3.1 Mangrove land use changes
For the purpose of detecting the mangrove land use a broad classification (mangrove forest, non-mangrove forest
and water bodies) was first made for each image. This was used to avoid possible problems that might occur if a
detailed classification with many classes was used. The result of the change detection analysis generated from
Landsat MSS 1982, Landsat TM 1994, and fused images (Landsat TM 1994 and ERS-1 1996 radar data) are
summarized in Table 3. The accuracy assessment of these various classifications is shown in Figure 3.
The mangrove forest area was seriously reduced from 1994 to 1996 in comparison with the period of 1982 to 1994.
From Table 3 it may be concluded that the evolution of mangrove areas was as follows: 96,228 ha, 91,827 ha, and
78,799 ha in 1982, 1994 and 1996 respectively. The trends of these changes in mangrove area are indicated in
Figure 4. From these results, it appears that about 13,028 ha (14.2% of the mangrove areas in 1994) disappeared in
the period of 1994 to 1996, whereas only about 4,401 ha (4.6% of the 1982 area) disappeared in period of 1982 to
1994. The total of mangrove areas that disappeared in between 1982 to 1996 is about 17,429 ha (18.1% since 1982).
Therefore, it may be concluded that the mangrove forest area has predominantly deteriorated between period of
1994 to 1996.
Figure 3. Accuracy of the broad classification used for the change detection analysis
Table 3. Delta Mahakam mangrove forest
land cover in 1982 to 1996
| Class Category |
MSS data
in 1982 (ha) |
TM data
in 1994 (ha) |
Fused image
of TM 1994
and ERS-1
1996 (ha) |
*. Mangrove
*. Non mangrove
*. Water bodies |
96,228
16,164
99,398 |
91,827
12,602
107,361 |
78,799
24,751
108,240 |
| Total |
211,790 |
211,790 |
211,790 |
Figure 4. Mangrove land cover changes
between 1982 to 1996
From the field data collection at the study area it was found that most of the deforested areas were converted to fish
and shrimp pond establishment. This evidence was supported by the images classification result in which there are
increasing in water bodies during the period of 1982 to 1996.
4. Conclusions
4.1 Detecting mangrove deforestation
Results of this research show that it is possible to differentiate between mangrove forest and non-mangrove cover
types with all types of optical remotely sensed data. Specifically, it was found that:
- all optical data were able to differentiate fishponds;
- Landsat MSS and Spot XS data were able to differentiate only one agriculture class;
- Landsat TM data was able to differentiate two agriculture classes;
- none of the optical data were able to differentiate oil pipeline establishment.
Using radar data to detect deforestation, it was found that:
- all radar data were able to differentiate fishponds and clear cut areas;
- all radar data were able to identify the establishment of oil pipelines (using both spatial and backscatter information)
- none of the radar data were able to differentiate any agriculture classes
- JERS-1 and Radarsat images were able to differentiate partly cut Nypa palm (fishponds under construction).
Given the accuracy percentage and the number of classes resulted from these classifications, it seems that Landsat
TM, Spot XS and all radar images, but especially JERS-1 and Radarsat all give reasonable results in detecting
deforested areas. There were no significant differences between the accuracy of these five data sets although, as
might be expected, the accuracy of the results from Landsat MSS is soemwhat lower. It is possible that if optical
data are fused with radar data they may even give better results. Further research is therefore required to investigate
whether optical and radar data can complement each other for detecting deforested mangrove areas.
4.2 Monitoring mangrove deforestation
Results of this research show that it is possible to monitor mangrove deforestation with a reasonable accuracy using
optical and radar satellite images.
- during period of 1982 to 1996 about 17,429 ha of mangrove forest disappeared and changed to another land
uses;
- mangroves deforestation in study area is mainly caused by shrimp ponds establishment;
In addition, there are two other human activities that cause deforestation of mangrove in the Makahan delta:
establishing agriculture fields and oil extraction activities. However, the area deforested due to these activities is
relatively small compared to that resulting from the establishment of shrimp ponds.
