Tank Model Application for Runoff and Infiltration Analysis on Sub-Watersheds in Lalindu River in South East Sulawesi Indonesia

Improper land management often causes flood, this is due to uncontrolled runoff. Runoff is affected by the management of the land cover. The phenomena also occurred in South East Sulawesi, Indonesia. This study aims to analyze the flow rate of water in watershed of Lalindu River in North Konawe, South East Sulawesi by using a Tank Model. The model determined the magnitude of the hydrologic runoff, infiltration capacity and soil water content several land uses were evaluated in the study area. The experimental and calculation results show that the runoff in the forest is 2,639.21 mm/year, in the reed is 2,517.05 mm/year, in the oil palm with a slope more than 45% is 2,715.36 mm/year, and in the oil palm with slopes less than 45% is 2,709.59 mm/year. Infiltration in the forest is 30.70 mm/year, in the reed is 7.51 mm/year, in the palm oil with a slope more than 45% is 24.13 mm/year and in the palm oil with slopes less than 45% is 29.67 mm/year. Runoff contributes to stream flow for water availability.


Introduction
Changes in land use in the watersheds (WS) provide a dominant influence on flood discharge [1,2]. The greatest influence changes in land use on the sustainability of water resources is the changes from forest land to other uses such as agriculture, farming, residence or industry. If these activities are not well-managed, it will result in water excess (flood) during the rainy season and drought during the dry season. Unwisely changes in land use and is not accompanied by conservation measures will largely become runoff. Watersheds can be viewed as a hydrological system in which precipitation is the input from streams, and evapotranspiration is the system output. Furthermore, the watershed is the place where the simultaneous processes happen and become a part of the hydrological cycle.
The expansion of oil palm plantations keeps going in Indonesia, including in Southeast Sulawesi, which is spread almost all over North Konawe Regency, especially in Wiwirano District [3]. The changes of land cover grow rapidly and it is one of areas used as a priority area of oil palm development. It can be viewed from the increasingly growing oil palm plantation during the last five years, both operated by the community and by private plantations. The land use ignoring sustainability aspect, such as deforestation will put impact on the reduced water discharge in Lalindu watersheds. The deforestation is a consequence of the land function transfer. Land use transfer has resulted in changes of land structure. During dry season, the soil becomes hard and barren due to absorption strength of the sun, penetrating into the soil. If various trees still exist, the sun heat will be muffled in the foliage and the soil will stay fertile for the stable humidity. In addition, travel time of the water is relatively short, thus the rate of water infiltration by land is slower than the runoff rate. It is influenced by a number factors, among others; by concave topography of Mount Wiwirano (although the nature of the soil supports the absorption of water into the soil) resulting in most rainwater that will be lost through runoff  and later run towards the river and eventually into the sea, before the rain water fills the underground water through infiltration and percolation process. A settlement is located surrounding this valley, thus during heavy rains, floods will occur. To investigate the problem, a tank model is applied to determine watershed parameters in this area. The tank model base on the hypothesis that the runoff flow and infiltration is a function of the water amount presents in the soil [4][5][6]. The tank model can be constructed in such a way, thus representing the function of sub-watersheds area, or represents the difference in the structure/ type of land in each layer. Besides explaining the lost of initial rainfall and the dependence on prior rain, the tank model can also present some of the components forming the runoff flow, which have specific period and time lag. The tank model structure is the closest model to each watershed [4,5]. A tank with wasting channel on its side represents runoff, lower wasting channel represents infiltration, and saving component represents the runoff processes in one watershed or in watershed partly. Several parallel similar tanks can represent a large watershed [6]. In this study flow rate is analyzed by using a model consisted of runoff parameter, infiltration capacity and ground water content.

Material and Method
The research was conducted on sub-watersheds in Lalindu, North Konawe, one of the districts in Southeast Sulawesi, Indonesia. Observations made on hydrological conditions on 4 sub-watersheds which were parts of the Lalindu sub-watershed. The placement of 4 sample plots in each sub-watershed represented the three types of vegetation or land cover in the sub-watershed. The vegetations are the forest, reeds and palm oil. Palm oil itself was divided into two by the steepness, i.e. palm with steepness > 15% and palm oil with steepness <15%. The model was verified by observing the sample plots to obtain parameter value of the first tank ( Fig. 1)

Figure 1. Tank model used in the research
The runoff parameters can be formulated as following equations. The basic equation for the first tank is as follows: The equation for the second tank is as follows: The equation for the third tank is as follows: The equation for the fourth tank is as follows: . a 4 (4) The runoff discharge from river (Q) is calculated by the following equation: The data used in this research were primary and secondary data. The primary data included water discharge data on sample plots, sub-watersheds and sub-watershed of one year observation. Secondary data was the climate data, soil-type data and biophysical conditions. The data analysis was performed on the amount of rainfall, infiltration, evapotranspiration and runoff. Furthermore, tank model creation was made to describe runoff processes occurring in the watershed.
Tank model was validated by using daily rainfall and daily actual discharge data from the results of direct measurements in the field. The actual discharge data used in this validation process was the actual discharge data in the sub watersheds of gardens and other lands located in the surrounding garden location of the research.

