A Multilayer Perceptron Model for Stochastic Synthesis

Time series analysis is a major mathematical tool in hydrology, with the moving average being most popular model type for this purpose due to its simplicity. During last 20 years, various studies have focused on an important statistical characteristic, namely long-term persistence and simultaneous consistency at all timescales, when different timescales are involved simulation. Though these issues been successfully addressed by researchers, solutions that suggested mathematically advanced, which poses challenge regarding their adoption practitioners. In study, multilayer perceptron network used obtain synthetic daily values of rainfall. order develop model, first, appropriate set features was selected, then, custom cost function crafted preserve properties time series. This approach applied two locations climatic conditions long record measurements (more than 100 years first more 40 second). The results indicate methodology capable preserving characteristics. advantage that, once it has trained, straightforward apply can be modified easily analyze other types hydrologic