Drought Disaster Forecasting Based On Rainfall Runoff Transformation Modelling (Case Studies In Tukad Petanu Watershed)

Ardana, Putu Doddy Heka (2014) Drought Disaster Forecasting Based On Rainfall Runoff Transformation Modelling (Case Studies In Tukad Petanu Watershed). In: CARE (Care About Risk And Environment) 2 Internasional Conference.

	Text Proceeding Care 2014 - Isi.pdf Download (621kB)
	Text Cover Prosiding_merged.pdf Download (1MB)

Abstract

Drought is one of the major weather related disasters. Persisting over months or years, it can affect large areas and may have serious environmental, social and economic impacts. drought is a normal, recurrent feature of climate, although many erroneously consider it a rare and random event. It occurs in virtually all areas, whatever their normal climate may be, and the characteristics of a drought may be very different from one region to another. Drought is difficult to define precisely, but operational definitions often help define the onset, severity, and end of droughts. An operational definition of drought helps people to identify the beginning, end, and degree of severity of a drought. Early detection of droughts helps to implement drought mitigation strategies and measures, before they occur. Therefore, drought forecasting plays an important role in the planning and management of water resources systems, especially during dry climatic periods. However, drought analysis and forecasting are not always easy. This study developed an objective drought analysis through a rainfaal runoff model which developed from simulated Feed Forward Backpropagation Neural Network to generate flow (discharge/runoff) the drought threshold with probability of 0,5 called Q (Q). From the resulted flow rate, the deficit and drought duration will be estimated and this way mean can be used to droughts analysisi effectively and forecast future drought conditions. In addition, based on the forecast discharge data, can also he used to predict the incidence of drought in the next years. From the results of the analysis to get the level of performance models, feed forward back propagation neural network method, type 3-10-5-1 architecture (model 5) provide the most optimum results to describe the rainfall-runoff relationships that occur in the watershed Tukad Petanu. The value of MSE network is 0,05584, "r" for training process is 0,8688 (86,88%) and 0,7536 (75,36%) for testing and AAE value is 0,1322 for training process and 0,1971 for testing process. That is mean the pattern of the target data with the prediction data shows a pattern elose to each other, which means it has a balance between the process of memorization and generalization of network. So, the rainfall runoff transformation can be used to calculate the drought deficit, duration and drought forecasting. From analysis of drought can be seen that the results between the Quli and Qaerasi Quermut showed a significant difference. This can be seen in the duration of the drought that occurred on average for a year. Based QaPal average drought duration obtained for 5.8 months while based Qu average drought duration obtained during 7.3 months. As for the dry months, based on the two analyzes (Qnervai and Quimadas). month began to dry starting with the average in July and ends in December. In the last in forecasting, from the analysis result obtained, in 2004 Qsimin will occur in May as big as 4,343 m3/dt and in August the discharge in the amount of 2,245 m'/dt. Dry (drought) season will start in July and will end in December. The value of the deficit amounted to 1,011 m'/dt. in 2005 Qmanimum will occur in January as big as 2,783 m3/dt and in August the discharge in the amount of 1,592 m'/dt. Dry (drought) season will start in July and will end in December. The value of the deficit amounted to 0,332 m'/dt Keywords : drought, rainfall runoff transformation, feed forward backpropagation neural network, forcasting

Actions (login required)

View Item