A Geospatial Assessment of the Destruction of the Mangrove Forests of Madagascar and Burma

Evan Johnson: University of California, Los Angeles
Justin Manduke: University of California, Los Angeles

Abstract

Over the course of the past decade, society has become increasingly aware of global ecological harm and habitat destruction. A popular environmental issue is the case of the rapidly declining acreage of mangrove forests. Mangrove forests typically are found along the coastlines of tropical regions and serve as a habitat for hundreds of terrestrial and aquatic ecosystems. However, global mangrove ecosystems are disappearing at an alarming rate. Many researchers attribute this to a series of events that include global climate change, increased shrimp aquacultures, and general anthropogenic processes. This paper will provide a geospatial assessment of mangrove forest health in two unique regions of the world. It has been chronicled in scholarly papers that the mangroves of Southeast Asia and Western Africa are especially endangered. The Mahajamba delta of Madagascar and the deltas of the Ayeyarwady region of Myanmar will be the primary study areas. This study will use Landsat 7 to generate a series of unsupervised classifications in each of these regions. From these classifications, it is possible to calculate and quantify the extent of the mangrove decline over the past decade, as well as which region is in greater threat.

Introduction

Many of the world’s ecosystems remain under serious threat of irreversible reduction, one of which is the mangrove forests of Africa and Southeast Asia. Mangrove forests are classified by a variety of evergreen trees thriving in the brackish coastal waters found in tropical and subtropical regions. (“Mangal…”, par. 1-3). The characteristically exposed roots of the vegetation in these forests become inundated by the salty ocean water daily, necessary for the survival of the trees which provide shelter to many species of dugongs, mollusks, and fish (Tognetti, 2001). Mangroves also shelter the soils of coastal regions from imminent threats of tidal erosion by supporting the sediment with extensive systems of roots (Oo, 2002). The tides, however, are also vital the well-being of a variety of mangroves, for the swells deposit ocean silt onto the roots, providing the mangroves with adequate nutrients. These forests, which are so vital to the livelihood of a great range of biodiversity, find themselves on the verge of mass destruction in many areas of the tropics at the hands of socioeconomic industry, namely shrimp aquaculture. Although this practice oftentimes proves economically viable for the parties in operation, conversion of the mangroves for purposes of raising shrimp has led to the compromising of various coastal communities, social conflict, coastal erosion, and the loss of both floral and faunal biodiversity (Primavera 1997). Shrimp aquaculture has expanded rapidly in the tropics since the middle of the twentieth century, and the effects of such endeavors have been seen very notably in the deltas of Burma and the northwestern bays of Madagascar. Global civil society has thankfully taken notice of these operations, however damage has already been done. This study aims to comparatively quantify the destruction of coastal mangroves at two sites in Madagascar and Burma in order to assess the severity of current vegetation clearance.

Methodology

Study Regions

For the purposes of this study, mangrove forests in two different geographical regions were studies. The two study regions were the Mahajamba delta of northern Madagascar and the deltas of the Ayeyarwady region of Myanmar. The following video gives a brief introduction to the Mahajamba delta and its mangroves:



While mangrove forests are threatened worldwide, these two geographical regions have been especially hard-hit. These two regions in particular have extensive fishery and shrimp aquaculture activity, thus leading to a hypothesis that the mangrove forests of Madagascar and Myanmar are more severely threatened than any others in the world.




The latitudinal and longitudinal coordinates of the Myanmar/Burma study site are 15 deg, 37', 34" N, 95 deg, 35', 27" E. The latitudinal and longitudinal coordinates of the Madagascar are 15 degrees, 17’, 32.48”S, 46 degrees, 50’, 47.23” E.

Data Collection and Analysis

Landsat 7 was the only satellite sensor used for this research. Images were gathered during the winter months of 1999 (Madagascar), 2000 (Burma), and 2010 (both images). Winter months consist of January, February, or March for the purposes of this study. Landsat images were downloaded from the United States Geological Survey (USGS) Global Visualization Viewer (Glovis). In total, four Landsat 7 images were downloaded. Bands 1 through 5 were needed for accurate analysis.

All four images were imported to the ENVI 4.7 software suite for remote sensing manipulation. To systematically process each image, a standard and repeatable algorithm was created. Beginning in 2003, Landsat 7 began to experience problems associated with data loss. Thus, the first step is to replace the bad values in the image using ENVI. Once the image has been repaired, it is possible to align the Landsat bands in an order that highlights the mangrove forests. It was determined that mangrove forests are best highlighted using a 4,5,3 band combination (Sinking). Under this combination, mangrove forests turn a unique shade of red, separating themselves from all other vegetation in the region. A step-by-step visualization of this algorithm is displayed below.

In order to assess the ability of Landsat 7 to detect mangrove forests, an unsupervised classification was conducted. Land and vegetation classes were narrowed to three different categories – mangrove forests, water, and all other land and vegetation. Following the pre-processing steps described previously, an unsupervised classification was generated for each of these four satellite images. To achieve greater detail and remove error, the land classes were separated into twenty separate classes.

The unsupervised classification of each image is used to determine the amount of mangrove forest area coverage. Once vegetation is divided into classes, it is possible to determine the percent and total area coverage of each vegetation class. Once the area of mangrove coverage is determined for each image, it is also possible to compare before and after images and determine whether mangrove forests have experienced increased destruction over the past ten years.

Visual walk-through of methodology:



Figure 1: Layering the 3 visible bands (1, 2, 3) of the Landsat 7 ETM+ sensor in a specific fashion (Red-3, Green-2, Blue-1) produces this full color image of the 2010 Burma site.



