چكيده به لاتين
Increasing greenhouse gas concentration has led to continuous climate change. Climate changes can lead to changes in catchment hydrological regime and performance of water resources systems in the future. So assessment of climate change impacts phenomenon is important and must be considered more than ever. For climate change impacts assessment, GCM models are used. GCMs scenarios are large-scale; In order to use the GCM outputs in the catchment scale, downscaling is required. The results accuracy of the assessment of the regional impacts of climate change is highly dependent on the accuracy of the downscaling method. so the development of an appropriate downscaling method is an important step in predicting the climate change impacts on catchment hydrological regime. Weather generators (WGs) models have major potential advantages for downscaling and assessing the hydrological impacts of climate change. Despite their many capabilities, the weakness of these models in: 1-reproduce low-frequency variabilities, 2- inconsistency of future scenarios with scenarios derived from GCMs, and 3- reproduce the extreme values, has limited the efficiency of these models.
In IWG model, many of the weaknesses of the WGs model (in terms of reproducing the extreme values, low frequency variabilities of temperature, and the impacts of rainfall occurrence changes on secondary variables in downscaling) have been resolved. However, low frequency variabilities of rainfall have not been corrected in this model. By correcting the weaknesses of this model, an efficient method for downscaling the climate variables in assessing the impacts of climate change, especially for hydrological purposes, can be achieved.
In this thesis, the performance of the IWG model is compared with that of the SDSM and LARS-WG models to reproduce a wide range of observational time series characteristics. Then, two methods are presented for correcting low frequency variabilities of this model. The first method is based on using the variable values of the model parameters during different months of the monthly time series. In this method, monthly statistics related to large-scale atmospheric variables are simulated, and are entered into the model for each month. In model validation, it was observed that this method overestimates the low frequency variabilities. In the second method, the bias of the theoretical distribution of produced variables in the monthly time scale was modified by the Quantile Perturbation Method (QPM), and the monthly distribution of produced variables was adjusted to the theoretical distribution fitted to the observational variables. Then, by a linear function, the daily produced values were matched to the monthly values. Also, a downscaling method appropriate to this method was presented, by the perturbation of model parameters based on GCM scenarios.
On the other hand, a method for modifying low frequency variabilities of LARS-WG model was proposed. Since this model uses semi-empirical distributions (instead of theoretical distributions in other WGs), the proposed method for modifying rainfall low-frequency variabilities based on QPM has been adapted for these distributions. Also, the method of downscaling the monthly rainfall of this model is based on QPM (instead of perturbing the parameters). To correct the low frequency variabilities of the temperature, the IWG model method based on the multivariate AR model has been used.
A wide range of statistical tests have been used to evaluate the performance of the models and compare it with the performance of the basic models. These tests include direct tests (comparing the characteristics of produced and observed weather series) and indirect tests (comparing river flow statistics simulated by a daily hydrological model based on produced and observed weather series). The performance of the models in downscaling was also evaluated. To perform these tests, 15 weather stations over a diverse range of climatic conditions were used.
The results showed that the proposed methods corrected the low frequency variabilities of IWG and LARS-WG models. It also improves the performance of models in reproducing a wide range of other statistics in observed series. Another advantage of this method is its applicability in any WG (without the need for access to the source code of the model or any structural change); because this method applies to WG outputs.
Keywords: Low Frequency Variabilities, Downscaling, Climate Change, Weather Generator, Hydrological Regime، Rainfall, IWG, LARS-WG, SDSM.
