Biologia 64/3: 522—526, 2009 Section Botany DOI: 10.2478/s11756-009-0074-y

Spatial-based assessment of land use, soil erosion, and water protection in the Jeneberang valley,

Sumbangan Baja1, Muhamad Ramli2 &SyamsulA.Lias1

1Department of Soil Science Hasanuddin University , Republic of Indonesia; email: [email protected] 2Soil Research Station Maros, Department of Agriculture, Republic of Indonesia

Abstract: Soil erosion by water is considered as one of the most significant forms of land degradation that affects sustained productivity of agricultural land use and water quality. It is influenced by a considerable number of factors (including climate, soil, topography, land use and types of land management), so that the information on the spatial distribution of soil erosion rate and its related effects can be effectively employed as a baseline data for land use development and water protection. The principal aim of this study is three-fold: (i) to map existing land use; (ii) to assess and map the spatial distribution of average annual rate of soil losses in the study area; (iii) to evaluate spatial matching between existing and proposed land use including a distance analysis from the water body (the Bili-Bili Dam). An analytical procedures used, respectively, include supervised classification of satellite imagery, application of RUSLE (Revised USLE), and overlay analysis in a raster GIS environment, utilising available information in the region covering some parts of Jeneberang catchment, South , Indonesia. The results suggest that the outputs of this study can be used for the identification of land units on a cell-basis with different land use types, rate of soil loss, inconsistency between proposed and planned land use, as well as the threat of land degradation to the main river and the dam. The analytical procedures developed in this research may be useful in other areas, particularly in the studies related to the assessment and mapping of land use and erosion for the importance of sustainable land use at a relatively large area. Key words: Jeneberang catchment; land use; soil erosion; GIS; RUSLE; spatial analysis

Introduction indicated as the area of “protection zone”, and in some parts long been recognised as the areas that provide Soil erosion by water is now becoming the main issue of fresh produce for urban region in the lower catchment. catchment management and development in Indonesia, Many types of horticulture production from this area and is considered as one of the most significant forms are marketed not only locally but also in other provin- of land degradation that affects sustained productiv- cial regions. ity of agricultural land and water quality. Factors af- Soil erosion and sedimentation are important phe- fecting soil erosion by water vary considerably; some nomena in this region, which threaten water quality are dynamic and changing in time and space, particu- in the main river and in the dam. A gigantic landslide larly in the regions where intensive use of land exists. took place on March 26, 2004 on the caldera wall of Mt. The Jeneberang valley located in the East of Makassar Bawakaraeng (2,830 m a.s.l.) in the uppermost reach of City, one of the largest cities in Indonesia, is part of the Jeneberang River. The huge mass of debris yielded Jeneberang catchment area having comparatively com- from this gigantic landslide traveled about 7 km down plex ecological characteristics, particularly in terms of the Jeneberang River. The volume of the slide mass geology and landform, river systems, and plant species. was estimated at about 240 million cubic meters with In addition, a considerable number of activities occur a head width of 1,600 m, a height 700 to 800 m, and within the catchment including forestry, intensive agri- a thick of approximately 200 m (Tsuchiya et al. 2004). culture and horticulture, rice field, livestock, and min- While there is lack of information on up-to-date land ing. use for the whole region, it is important to study the The multi-purposed Bili-Bili Dam with a catch- spatial distribution of land use and its related effects on ment area of around 38,400 ha exists in this area. The erosion, and to assess how they may influence the dam dam provides clean water for domestic (more than 1 that exist in the lower catchment. This paper presents million people) and industrial uses, as well as for about a result of study involving spatial analyses of land use, 24,000 ha of paddy fields. For clean water, the dam soil erosion by water, and current status of land use and supplies 3.300 litres per second, through 17 km under- land management that affects water regulation and thus ground piping systems. The uppermost catchment is protection in this area.

