Cocoa plant mapping in Aceh Tamiang regency, Aceh Indonesia: an object-based image analysis approach

Cocoa is one of the leading commodities of plantations that has a vital role in the Indonesian economy. Due to the unavailability of detailed cocoa area data in Aceh Tamiang Regency, mapping was carried out by utilizing remote sensing technology. This study aims to map and determine the distribution of cocoa plants in Aceh Tamiang Regency using the object-based image analysis (OBIA) method and the accuracy test of the OBIA method to map cocoa plants. The object-based image analysis (OBIA) classification results for cocoa plants covered 946.87 ha and cocoa plants spread across 12 districts in Aceh Tamiang Regency. The widest distribution of cocoa is in Karang Baru District, 139.22 ha with a percentage of 14.70%, and the smallest area in Kota Kualasimpang District, covering an area of 6.30 ha with a rate of 0.67%. The results of the accuracy test to map cocoa plants using the OBIA method obtained an overall accuracy value of 95% and a kappa value of 86%.


Introduction
Cocoa is one of the leading commodities of plantations that has an essential role in the Indonesian economy.Besides being the country's foreign exchange earner, a source of income for farmers, and a creator of jobs, cocoa is also the only main ingredient in chocolate-making.
Indonesia is one of the largest cocoa-producing countries, producing 732,256 tons in 2022 [1].Indonesia's cocoa production center is the Sulawesi archipelago, covering Central Sulawesi, Southeast Sulawesi, South Sulawesi, and West Sulawesi in addition to the Sulawesi islands, Aceh Province which agroecosystem-wise has excellent potential for cocoa development [2].One of the districts in Aceh that has the potential to develop cocoa is Aceh Tamiang Regency.
According to data from the Badan Pusat Statistik (BPS) in 2022, cocoa is the third largest crop after oil palm and rubber in Aceh Tamiang, with an area of 1,255 ha.The planting area of this commodity changes every year.In 2018, the cocoa planting area in Aceh Tamiang was 1,674 ha.In 2019, it decreased to 1,337 ha.In 2020, it increased again to 1,515 ha; in 2021, the data changed to 1,255 ha [3].There is no detailed data on the area of cocoa plants, so it is necessary to do mapping by utilizing remote sensing technology to obtain data on the distribution of cocoa plants in Aceh Tamiang.Remote sensing is the science or art of obtaining information about an area, phenomenon, or object by analyzing the data obtained using the tool, area, or symptom to be studied without direct contact with the object [4].One analysis to obtain accurate information and results in representing the suitability between images and conditions in the field is by classification techniques.
Remote sensing technology that can be utilized to map cocoa plants is an object-based classification method.This classification method minimizes the weaknesses of pixel-based classification, which still experiences salt and pepper effects at the classification stage, as well as still subjective visuals [5].Over time, remote sensing image information extraction has various forms of analysis, one of which is Object-Based Image Analysis (OBIA) [6].
Object-based image analysis (OBIA) is an approach whose classification process not only considers spectral but also spatial aspects of objects.In general, the classification process with the OBIA method goes through two stages, namely, segmentation and classification of each segment [7].This object classification must begin with segmentation, which aims to separate classification objects with certain conditions, where three factors affect segmentation: scale, shape, and compactness [8].
OBIA has its advantages compared to pixel-based methods because the pixel-based methods only consider spectral aspects.In contrast, OBIA methods are much more sophisticated because they consider spectral, shape, and texture aspects [9].The OBIA method has several disadvantages compared to the pixel method, including that the accuracy of segmentation decreases with increasing segmentation scale, the negative impact of segmentation errors becomes greater when it is on a large segmentation scale, and the overall accuracy of object-based classification is highly dependent on the scale of segmentation [10].
This study aims to map and determine the distribution of cocoa plants in Aceh Tamiang Regency using the object-based image analysis (OBIA) method and the accuracy test of the OBIA method to map cocoa plants.

