Five years after the devastatingAsian tsunami of December 2004, researchers and practitioners are still wondering if mangrove forests played any role in saving lives and property. A maj-ority of studies on the question provided evidence that a high-density mangrove forest better attenuates tsunami waves than a low-density forest and highlighted the need for better information on the status and conditions of mangrove forests. Information on the extent, spatial configuration, density, height, and species composition were deemed essential. At the same time, funding agencies, governments, and non-governmental organizations (NGOs) were looking for potential areas to rehab-ilitate mangroves. In this context, USGS/EROS, with support from USGS Director’s Venture Capital Fund, started this project to study the distribution and dynamics of mangrove forests in the tsunami-impacted region of Asia. (See Figure 1.)
The tsunami-impacted region consists of coastal areas of Indonesia, Malaysia, Thailand, Myanmar, Bangladesh, India, and Sri Lanka. The region comprises ~10% of the total mangrove forests of the world, including the largest remaining contiguous tract of mangrove forests, the Sundarbans in Bangladesh and India. Strong demographic pressure and diverse climatic conditions have created a mosaic of mangrove diversity in this region, which is undergoing constant changes. In fact, the Asia Pacific region is the epicenter of mangrove biodiversity and consists of many existing and planned national parks, biosphere reserves, and world heritage sites.
The objective of the USGS study was to determine the rate, causes, and distribution of mangrove forests using multi-temporal Landsat data and field observations. Analysis explored simple research questions such as how much mangrove forest remains, where is it located, what is the spatial and temporal rate of change, and what are the main reasons for the change? More importantly, the study identified potential areas for rehabilitation and regeneration.
We interpreted Landsat MSS, TM, and ETM+ satellite data using hybrid supervised and unsupervised digital image classification techniques. Geometric correction was performed to improve the geo-location to ± ˝ pixels needed for the change analysis. Each image was normalized for variation in solar angle and Earth-sun distance by converting the digital number values to the top of the atmosphere reflectance. Ground truth data and existing maps and databases were used to select training samples and also for iterative labeling. Change analysis was performed using post-classification approach.
Our analyses show that the region lost 12% mangrove forests from 1975 to 2005 to its present extent of ~1,670,000 hectares (ha.). This rate is much less than earlier estimates. As expected, the rate of deforestation varied in both spatial and temporal domains. Burma experienced the highest rate of annual deforestation (1%) and the highest regional rate of deforestation occurred during 1990-2000. Major factors responsible for regional deforestation were agricultural expansion (81%), shrimp farming (12%), urban development (2%), and other factors (5%) (See Figure 2.) This finding is contrary to the widespread belief that shrimp farming is the primary cause of deforestation.
The deforestation rate varied during observation periods. The annual rate of deforestation from 1975-2005 was highest (~1%) in Burma (see Figure 3) compared to Thailand (-0.73%), Indonesia (-0.33%), Malaysia (-0.2%), and Sri Lanka (-0.08%). In contrast, mangrove forests in Bangladesh (+0.14%) and India (+0.04%) remained essentially unchanged or slightly expanded during this period. The increase in mangrove area that we found in India is consistent with reports from the Forest Survey of India, which stated that mangrove forest cover has increased or remained unchanged since 1995. However, almost all the mangrove areas in India are severely degraded, with reduced or negligible vegetation cover. Bangladesh has started ambitious mangrove rehabilitation programs, and mangrove forest areas have also increased by aggradation. The reforestation programs in both India and Bangladesh were initiated by the government and local communities.
At the local level, both deforestation and forest regeneration occurred with varying intensities, with localized hotspots of rapid change. We identified the major deforestation fronts that are located in the Ayeyarwady Delta, and in the Rakhine and Tahinthayi provinces of Burma; Sweetenham and Bagan in Malaysia; Belawan, Pangkalanbrandan, and Langsa in Indonesia; and Southern Krabi and Ranong in Thailand. Major reforestation and afforestation areas are located on the southeastern coast of Bangladesh, and in Pichavaram, Devi Mouth, and Godavari in India.
Our spatio-temporal analysis shows that, despite having the highest population density in the world in its periphery, areal extent of the mangrove forest of the Sundarbans (see Figure 4) has not changed significantly (approximately 1.2%) in the last 25 years. However, the forest is constantly changing due to erosion, aggradation, deforestation and mangrove rehabilitation programs. The net forest area was increased by 1.4% from the 1970s to 1990 and was decreased by 2.5% from 1990 to 2000. The change is insignificant in the context of classification errors and in the dynamic nature of mangrove forests.
The strong commitment of governments under various protection measures such as forest reserves, wildlife sanctuaries, national parks, and international designations, is believed to be responsible for keeping this forest relatively intact (at least in terms of area). This lack of significant loss in terms of area is an excellent example of the co-existence of humans with terrestrial and aquatic plant and animal life. While the measured net loss of mangrove forest is not high, the change matrix shows that turnover was much greater than net change, and the forest is under threat from natural and anthropogenic forces leading to forest degradation, primarily due to top-dying and overexploitation of forest resources.
Lessons learned from this study proved useful enough to expand the study at the global scale. NASA’s Land Cover and Land Use Change Program and the U.N. Environment Programme supported the global study with the following research questions:
How can we use historical and current satellite data and state-of-the-art image processing and geospatial modeling tools and methods to characterize mangrove forest attributes and dynamics more effectively?
What is the present status of mangrove forests of the world and how have the extent and characteristics of mangrove forests changed in the last 15 years?
What are the causes and environmental and socio-economic consequences of mangrove deforestation?
