| dc.description.abstract | The coral reef in the Coral Triangle Asia has been directly threatened by human activities and natural threats. The coral reef destructions have also happened in Spermonde Archipelago especially Barrang Lompo Island (COREMAP II 2010). Barrang Lompo Island contributes highly to the society, the most livelihoods of which depends on its shallow water. Therefore, spatial dynamic mapping and spatial prediction model for future trend of coral reef are needed to create good spatial planning for coastal area using remote sensing data and M-CA Model. The objective of this research was to analyze dynamics of coral reef by using remote sensing data and also to predict the future trend of coral reef change. The research was conducted from November 2012 to December 2013 at the shallow water of Barrang Lompo Island. There are several data used in this study. Landsat TM / ETM+ of 1993, 1997, 2002, 2007, and 2012 were processed using image classification for coral reef map. Probability maps were derived from multiplication maps of salinity, sea surface temperature (SST), Dissolve Oxygen (DO), pH, total suspended sediment (TSS), water clearness, bathymetry, and the constraint map used were land area and deep water in Barrang lompo with the assumption that the area will not change into coral reef for 10 years. This research has been conducted in several steps: a) ground truth, b) image processing, c) Markov chain analysis, d) Multi criteria evaluation (MCE) analysis, e) Cellular Automata (CA) analysis, f) validation and, g) coral reef change prediction. The main measurement data was derived from ground truth. Ground truth data from the field were used for image processing. They were applied on the required habitat types of the whole study area by using Global Position System (GPS). Image processing was used to classify coral reef at different times. The classified images were used as input for Markov chain analysis in order to get transition area matrix. MCE analysis was used to create a probabilities map using several parameters. The results of image processing, Markov chain analysis and MCE were used as inputs to perform the CA model. The final result of this model was the coral reef change prediction. The result of simulation from CA process was validated using the actual coral reef map obtained from Landsat image to evaluate the prediction result. If the result validation standard agreement is >75%, then it can be used to predict the future trend coral reef for the next 10 years. Based on the confusion matrix, the result classification in 2012 showed the overall accuracy was about 80.24%. The result of analysis dynamic that occurred during 1993 to 2012 showed live coral and seagrass area decreased approximately by 11.21 ha (12.45%) and 3.57 ha (8.02%) while dead coral increased up to 5.05 ha (86.18%), rubble increased 5.14 ha (68.72%), and sand increased 4.59 ha (35.23%). Live coral and seagrass areas decreased from 1993 to 2012, and the greatest decrease occurred from 2002 to 2007 covering the area of 4.51 ha and 1.47 ha, respectively. The second greatest decreases for coral and seagrass was 4.15 ha and 1.46 ha from 2007 to 2012, respectively. While dead coral, rubble, and sand has been from 1993 to 2012. Based on observations in the field, the coral reef habitat destruction in Barrang Lompo Island was largely caused by human activities. It was proven by many fragments of rock (rubble), and the activities by explosives fishing such as bombs and chemicals in shallow waters Barrang Lompo (COREMAP II 2010). According to Yusuf and Jompa (2012), the phenomenon of coral bleaching that occurred at the end of 2009 until middle of 2010 was caused by La Nina phenomena increasing the SST by the westward warmpool movement from the Central Pacific to the Indonesian Seas, due to the strengthening of trade winds that significantly decreased the quality of coral reefs in the Spermonde archipelago. The simulation result of coral reef change used the number 5 times as the total iterations, and filtering type 5x5 cell contiguity filtering was used to predict in 2002, 2007 and 2012, while to predict 2022 the number 10 iterations was used. Before simulation model to predict in 2022, the simulation result of coral reef change of 2002, 2007, and 2012 then compared with classification result of coral reef obtained from Landsat images. This study used Kappa Index Agreement (KIA) and spatial distribution was indicated by Post-Classification to see the accuracy of model. Calculation results obtained were 2002 (89.41%), 2007 (88.86%), 2012 (87.71%) overall kappa. Although the results of the model M-CA every year was over-estimated, the value of validation was good enough to predict the future trend of coral reef of 2022 because the achievement was higher than the standard value of 75% (Montserud et al. 1992 in Wassahua 2010). Based on the simulation model trend of coral reef change from 2012 to 2022, the total area of coral reef was predicted to decrease from 78.86 ha to 63.55 ha. In this case, coral reef had a higher change rate and if it always occurs the coral reef area will disappear or be broken in Barrang Lompo Island. On the other hand, dead coral and rubble were predicted to increase from 5.86 ha to 15.83 ha and 7.48 ha to 13.62 ha, respectively. From the fore going discussion, the conclusion can be described as follows. The live coral condition decreased from 1993 to 2012. The greatest decreases occurred from 2002 to 2007 covering about 4.51 ha and second greatest decreases was from 2007 to 2012 about 4.15 ha. Relatively live coral areas tend to become dead coral and rubber. The result model M-CA showed the condition of live coral area from 2012 to 2022 was predicted to decrease about 15.31 ha (19.41%), and dead coral and rubble predicted to increase from 5.86 ha to 15.83 ha and 7.48 ha to 13.62 ha, respectively. Furthermore, the recommendation from this research is that SST parameters was used as a scenario, the result of this spatial model M-CA can be used to predict the future trend of coral reef change. However, this model has not used human activity parameter as input. Hence, future research on predicting the future trend of coral reef should use a model that considers the human activities as input. | en |