Materials and methods
The study area in this study is Aceh Tamiang Regency.Spatially, the study area can be seen in Figure 1.This research was conducted by quantitative descriptive analysis and image data extraction using the object-based image analysis (OBIA) method.The study used laptops, GPS, eCognition 9.0 software, ArcGIS 10.8, Global Mapper 20, cameras, and Google Earth Imagery acquired in 2019, 2020, and 2021 Aceh Tamiang Regency.The data for 2019 to 2022 was selected because Google Earth imagery data for Aceh Tamiang District were available in those years.
In the first step, image analysis is a pre-analysis process that converts image data in Enhanced Compression Wavelet (.ecw) format into Tagged Image File (.tif).This data conversion is done in the Global Mapper software.Furthermore, data analysis in this study is segmentation and image classification using eCognition Developer software.Segmentation is the initial stage before image classification with the OBIA method.The algorithm used for segmentation in this study is the multiresolution segmentation algorithm.This segmentation separates the image into regions or objects based on predetermined parameters [11].
The segmentation process is run based on three parameters: scale, shape, and compactness.In this study, the parameter values used, namely the parameter scale of 150, the shape of 0.3, and compactness of 0.7, resulted in 574,761 objects that could separate objects in detail well.
Furthermore, the classification process uses membership functions and classification algorithms.Classification begins by selecting a sample into a class of segmented objects.After the sample selection process, determine the membership function to use.Then, the classification is done with the classification algorithm.Post-analysis conducted in this study is editing class if the classification results are found in inappropriate classes.Ground check is carried out at this stage by taking points in the field as sample points to validate the results of classification and accuracy tests.
Accuracy test to see the level of error that occurs in the classification to determine the level of accuracy of the method against the results of the classification that has been done.The accuracy test used a confusion matrix table, which connects classification results and data taken from the field.This test produces overall producer, user, and kappa accuracy.

Image classification results
The classification results mapping cocoa plants in Aceh Tamiang Regency using the OBIA method obtained an area of 946.87 ha.Based on BPS data from Aceh Tamiang Regency in 2022, the cocoa area in Aceh Tamiang is 1,255 ha.The difference between BPS data and classification results is 308.13 ha.Currently, the area of cocoa in Aceh Tamiang has begun to decline in several regions.Based on the results of interviews in the field, people who previously cultivated cocoa plants are now switching to oil palm.This is because pests often attack cocoa plants, so they prefer to plant oil palm because the cost of palm oil care is cheaper and maintenance is more manageable.As is known, oil palm is a plantation commodity with the largest area in Aceh Tamiang Regency.The difference in area is also influenced by the satellite images used; in this study, the images used are 2019, 2020, and 2021.Meanwhile, cocoa area data from BPS is from 2021.In this study, field surveys have been carried out where Aceh Tamiang has been reduced a lot so that it affects the area obtained from the image classification process.
It is hoped that further research will be carried out with the same method by trying a better combination of membership functions to get better and more accurate classification results on cocoa plant mapping in Aceh Tamiang Regency.

Distribution of cocoa plants in Aceh Tamiang Regency
Based on the classification results, cocoa plants were distributed in Aceh Tamiang Regency, covering an area of 946.87 ha, and spread in 12 districts in Aceh Tamiang Regency.The distribution of cocoa plants can be seen in Table 1.
From the classification results, the distribution of cocoa with the largest area in Aceh Tamiang is in Karang Baru District, which is 139.22 ha with a percentage of 14.70%, and the smallest cocoa area of 6.30 ha with a percentage of 0.67% in Kota Kualasimpang District.Spatially, the distribution of cocoa in Aceh Tamiang District can be seen in Figure 2.

Accuracy test
From the results of classification using the OBIA method, the level of accuracy with the confusion matrix table is obtained.The accuracy test results can be seen in Table 2. Accuracy testing produces a producer's accuracy value for the cocoa class of 96% for the user's accuracy in the cocoa class of 98%.In the non-cocoa class, the producers' accuracy value obtained is 93%, while the resulting users' accuracy is 86%.The overall accuracy value produced by both classes is 95%, with a kappa value of 86%.The acceptable result of the classification process is that the kappa accuracy value is more than or equal to 85% [12].The higher the accuracy and kappa coefficient, the better the classification results.
Classification carried out using resolution images with the object-based image analysis method produces a high accuracy value; this can facilitate the classification of cocoa plants.However, with high accuracy results, this method can also be used to classify plants with other shades, such as coffee plants.

Conclusions
Based on the results of classification using the Object-Based Image Analysis (OBIA) method to map cocoa plants in Aceh Tamiang Regency, an area of 946.87 ha was obtained and spread across 12 districts in Aceh Tamiang Regency.The distribution of cocoa's most significant area is in Karang Baru District, with an area of 139.22 ha with a percentage of 14.70%, and the smallest area of 6.30 ha with a percentage of 0.67% in Kota Kualasimpang District.The accuracy of the OBIA method for cocoa plant mapping is 95% for overall accuracy and 86% for kappa accuracy.

3 Figure 1 .
Figure 1.Map of the study area.

Figure 2 .
Figure 2. Map of cocoa distribution in Aceh Tamiang Regency.

Table 1 .
Area and distribution of cocoa plants from OBIA classification in Aceh Tamiang Regency.

Table 2 .
Cocoa classification accuracy test results using OBIA.