(http://www.naucrates.org/), a group that has been working locally for 10 years.
Project Methodology (650 words max) *
Mangrove ecosystem analysis is commonly performed through remote sensing and GIS techniques.
Many published articles propose methods based on intensive visual interpretation to map vegetation
class boundaries using aerial photos and satellite images. Simple GIS overlay is frequently used to
quantify changes among the identified classes. Automatic classification is also used to produce
digital maps both using unsupervised and supervised algorithms. Many studies concern land use
land cover (LULC) changes that occurred in the last 20 years. For this reason, different data from
aerial photographs to satellite images are used. Those data do not allow automatic analysis of
spectral signature changes and require extensive local knowledge. LULC changes can be assessed
systematically and with less bias with automatic analyses when 1) satellite imagery of the same
sensor are available, and 2) some capacity for ground truthing exists.
This project intends to determine the impact of a rapid strong event like the December 2004 tsunami
on coastal vegetation ecosystems on Koh Phra Thong (Thailand). We expect to find two kinds of
effects: 1) immediate, severe damage to trees due to mechanical effect of tsunami waves; 2)
progressive degradation of standing biomass due to secondary effects including sedimentary
deposition and elevated salinity. For these reasons we intend to perform two kinds of analysis.
The island ecosystems have been monitored by the Naucrates since late 1990s. In 2002 an intensive
field campaign was begun with the aim of producing a vegetation geo-database. Preliminary
analyses were conducted using available cartography. An island bathymetric map (~ 1970) was
digitized and georeferenced to represent island landforms in a GIS. Moreover we obtained a cloud-
free Landsat-7 ETM+ image from 15 January 2002. We surveyed most of the island by jeep, by
boat, and by foot during 14 excursions to obtain a characterization of the island’s vegetation zones.
We obtained GPS coordinates, photographs, and observations including vegetation, landform
features, and soils. Other cartographic maps recently produced by SEUB (www.seub.or.th) where
For the first objective (immediate change) we will apply change detection techniques to two
images: one acquired a few weeks before the tsunami (t=0), and the second a few days after (t=1).
Two methodologies have to be tested: 1) change detection matrix based on automatic classification,
and 2) a pixel by pixel spectral comparison. These different methodological approaches will
integrate uni-temporal post tsunami imagery analysis and bi-temporal (pre/post event) change
Pre-processing methods regarding georefencing and topographic and atmospheric correction will be
applied as needed. Automatic supervised classification, using the geo-databse of ecosystems ground
truth, will be performed to produce pre impact and post impact land cover maps. This map will be
analysed in a GIS environment to produce the change detection matrix. Normalized difference
vegetation index (NDVI) and wide dynamic range vegetation index (WDRVI) (Gitelson, 2004) will
be calculated to analyse VIs changes after the impact. This analysis will emphasize area where
vegetation cover has been lost but also areas where vegetation is likely to be damaged by the event.
The second objective will be achieved by analysing the vegetation index (VI) signal of surviving
vegetation patches. At least two images will be analyzed within the successive twelve months. A
segmentation procedure based a post-tsunami land cover map will allow us to analyse ecologically
homogeneous areas. Pixel by pixel changes in the VI signal will be weighted on the average
behaviour of the initial class/patch to highlight areas were it is highly probable that a change has
been occurring. This time-series analysis intends to identify areas that are further degraded or in
recovery. GIS analysis will allow us to generate hypotheses regarding processes, such as salinity,
sedimentation, or soil erosion, that are likely to have a delayed impact.
Validation of the change detection analysis will be performed by NGO members during 2007 winter
field campaign. Products of this study will be provided to local authorities and NGO working on
ecosystem restoration on the island.
Experience (300 words max)*
Mirco BOSCHETTI received his PhD from the Agronomy Faculty of the State University of Milan
(Italy), in 2006. The research was conducted with the aim of studying the contribution of remote
sensing data, acquired in field with spectro-radiometer or provided by satellite sensor, to the
monitoring of cropping systems for analyse the crop phenology and to assess crop production.
In 1998 he was graduated from the State University of Milano, major in Environmental Science,
acquiring the know-how on the methodology for the monitoring and analysis of environmental
compounds and their interactions. Specialisation degree in Environmental Resources Analysis and
In the last 8 years he worked as research collaborator at the IREA-CNR in particular he was
involved in the:
- GEOLAND project (http://www.gmes-geoland.info/) of the 6th Framework Programme of the
European Commission. The research focuses on global environmental protection and global
change issues. A pre-operation method to perform continental scale assessment of Africa
ecosystems was developed integration EO data in GIS environmental (2004 - present);
- GLASNOWMAP project (http://subzero.irea.cnr.it/glasnowmap/) founded by ESA joining the
summer field campaigns on alpine glacier addressed to measure spectral optical properties of
different snow and ice surfaces (2002-2003).
- DARFEM experiment (http://milano.irea.cnr.it/hysens/hysens.htm) of HySens project (DLR-
Germany). Research was designed to retrieve biophysical vegetation parameters by
hyperspectral-directional observations. (2001-2002);
He has good knowledge in remote sensing data (airborne and satellite) processing and in GIS and
extensive experience in field measurements and organization of field campaign. He regularly gives
lectures in courses at university level on Cartography, Remote Sensing and GIS. He is also involved
in scientific divulgation at different level and interested in collaborating with NGO on
environmental issues. In particular in 2002 he joined NAUCRATES NGO in Thailand to produce a
digital cartography of Phra Thong island ecosystems by field survey and GIS mapping.
Remote sensing and GIS affords the opportunity to analyze landscape data in many ecologically
relevant and efficient ways. Our study is similar to others that have successfully used the
combination of remote sensing and field surveys to quantify changes in mangrove plant cover over
time. We expect to be able to differentiate between different types of mangrove communities.
Following our analyses, we will report the percent of mangroves that were lost during the tsunami,
whether certain mangrove community associations or geographic locations were affected more than
others, any differential sensitivity between mangroves and other ecosystems to the tsunami-related
flooding. With our results, we may also be able to detect areas that may be targets for future
Sources of financial funding*
Other Project Discipline/Cathegory
Stone Richard (2006), A Rescue Effort for Tsunami-Ravaged
Mangrove Forests, Science VOL 314 pp. 404, 20 OCTOBER 2006
Gitelson A.A. (2004) –“Wide Dynamic Range Vegetation Index for Remote Quantification of
Biophysical Characteristics of Vegetation” J. Plant Physiol. 161. 165–173.