Land cover changes in Bila Watershed, South Sulawesi, Indonesia

Changes in land cover are characterized by changes in land use conversion. One of the factors that cause changes in land closure is the increase in the number of residents so the need for residential land also increases. This study aims to identify changes in land cover in 2017 and 2021 in the Bila River Basin. Land cover change analysis is carried out using the Geographic Information System approach using the overlay analysis method in ArcGIS software. The results showed that in 2017-2021, secondary dryland forest land cover experienced a reduction in the area of 915.96 hectares or from the total area of watersheds converted to mixed dryland agriculture. Meanwhile, the largest addition occurred in the dry land closure class covering an area of 634.08 ha of the total watershed area converted to dryland farming. This is due to the increasing need and dependence of the community on land, with community activities in managing land and utilizing the land to increase agricultural yields as well as a source of livelihood for the community.


Introduction
Such rapid population growth and development led to changes in land use functions.The increase in population certainly requires the provision of facilities and infrastructure to meet the needs which ultimately demands land conversion [1].The Ministry of Home Affairs through the Directorate General of Population and Civil Registration released population data for the second semester of 2021 which contains information on the number of Indonesian residents, namely 273,879,750 people.This number has increased by 2,529,861 people compared to 2020.In line with the demand and fulfillment of land needs, resulting in a shift in the function or conversion of land functions that were previously intended as forest areas, water catchment areas, and plantation agriculture changed their function to commercial areas.The transition of the function of an area that is able to absorb water (previous) to an impervious area will result in hydrological imbalances and negatively affect the condition of the watershed [2].Land cover is influential in the movement of water in the watershed.The vegetated land cover serves as a soil protector against the impact of rainwater grains as well as against the carrying capacity of surface flow water (runoff), as well as increasing water infiltration into the soil [3].Forest cover in a watershed is the most effective land cover naturally to control the movement of rainfall so as to reduce erosion and sedimentation, increase the absorption of water into the soil and reduce other parts to surface water, as well as flood and drought control [4].
Land use change has caused negative impacts on land and water resources that occur in watershed areas.Land use change in watershed areas affects watershed hydrological conditions such as erosion, land degradation, flooding, and increased surface runoff coefficient [5].
One of the areas in Indonesia that are included in the flood-prone area during the rainy season is Wajo Regency, South Sulawesi.The flooding that occurred was caused by the overflow of Lake Tempe.Based on the news reported by Antara news, on August 28, 2021, it was stated that rain with moderate to heavy intensity over the past two days (25-26/8/2021) caused water discharge from Lake Tempe to increase to overflow which resulted in dozens of districts in Wajo Regency to Bone Regency being The Bila watershed, which is one of the watersheds that empty into Lake Tempe, is also a factor in causing flooding in Lake Tempe.This is due to the existence of water catchment areas that are experiencing land conversion [6].Land that is supposed to function as a buffer area and water catchment experiences a decrease in land quality, thereby increasing the rate of surface flow, erosion, and sedimentation.The accumulated sediment makes Lake Tempe shallower and reduces water volume capacity [7].
Based on the description of the problem, it is necessary to use a model to help predict the amount of surface runoff that occurs in the Bila watershed and the effect of land cover changes on surface runoff.One of the models that can be used is the Soil and Water Assessment Tool (SWAT), this model allows it to be applied in various analyses and simulations of a watershed to see the quality of the watershed and its prolonged impact based on specific climate information, soil characteristics, topography, vegetation, and land management in a watershed [8].In addition, the SWAT model can be applied in assessing the hydrological response of a catchment area with a water balance approach based on land use change scenarios [9].

