Resumen

CAMPOS-ARANDA, Daniel Francisco. Best PDF in 19 large annual series of MDP from the San Luis Potosi state, Mexico. Tecnol. cienc. agua [online]. 2019, vol.10, n.5, pp.34-74.  Epub 15-Feb-2020. ISSN 2007-2422.  https://doi.org/10.24850/j-tyca-2019-05-02.

All hydraulic works are planned and designed based on Floods Design. Without hydrometric information, these predictions are estimated using hydrological methods that yield the sought flows by means of design rainfalls. Design rainfall is estimated based on pluviometer records of annual maximum daily precipitation (MDP) due to the shortage of pluviographs. The probabilistic analysis of the annual MDP series is identical to that of the floods; however, neither adequate probability distribution functions (PDFs) nor those that should be applied by precept have been defined so far, hence the need to try several. First, the best PDF was searched for using the L-ratio diagram, which includes six models with three fit parameters. An objective selection is made by using the weighted absolute distance, in the 19 annual MDP records with more than 50 data from the state of San Luis Potosi, Mexico. Then eight descriptive ability (DA) indexes are described and applied to the eight PDFs that were compared, in each of the 19 PMD records. The results are concentrated and analyzed for geographic areas of the state: Potosino Plateau and Middle Zone. Results show that Wakeby PDF is a model having high DA and for that reason, its application is suggested as precept. The two best PDF options are also highlighted in each of the 19 records processed, according to the eight DA indexes. Finally, a comparison of predictions with periods of return of 50, 100, 500 and 1000 years is carried out to explore shallowly the predictive ability of the PDFs found as best options. In each registry four PDFs are applied, the one obtained according to the L-ratio diagram; the two best PDFs according to the eight DA indices and the Wakeby distribution. It is concluded that the use of the L-ratio diagram and the application of the eight DA indexes are adequate and lead to a good approximation, since it was not difficult to select the adopted predictions, besides the similarity of the predictions calculated in each register promotes confidence in such estimations.

Palabras llave : L-ratio diagram; standard error of fit; relative standard error of fit; mean absolute error; maximum absolute error; Akaike information criterion; Q-Q correlation coefficient; concordance indexes and predictive ability.

        · resumen en Español     · texto en Español | Inglés     · Español ( pdf ) | Inglés ( pdf )