The global study is important because current estimates of mangrove forests of the world range from 110,000 to 240,000 km2. It is believed that the present extent of the forest is less than half of what it once was, and much of what remains is in degraded condition. The forests continue to decline due to conversion to agriculture, aquaculture, tourism, and urban development. About 35% of mangroves have been lost over the last two decades alone, and the forests have been declining at a faster rate than inland tropical forests and coral reefs. Predictions suggest that 30-40% of coastal wetlands and 100% of mangrove forests could be lost in the next 100 years if the present rate of loss continues. As a consequence, important ecosystem goods and services (e.g. natural barrier, carbon sequestration, biodiversity) provided by mangrove forests will be diminished or lost.
Mangrove forests are among the most productive and biologically important ecosystems of the world. The forests help in stabilizing shorelines and in reducing the devastating impact of natural disasters such as tsunamis and hurricanes. They also provide breeding and nursing grounds for marine and pelagic species, and food, medicine, fuel, and building materials for local communities. It is estimated that the forests, including associated soils, can sequester approximately 22.8 million metric tons of carbon each year. Covering only 0.1% of the continent’s surface, the forests account for 11% of the total input of terrestrial carbon into the ocean, and 10% of the terrestrial dissolved organic carbon (DOC) exported to the ocean.
Global Mangroves Forest Map
We have been using recently available Global Land Survey (GLS) Landsat data of 2000 to prepare the first wall-to-wall map of the mangrove forests of the world. The GLS is a global dataset of Landsat 30-m resolution satellite imagery prepared in a partnership between USGS and NASA in support of the U.S. Climate Change Science Program (CCSP), Group on Earth Observations (GEO), and the NASA LCLUC Program. These data are freely available from http://glovis.usgs.gov. Without GLS data, this study would not have been possible, because for a single researcher it is extremely difficult and expensive to process thousands of Landsat data for the tropical and sub-tropical regions of the world where persistent cloud cover is an issue.
We have been interpreting ~1,000 Landsat scenes using a hybrid supervised and unsupervised digital image classification technique. The GLS 2000 data were collected between 1997 and 2000. The multi-satellite, multi-year and multi-seasonal data used in our study are typical of global and continental land use and land cover change studies. It is unrealistic to obtain a global coverage of Landsat data of the same season or year.
Geometric correction was performed to improve the geo-location to Root Mean Square error of ± ˝ pixels needed for subsequent change analysis. Each image was normalized for variation in solar angle and Earth-sun distance by converting the digital number values to the top of the atmosphere reflectance. Ground truth data and existing maps and databases were used to select training samples and also for iterative labeling. Results validation was performed using existing GIS data and published literature.
We have been mapping “true mangroves” defined as trees, shrubs, and palms that exclusively grow in the tidal and inter-tidal zones of the tropical and sub-tropical regions of the world. The minimum mapping unit used in this study will be .08 ha. (See Figure 5.)
Besides preparing a wall-to-wall map of the globe, we are also performing in-depth analysis for some countries, including Madagascar. The mangrove forests of Madagascar are declining, albeit at a much slower rate than the global average. The forests are declining – logging activities, over-exploitation, clear cutting, degradation, and conversion to other land uses all are occurring at the expense of the forests. However, accurate and reliable information on their present distribution, and their rates, causes and consequences of change is not available. Earlier studies used remotely sensed data to map and, in some cases, to monitor mangrove forests at a local scale. Nonetheless, a comprehensive national assessment and synthesis was lacking.
Giri, C.P.; Pengra, B.W.; Zhu, Z.; Singh, A.; and Tieszen, L.L., 2007, “Monitoring Mangrove Forest Dynamics of the Sundarbans in Bangladesh and India Using Multi-temporal Satellite Data from 1973 to 2000”: Estuarine, Coastal and Shelf Science, v. 73, no. 1-2, p. 91-100.
Giri, C.P.; Zhu, Z.; Tieszen, L.L.; Singh, A.; Gillette, S.; and Kelmelis, J.A., 2008, “Mangrove Forest Distributions and Dynamics (1975-2005) of the Tsunami-affected Region of Asia”: Journal of Biogeography, v. 35, no. 3, p. 519-528.
We interpreted time-series Landsat data of 1975, 1990, 2000, and 2005 using a hybrid supervised and unsupervised classification approach. Landsat data were geometrically corrected to an accuracy of +/- 0.5 pixel, an accuracy necessary for change analysis. We used a post-classification change detection approach (see Figure 6).
Our results showed that Madagascar lost 7% of mangrove forests from 1975 to 2005, to a present extent of ~2,797 ha. Deforestation rates and causes varied both spatially and temporally. The forests increased by 5.6% (212 square kilometers) from 1975 to 1990, decreased by 14.3% (455 square kilometers) from 1990 to 2000, and decreased by 2.6% (73 square kilometers) from 2000 to 2005. Similarly, major changes occurred in Bombekota Bay, Mahajamba Bay, Coast of Ambanja, Tsiribihina River, and Cap St. Vincent. Main factors responsible for mangrove deforestation include conversion to agriculture (35%), logging (16%), conversion to aquaculture (3%), and urban development (1%).
Our preliminary findings suggest that the moderate resolution satellite data such as Landsat contain enough detail to capture mangrove forest distribution and dynamics. However, very small patches (<900 square meters) of mangrove forests found along the coast and canals will not be identified from this data. High-resolution satellite data (e.g. IKONOS, QuickBird) or aerial photographs are needed to assess and monitor those areas. However, those very small areas will not make a big difference in the global total. We will incorporate the missing information from available GIS and statistical data. Availability of pre-processed and free data such as GLS is critical to generate important information needed for resource conservation and planning. The methodology developed and the resulting database provide guidance for regional and global efforts to allocate conservation resources, to perform carbon accounting, to implement ecosystem-based management, and to inform mangrove spatial planning, education, and basic research.