Study area
The Bila watershed is one of the watersheds in South Sulawesi.The research site is located in the Bila watershed which administratively crosses three districts, namely Enrekang Regency, Sidenreng Rappang, and Wajo Regency with an area of 164,735.78hectares (ha).The geographical location of the Bila watershed is 3°44'9.20"BT and 119°59'27.23"LS.This research was conducted for four months, namely the Month of July until November 2022.Data processing is carried out at the Watershed Management Laboratory, Faculty of Forestry, Hasanuddin University.The map of the study location can be seen in Figure 1.The method of this research begins with setting the boundaries of the research location, then continues with data collection, image interpretation, checking, and data collection in the field and the last is the accuracy test.

Material and methods
3.1.Preparation data 3.1.1.Accuracy test.The accuracy test is a comparison between two data from image interpretation and field conditions.The method of calculating the accuracy value of image interpretation results to test the magnitude of accuracy using the kappa accuracy confusion matrix is used as the first step in describing differences and statistical analysis techniques to assess the accuracy of the map [10].In the confusion matrix, the data from the interpretation of the image and the data from field checks are arranged in a percentage comparison table.The result of an acceptable level of accuracy of image interpretation is a classification process that has a kappa accuracy value of more than or equal to 85% or a coefficient of 0.85 [11].the confusion matrix table can be seen in Table 1.Kappa accuracy is the accuracy that considers all the elements present in the confusion matrix.The accuracy of kappa is often considered the most relevant because it includes its omission and commission error [12].At this time, the recommended accuracy is kappa accuracy, because overall accuracy, in general, is still overestimated.The accuracy of kappa is often also called the kappa index.Mathematically the accuracy of kappa is presented as follows:

Land cover 2021
The results of the classification of land cover contained in the Bila watershed in 2021 are primary dryland forests, secondary dryland forests, open land, grasslands/savannas, settlements, dryland agriculture, rice fields, shrubs, swamp thickets and water bodies with the largest land closures, namely rice fields covering an area of 19,729.80 or 37.79% or of the total area.The area and percentage of land cover in 2021 can be seen in Table 3.The results of the interpretation of land closure in 2021 then carried out an accuracy test of the classification results obtained by comparing land closures in the field through a field survey (ground check).There are 212 coordinate points in the land cover class determined by taking into account the accessibility ability of researchers to retrieve coordinate points.The confusion matrix table can be seen in Table 4.As a result of the confusion of the field checkpoint matrix of each class of land closure, it is known that there are 212 sample points with the correct number of points in the field as many as 197 points.From these results, the level of confidence obtained from the interpretation of the image of the overall accuracy value of 92.92% and the kappa accuracy index value of 0.91 or 91.12%.The acceptable accuracy of image interpretation is 85% [13] and the Kappa value is said to be very good if the Kappa value > 0.80 [14].This suggests that the interpretation of the Sentinel-2 image is acceptable.