c 2009 Institute of Botany, Slovak Academy of Sciences Assessment of land use, soil erosion, and water protection in the Jeneberang valley 523

shrubs, mixed agriculture, paddy field1 (irrigated), paddy field2 (rainfed), water body, and rural residential. A spec- tral based classification strategy with supervised method was undertaken with the assistance of visual analysis of a displayed colour composite, vegetation index, plus ancil- lary information. Ground data collection was conducted to study land use patterns and characteristics in relation to their spectral response patterns on the satellite images. Us- ing a “maximum likelihood” technique, image classification was carried out on seven spectral bands and employing the signature files generated from the training areas. Classifica- tion processes involved the use of an automated technique first, and were followed, when subtle information existed, by binary masking (McKendry et al. 1992) with the help of ancillary information or visual interpretation. As many as 216 representative samples were used for a site-specific method of accuracy assessment in this project. Ground data collection was conducted to record the ac- tual land use at the sites chosen in the sampling procedure. Then, an error matrix containing errors of commission, er- rors of omission, and overall accuracy was established (see Fig. 1. Location of study area. Stehman 1999). Errors of commission relate to the cases where the algorithm applied has defined categories that do not exist on the ground. This measure, called user’s ac- curacy or reliability, is indicative of the probability that a Material and methods pixel classified on the map/image actually represents that Study area and database category on the ground. On the other hand, errors of omis- In this study, the lower parts of the Jeneberang catchment sion occur when certain categories that actually exist on area (in , Indonesia, see Figure 1) consisting the ground are omitted. This accuracy measure is called of 32,385 ha where the dam exists was selected and mapped producer’s accuracy. The relationships of such terms could using Geographic Information Systems (GIS) (Burrough & be expressed as follows: McDonnell 1998). The data used for analyses consist of 100 (i) Landsat ETM+ images (2004); (ii) soil data (Research User’s accuracy (%) = % – error of commission (%) 100 Center Hasanuddin University 1999); (iii) Digital Elevation Producer’s accuracy (%) = % – error of omission (%) Model (DEM) generated form 25 m contour lines with a grid (1) cell size of 30 m; (iv) rainfall data; (v) local government pro- posed land use map (Pemerintah Kab. Gowa, 2004); and (vi) Overall map accuracy is calculated by dividing the sum ground truth data obtained from field surveys. of the entries that form the major diagonal (i.e., the num- ber of correct classification) by the total number of samples taken. Methods Analysis methods used consist of three components. First, Calculation of annual soil loss classification of land use utilizing remotely sensed imagery Annual soil loss was calculated on a spatial basis, using the (using Landsat ETM+ images, with the date of acquisi- Revised Universal Soil Loss Equation (RUSLE), as follows tion of 21/08/2004 – soon after the gigantic landslide took (Renard et al. 1997): place in the uppermost area). Ground data collection was conducted to study land use patterns and characteristics A = RKLSCP (2) in relation to their spectral response patterns on the satel- lite images, and the data were necessary for training areas where A = average annual soil loss (t/ha/y) R = index of before spectral based classification techniques were used to rain erosivity, K = index of soil erodibility, LS =slopefac- derive thematic information. Second, calculation of annual tor (L = length, and S = gradient), C = land cover and soil erosion rate using the RUSLE (Revised Universal Soil management factor, and P = conservation practice (sup- Loss Equation) (Renard et al. 1997). Third, determination port) factor. (on the spatial basis) of land use types around the buffer Each erosion factor was generated in a rater GIS data zone of the Bili-Bili Dam, to reveal water protection from layer. Erosivity index (R) was calculated from rainfall data sedimentation around the catchment area, and to see the (represented only one station), with average annual rainfall consistency between proposed (planned) and existing land of 2.418 mm/year. Erodibility index (K) was derived from use types. soil great group (Table 1), LS factor was from DEM, and C and P factors were calculated from land use map. Therefore, Land use/land cover classification by using an overlay method with multiplication function the A methodological framework of thematic information ex- average annual soil loss was depicted on a cell-by-cell basis. traction (developed by Baja et al. 2002) using Landsat ETM+ images was used to produce a land use/land cover Generation of buffer zone and spatial matching map. In this study, three bands of Landsat ETM+ im- A map of buffer zone around the dam was done using raster ages (bands 1, 2, and 3) were employed. Following a site image calculation employing distance analysis (as demon- inspection, seven classes of relatively static land use/land strated by Baja et al. 2007). The distance intervals used cover were defined to establish the map legend: dense forest, are 0–100 m, 100–250 m, 250–500 m, 500–1000 m, and 524 S. Baja

(a) (b)

Fig. 2. Landsat ETM+ colour composite (a), and land use map (b).