Figure 2: Layering the red, near-infrared, and mid-infrared bands (3, 4, 5) of the Landsat 7 ETM+ sensor in a specific fashion (Red-4, Green-5, Blue-3) produces this image of the 2010 Burma site that allows for the easy viewing of mangrove vegetation (bright orange).



Figure 3: This subset of the full Burma Landsat image will be the geographic location for which the quantifiable mangrove decrease will be calculated.



Figure 4: This image differs from the previous step of the processing algorithm insomuch as the scan line gaps from the faulty ETM+ sensor have been corrected through data extrapolation. This image is not entirely accurate, however the estimation helps in the imagery analysis.



Figure 5: The corrected subset image undergoes an unsupervised Isodata classification in order to isolate the different land cover types (wavelengths) represented in the image.



Figure 6: Various classes merge together in order to isolate 3 main classes of land cover information: land (black), mangroves (red), and water (blue). The “class statistics” of this image now can quantify the area and coverage percentage of the mangroves in the Burma delta in comparison to the 2000 image.

It is important to note that images of each study region were subset identically to accurately calculate acreage and detect change.

Results

Displayed below are the final unsupervised classification images from each study region . Red colors represent mangrove forest, blue colors represent water, and black colors represent land and all other vegetation.

Unsupervised 1999 - Madagascar


Unsupervised 2010 - Madagascar


Unsupervised 2000 - Burma


Unsupervised 2010 - Burma


As displayed in these classifications, mangrove forestation has declined precipitously over this ten year period. Using these classifications, it is possible to calculate and quantify the extent of this decline.



In January of 2000, mangrove forests covered 772,936.2 square kilometers, or 16.312% of the Myanmar study region. By 2010, this value decreased to 455,506.2 square kilometers, or 9.613% of the study region. This represents a 41.068% decrease in mangrove forests over the ten year period.

Similar results exist for the Madagascar site. In January of 2000, mangrove forests covered 517,886.1 square kilometers, or 18.783% of the study area. By 2010, this value had decreased to 429,032.7 square kilometers, or 15.579% of the region. Overall, this represents a 17.157% decrease in mangrove forests over the ten year period.

A bar graph displaying this data is displayed below.



The gross decrease in mangrove area was significant in both regions (see table for raw data). Overall, however, the Myanmar study area experienced much greater mangrove forest destruction proportionally than did the Madagascar study area.

Discussion

Our findings have quantified the extent to which mangrove mangrove forests have been converted in our Madagascar and Burma study sites. Shrimp aquaculture has driven industry to clear the vegetation to make room for shrimp farms. Over 17% of Madagascar mangroves and over 40% of Burma mangroves in the study were lost. These numbers are disturbingly large and represent a morbid trend.

Some limitations and assumptions must be taken into account when assessing the results of our study. For example, the more recent Landsat 7 images had data gaps due to a broken sensor, so the information was extrapolated from surrounding areas. This may have slightly compromised the accuracy of our data. Also, even though all images were taken from winter months, climatic variability in vegetation area cannot be perfectly accounted for without data pertaining to temperature, rainfall, and other physical elements. In fact, some high resolution Landsat images that would have worked well with our study, however they were unable to be used due to weather conditions (i.e. clouds). It is also important to note that a major hurricane occurred in the Burma study region during May of 2008. Inevitably, many mangrove forests were lost during this natural disaster. Thus, cause for mangrove loss is not completely an anthropogenic issue.

Research has been conducting on our study sites before, many with the SPOT and SIR-C systems. Mahajamba bay has undergone much observation and classification previously, however little quantitative data has been published. In these cases, decrease in mangrove area has also been attributed to shrimp aquaculture in addition to land conversion for housing (Blasco, 1998 and Pasqualini, 1999).

Given more time and extensive resources, several other methods should be considered. For in-depth analysis, a field study is necessary to accurately obtain ground truth data. This ground truth data could then be used to run a supervised classification that would deliver more accurate results.

Several other sensors should also be used. NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) should be used to combine Land-Surface Temperature (LST) imagery with Enhanced Vegetation Index (EVI) imagery to generate a Disturbance Index. Future studies should also include an extensive Normalized Difference Vegetation Index (NDVI) based upon Landsat imagery. Such a study would analyze random points within an NDVI.

Overall, it becomes clear from these results that mangrove forests in these two ecologically important regions are disappearing incredibly rapidly. If mangrove destruction continues at this current rate, these vital forest ecosystems may soon cease to exist.

References

Blasco, F. "Recent Advances in Mangrove Studies Using Remote Sensing Data." Marine and Freshwater Research 49 (1998): 287-96. Print.

"Mangal (Mangrove)." Mildred E. Mathias Botanical Gardens. UCLA Botany Department. Web. 23 Nov. 2010.

Oo, Nay Win. "Present State and Problems of Mangrove Management in Myanmar." Trees. 16 (2002): 218-23. Print.

Pasqualini, Vanina. "Mangrove Mapping in North-Western Madagascar Using SPOT-XS and SIR-C Radar Data." Hydrobiologia 413 (1999): 127-33. Print.

Primavera, J.H. "Socio-economic Impacts of Shrimp Culture." Aquaculture Research. 28.10 (1997): 815-27. Print.

Sinking, Ramphing, and Thanakorn Sanguantrakool. “Monitoring of Mangroves in Trang Province, South of Thailand by Using Multi-temporal LANDSAT-5 TM and SPOT-5 Data." Satellite Based Resources Information Center. Web.

Tognetti, Sylvia. "Madagascar Mangroves (AT1404)." WWF Ecoregions. World Wildlife Fund, 2001. Web. 23 Nov. 2010.