Results and Discussion
Daily rainfall and evapotranspiration (ETc) data from the calculation results of climate data in the garden in 2011 are presented in Fig. 2. The data is used as an input value in the analysis of the tank model. The discharge measurement of sample plots on four types of land cover is forest, reeds, and palm oil land, with a steepness > 15% and <15%. The discharge data for sample plots of the size 16 m 2 is shown in  In addition, the discharge measurement is conducted at sub-watersheds in the area of 20.000 ha. The discharge measurement data at the sub-watersheds is influenced also by run-off from the garden area in the surrounding area of the watershed. The discharge data of sub-watersheds from direct measurement is presented in Fig. 4.

Tank Model Calibration in Sample Plots
The measurement result data in sample plots is used as a basis for calibrating the tank model to determine the magnitude of the tank model coefficient in the first tank. The calibration of each type of land result on sample plots for the first tank on tank models are presented in Table 1. The analysis of tank model of on the sample plots is conducted to determine the land cover characteristic towards the water.

Tank Model Calibration in Sub-Watersheds
In Table 2, the parameter coefficient values of the tank model calibration results in sub-watersheds is determined by using the data of 2011. From the calibration results, the coefficient of determination is obtained to be more than 70% for each land use. Based on coefficient values of tank model from the calibration result, it is shown that the coefficient of the tank outlet is smaller to the lower part. It is due to the deeper soil layers, the ability of the soil to have water run-off is smaller. Similarly, the coefficient value of the tank downward (z) is smaller downward. That is due to the deeper layers of the soil, the soil capacity to carry water into the deeper layers (percolation) is getting smaller. From the results of the calibration tank model, it is known that ground water content (xx) is mostly found in the fourth tank. The water content in the soil at each level of the tank depth is greatly influenced by the type of plants that living above, as each model of discharge model outcome approaches the actual discharge model. In the process of model validation, the coefficient value of tank models that have been obtained from the calibration results in 2011 from the previous stage is used. This validation process is obtained from the model discharge in the sub-watersheds of gardens and other surrounding land, describing the response towards the rainfall. The validation result has determination coefficient of 74.52%, and a tank model is then used to perform scenario of land conversion into oil palm plantation using conservation methods.

The Analysis of Land Cover Changes on Water Supply
The next stage is to analyze the model in order to determine the total runoff and the total infiltration value of each different land cover. The total runoff and total infiltration value from the model analysis results in sub-watersheds at each different land cover is presented in Figure 6 and 7. The amount of runoff and infiltration rate is used to determine the state of water in Lalindu sub-watersheds for the amount comparison of the runoff value, and the infiltration is being simulated for area of 500 hectares from each land cover type.

Analysis of the relationship among soil components, water and vegetation
The runoff total and infiltration on model analysis are then used to examine the relationship among land of soil components, water and vegetation of each unit. There is a close relationship among the soil components, water and vegetations. Land is a medium for the vegetation growth. Different soil types will have distinctive characteristics in terms of the soil physical, biological, and chemical properties. Soil properties can determine the type of nutrients in the soil, amount of water that can be stored in the soil, and root systems reflecting the circulation of water movement in the soil. The soil ability in absorbing water is reflected in the vegetation types at the ground level. The vegetation function can effectively reflect the ability of the soil to absorb rainfall, maintain or increase the infiltration rate, and demonstrate the ability in restraining water or water retention capacity [7,8].

Conclusion
Tank Model was successfully applied in Lalindu watersheds to determine magnitude of the hydrologic runoff, infiltration capacity, and soil water content.
1. The hydrological condition in Lalindu sub-watersheds can be presented by Tank Model with coefficient calibration (R 2 ) of 74.52%. 2. The analysis of runoff the Tank Model at each land cover showed that the runoff discharge in the forest is 2,639.21 mm/ year, in the reeds is 2,517.05 mm /year, in the oil palm with a steepness of more than 45% is 2,715.36 mm/ year and in the oil palm with steepness less than 45% is 2,709.59 mm/ year. 3. The analysis of infiltration in the tank model at each land cover resulted that the infiltration in the forest is 30.70 mm/ year, in the reeds is 7.51 mm/ year, in the palm oil with steepness more than 45% is 24.13 mm/ year and in the oil with steepness of less than 45% is 29.67 mm/ year.