Table 1. Soil type (great group) of study area. and quantity in the main river and in the Bili-Bili Dam, No Soil Great Group Area (ha) Percentage particularly since a very remarkable natural disaster – 1 Distropepts 8420 26,0 a gigantic landslide took place on March 26, 2004 on 2 Tropofluvents 3886 12,0 the caldera wall of Mt. Bawakaraeng (2,830 m a.s.l.) 3 Distropepts 6801 21,0 in the uppermost reach of the Jeneberang River. The 4 Haplustalfs 1295 4,0 volume of the slide mass was estimated at about 240 2915 9,0 5 Ustropepst million cubic meters with a head width of 1,600 m, a 6 Rhodustalfs 1263 3,9 7 Fluvaquents 5505 17,0 height 700 to 800 m, and a thick of approximately 200 8 Haplustalfs 1943 6,0 m (Tsuchiya et al. 2004). This landslide has brought 9 Epiaqualfs 162 0,5 about many adverse effects to land and water ecology 10 Ustifluvents 194 0,6 and human activities in the region. Total Area 32385 100,0 Based on the elevation data generated, as depicted in Digital Elevation Model (DEM) (see Fig. 3a), sedi- ment delivery potentially continue to occurs both from > 1000 m. An analysis of “spatial matching” (Baja et al. steep dryland around the dam and through the main 2005) between buffer zones for the Bili-Bili Dam and exist- river (Jeneberang) to the middle parts down to the Bili- inglandusewasdoneonacell-basedinrasterGISdata,so Bili dam. Based on the calculation of potential annual that the potential risk of sedimentation to water body can soil erosion rate, it was found that the dominant soil be explained on a spatial basis. The same technique was erosion rate is that area with a very low category (< 5 also applied to evaluate the matching between existing and ton/ha/yr) comprises 20,727 ha, followed by very high proposed land use information. category (> 100 ton/ha/yr) covers a total area of 5.330 ha, medium category (10–50 ton/ha/yr) 2,608 ha, and Results and discussion high (50–100 ton/ha/yr) 1,997 ha (Fig. 3b). The dam covers an area of 1,737 ha. The data reveal that with The colour (R+G+B) composite of Landsat ETM+ such a high potential annual erosion rate, regulation of images, is depicted in Figure 2a. Following a site in- land cover in the area around the river systems and the spection, seven classes of relatively static land use/land dam is indispensable to prevent excessive sedimentation cover were defined to establish the map legend in image in the water. classification: dense forest, shrubs, mixed agriculture, Spatial matching (obtained from overlay proce- paddy field 1 (irrigated), paddy field 2 (rainfed), water dure) between buffer zones for the Bili-Bili dam (see body, and rural residential (Fig. 2b). Based on the error Fig. 4a) and existing land use (see Fig. 2b) shows that matrix generated from an accuracy assessment, most a considerable number of land areas are inconsistent types of land use produce both user’s and producer’s with their recommended function in Spatial Planning accuracy of more than 80%. The overall accuracy of Law (Law No 26 2007). As seen in Figure 4b, within the map produced in this exercise is 89.5%. the 100 m buffer mixed agriculture dominates the area Soil erosion in this area has become one of very (i.e., 190 ha from a total buffer area of 291 ha). Accord- important factors affecting water flow, water quality, ing to the Spatial Planning Law, such areas as 100 m Assessment of land use, soil erosion, and water protection in the Jeneberang valley 525

(a) (b)

Fig. 3. Map of DEM (a), potential annual soil loss (b).

Buffer Zone and Land Use 600 Forest Shrubs Mixed Agr 400 Rice Field Water Resident Area (ha) 200

0 1-100m 100-250m 250-500m 500-1000m Buffer Zone

(a) (b)

