Climate change affects different areas of our lives. The rise in the mean sea level, increases in temperature and changes in intensity and frequency of extreme events attributed to climate change lead to many socio-economical complications. Water management systems depend on the hydrological inputs for the engineering design. The main input for such designs, the IntensityDuration-Frequency (DF) curves, which were thought to be stationary is actually altered significantly with the changes in the extreme rainfall events. This paper discusses the possible variations in the IDF curves due to the climate change in Tawi Atair, Dhofar. It begins with the development of historical DF curves and then utilizing two General Circulation Models (GCM's) from the Couple Model Intercomparison Phase 5 (CMIP5) for the future projections of IDF curves during the years 2040-2059 and 2080-2099 by considering two emission scenarios, RCP4.5 and RCP8.5. A two-stage downscaling-disaggregation method was applied. Firstly, low spatial resolution GCM results were downscaled to the site-specific scale using weather generator (LarsWG). Secondly, a non-parametric disaggregation technique, known as the K-Nearest Neighbor (KNN), was used to disaggregate GCM daily time series to the hourly and sub-hourly time scales. The two GCM's were selected among 5 GCM's based on their performance to simulate the local precipitation in the Salalah region during 1950-2005. Accordingly, CNRM-CM5 and MRICGCM3 models were selected. The Lars-WG was calibrated with the observations of the study area for the years 1993-2009. The K-NN optimum disaggregation models for hourly and subhourly disaggregation were obtained based on the window size approach. The optimum window sizes for the hourly and sub-hourly disaggregation were found to be 28 days and 90 hours, respectively. The results show that the rainfall intensities are increasing especially for the short duration storms. For example, rainfall intensity of 5-min and 1-hr storms with 25-yr return period have increased by 51% and 14% respectively. The level of uncertainty related to different factors such as the methods used, the GCM's and RCP's selected and the statistical fitting of the Annual Maximum Rainfalls (AMR's) to the probability distribution function were discussed and presented in different formats for broader understanding.