Land Cover Changes
Land cover resulted from the interpretation of imagery in 2017 and 2021 Sentinel-2 Imagery.The results of the interpretation of this image, show that there are 10 classes of land cover located in the Bila watershed area with different areas.The class and extent of land cover in the Bila watershed can be seen in Table 5. Information : (+) = Area Addition (-) = Area Reduction It can be seen that the condition of land closure of the Bila watershed from 2017 to 2021 is most dominant in dryland agriculture and primary dryland forests.Meanwhile, the land cover that has the lowest area is open land and grassland/savannah.The condition of land closure in the watershed when has undergone changes in both the addition and reduction of area.The increase in the area occurs in the classes of open land, pastures/savannas, settlements, dryland agriculture, rice fields, and water bodies.The largest reduction occurred in the secondary dryland forest cover class of 915.96 ha, this is because the land is converted to dryland agricultural land cover.Forest conservation is threatened by disturbances caused by the conversion of forests into non-forest areas, especially agricultural and plantation businesses due to the increase in population and needs [15].The change from secondary dryland forest to dryland agriculture occurs due to an increase in population accompanied by the fulfillment of living needs so there is a change in land use due to human intervention to grow crops [16].
Perceived susceptibility and severity of land degradation strongly influences farmers' awareness of, and attitude toward environmental problems.This rising rate can attribute to the growing population in need of agricultural land in the study area.The main occupation of social community in that area is farming and located in the mountainous area.It is hard to avoid this agricultural practices because this has become people's culture for agriculture in mountainous region sand have made it hereditary.In Indonesia, where many upland areas can found, land use/cover change for the extension of agriculture activity commonly occurs [17].
These changes can have many negative impacts such as natural disasters of erosion and landslides.Changes in land cover from tight to open land can lead to erosion [18].The secondary dryland forest at the study site is mostly on steep marble, in the event of land conversion on this slope, the chances of erosion and landslides are very likely.The steeper the slope of the slope, the greater the chance of a landslide [20].The largest increase in the area occurred in the class of closure of dryland agricultural land, rice fields, and settlements.Some of the factors that cause this change are due to the increase in population from year to year which affects the increase in residential land because the community needs land as a place to live.The increase in population is always followed by an increase in the need for land [19].
Land conversion in some of the land cover is due to the increasing need and dependence of the community on land, with community activities in managing land and utilizing the land to increase agricultural output as well as a source of livelihood for the community.Land use change is a logical consequence of the growth and transformation of changes in the socioeconomic structure of developing societies [20].The land cover change reflects human nature that always seeks to utilize and manage the surrounding land resources for agricultural activities [21].The map of land cover changes for 2017 -2021 can be seen in Figure 4.

Conclusion
The results of the classification of land cover in the Bila watershed consist of ten classes of land cover, namely primary dryland forests, secondary dryland forests, open land, grasslands/savannas, settlements, dryland agriculture, rice fields, shrubs, swamp thickets, and water bodies.The land cover that dominates in this region is dryland agriculture and the smallest is open land.In 2017 -2021, it was identified that the dominant land area that experienced the largest reduction was secondary dryland forest covering an area of 915.96 ha, while the largest addition occurred in the dry land agricultural land closure class covering an area of 634.08 ha of the total watershed area converted to dryland farming.This is due to the increasing need and dependence of the community on land, with community activities in managing land and utilizing the land to increase agricultural output as well as a source of livelihood for the community.
of thousands of houses and hundreds of hectares of rice fields being flooded.

3 (
The data used include Sentinel-2 Imagery in 2017 and 2021, Watershed Boundaries in 2018 from the Director General of Watershed Control and Protected Forests, Provincial Administration Map, road network data, and river networks of South Sulawesi from the Indonesian Geospatial Portal.The tools used are laptops that are scarce with Geographic Information System (GIS) applications, GPS Receivers 1230 (2023) 012045 IOP Publishing doi:10.1088/1755-1315/1230/1/012045Geographic Positioning System), cameras, and stationary writing.

Figure 4 .
Figure 4. Distribution of flood event points.

Table 2 .
Analysis of land cover changes is carried out by comparing the land closure maps for 2017 and 2021 by overlaying them so that it will be seen what closures have changed over a period of five years.The changes that occur next are made in the form of tables and graphs to make it easier to see the changes in land cover that occur.1230(2023)012045Theresults of the classification of land cover using Sentinel-2 imagery in 2017 contained in the Bila watershed consist of ten classes of land cover, namely primary dryland forest, secondary dryland forest, open land, grassland/savannah, settlement, dryland agriculture, rice fields, shrubs, swamp thickets, and water bodies.The dominating land cover in this area is dryland agriculture covering an area of 48,754.37 ha or 29.60% of the total area of the Bila watershed.The area and percentage of land cover in 2017 can be seen in Table2.Land cover area in 2017 Total 164,735.78100.00 Figure 2. Land Cover Map in 2017.

Table 3 .
Land cover area in 2021

Table 4 .
Confusion matrix of land cover sample points in 2021

Table 5 .
Area of Bila Watershed Based on Land Cover in 2017 and 2021