Fig. 4. Map of dam buffer zone (a), and the histogram of land use distribution on different buffer zones (b). buffer zones around the dam should be well-protected land use types is depicted in Fig. 5. The area recom- (with forest as the main land use type). Further, buffer mended as ‘any type’ is mostly found in the west of areas of 100–250 m and 250–500 m are also dominated study area. From this match, it can be seen that shrubs by mixed agriculture and shrubs. dominates the area (1563 ha), followed by mixed agri- The study shows that the lower Jeneberang Valley culture, paddy fields, and forest. It is interesting to see around the Bili-Bili dam at the present time generally that for the area proposed to be forested (2126 ha), in undergoes a low erosion rate, as more than 60% of the fact forest only occupies 50 ha (2%), while paddy fields area tested has a potential soil loss rate of less than 5 comprises more than half (i.e. 60%) of recommended ton/ha/year. However, the area around the dam expe- areas. It also shows that the area recommended for an- riences relatively high potential soil loss that may cause nuals (with a total of 1544 ha) comprises 253 ha forest, high sedimentation rate into the water bodies. There- 396 ha shrubs, 528 ha mixed agriculture, and 364 ha fore, it requires an appropriate land management in the paddy fields. land area having a high potential risk as indicated in It should be noted that use of land should be di- potential soil loss map. Present land use classification rected to what recommended in the proposed map or in reveals that the area that should be allocated as 100 a certain case, it should anticipate land use occurrence m buffer zone around the dam consists of various land on areas more sensitive to the recommended types. For use types, but is dominated by mixed agriculture. Very example, areas recommended for ‘any type’ could fit limited buffer area of forest and scrubs is found around for any type of land use, but those recommended to be the water bodies, in order to meet the requirements of forested can fit only to forest or in some circumstances 30% of protected area on a catchment. to shrubs. Other areas proposed for annuals may fit to Spatial matching between proposed and existing any type of land use (see Baja et al. 2005), although 526 S. Baja

1600 Forest Shrubs 1200 Mixed Agr Paddy1 800 Paddy2 Area (ha)Area

400

0 Any-Type Forest Annual Perennial Recommended Land Use (Regulated)

(a) (b)

Fig. 5. Local government proposed land use map (a) and histogram of spatial match between proposed and existing land use (b).

priority must be given to annual crops. Such scheme is Burrough P.A. & McDonnell R.A. 1998. Principles of Geograph- used for the importance of sustainable land use (both ical Information Systems. Oxford University Press Inc., New in terms of productivity and degradation potential con- York. Law No 26 2007 on Spatial Planning. siderations). McKendry J.E., Eastman J.R., Martin K.S. & Fulk M.A. 1992. Explorations in Geographic Information Systems Technology: Applications in Forestry. Vol.2. United Nations Institute for References Training and Research. Switzerland. Pemerintah Kab. Gowa. 2004. Peta Penggunaan Lahan 2004. Baja S., Chapman D.M. & Dragovich D. 2002. Using GIS and re- Bappeda Kab. Gowa, Sungguminasa. mote sensing for assessing and mapping the present status of Renard K.G., Foster G.R., Weesies G.A., McCool D.K. & Yo- land use and land qualities in the lower Hawkesbury-Nepean der D.C. 1997. Predicting Soil Erosion by Water: A guide to Catchment, Australia. Geocarto International 17: 15–24. Conservation Planning with the Revised Universal Soil Loss Baja S. & Ramli M. 2005. Fuzzy modelling of environmental suit- Equation (RUSLE). USDA Agriculture Handbook No 703, ability index for land use development: application to lower Washington D.C. Jeneberang catchment area. Ecocelebica 1: 118–130. Research Centre, Hasanuddin University. 1999. Agricultural Soil Baja S., Dragovich D. & Chapman D. 2007. Spatial based com- Survey of Bili-Bili Irrigation Project Area. Research Centre, promise programming for multiple criteria decision making Hasanuddin University, Ujung Pandang. 89 pp. modeling in land use planning. Environ. Model. Asses. 12: Stehman S.V. 1999. Basic probability sampling design for the- 171–184. matic map accuracy assessment. Int. J. Remote Sens. 20: Baja S., Ramli M. & Jayadi M. 2005. Spatial matching be- 2423–2441. tween land use and land quality in the Jeneberang valley: Tsuchiya S., Koga S., Sasahara K., Matsui M., Nakahiro M., An analysis within a spatial planning perspective. In: Proc. Watanabe H., Shima H. & Yoshida K. 2004. Reconnaissance 2nd ASEAN Subcommittee on Space Technology and Appli- of the gigantic landslide occurred on Mt. Bawakaraeng in the cations Conference on Space Application Technology towards south Sulawesi state of Indonesia and unstable debris sedi- Competitive ASEAN, 5–11 August 2005. mentation (prompt report). J. Jap. Soc. Erosion Contr. En- gineer. 57: 3–4.

Received December 1, 2008 Accepted January 22, 2009