Fish Abundance Estimation in North of Central Java Sea by Means of Acoustic and Swept Area Methods: A Geostatistical Approach

Abstract

This research was intended to apply geostatistical approach on fish abundance estimation in the north of Central Java waters. Analysis of trawl and acoustic data structure was performed using geostatistical analysis, estimation results of the experimental semi-variogram were then used to infer the characteristics of fish abundance in the north of Java waters. Swept area data obtained using bottom trawl while acoustic data obtained using SIMRAD EK60 with operating frequency at 120 KHz. Data processing includes the standardization of catch, integration of acoustic data extraction, transformation of geographic position in to UTM (Universal Transverse Mercator) format, variogram fitting model, abundance estimation using the model, and fish distribution mapping by kriging. Results of structural analysis and models fitting using geostatistical analysis showed that the most suitable model with all data used were spherical model (spherical) with different parameters and R2 values from different models, models with the highest R2 value 0.940 was obtained from 2005 acoustic data, while the model with lowest R2 value 0.559 was obtained on 2006 swept area data set. The models were then used to estimate the value of fish abundance and to obtain information on the points with no abundance of information through kriging process. Validity of a model in estimating and interpolation can be known through calculation of R2 values from crossvalidation process between the estimated value with the actual value, the higher the value of R2 from cross-validation process the better the model is doing prediction. From the calculations of cross-validation, R2 values obtained varied for each data set. The highest cross-validation R2 values obtained from 2005 acoustic data set that is equal to 0.911, while the lowest R2 value 0.316 was found in 2006 swept area data set. Differences in the estimated level of validity from each set data can be used in consideration in the implementation of data collection on fisheries surveys, especially to get a better result. Next step of the geostatistical application is estimating and mapping of fishery resources based on a model that has been obtained, so it can provide a snapshot of fish abundance and distribution condition in the north of Java waters. From geostatistical analysis abundance estimation results in this study can be seen that the abundance of pelagic fish resources in 2005 has a higher value compared with the year 2006, this indicates a decline in the abundance of pelagic fish resources, while from the prediction of the abundance of demersal fish resources showed a higher abundance for year 2006. Abundance difference between the prediction made in 2005 and 2006 can also due to seasonal differences caused by different time-survey. Fish distribution patterns obtained from the mapping can be used as information about the specific fish behavior to environmental conditions. Estimation of abundance using geostatistical analysis can provide estimates of abundance of fish resources with smaller coefficient of variation and standard deviation than ordinary arithmetic calculations. This indicates the abundance estimation using geostatistical analysis that use the spatial structure into account in the analysis is a better estimator than the ordinary arithmetic calculation (non-geospatial). To improve the accuracy of geostatistical prediction-based can be done through increased coverage or increasing the sampling intensity and minimize the distance between samples, particularly for swept area data. In order to get more comprehensive condition of fish abundance it necessary to conduct regular monitoring and sample repetition.