Effect of disjunct sampling on calibration of design wind speed

Three main tasks consist of calibrating design wind speed. These include obtaining high quality extreme wind speed records, selecting the most appropriate probabilistic model, and applying the best fitting method to estimate the distribution parameters. The last two aspects have critical impact on the calibrated design wind speeds and have been intensively studied periodically. However, data quality is still the primary factor that affects the reliability of the calibrated design wind speed. The data is typically screened and processed before carrying out extreme value analysis. This includes removing erroneous data and conducting standardization. Besides quality control and standardization, the total number of years of data and the sampling intervals of data in each recording day also need to be considered. The former concern, also known as sample size effect, is well acknowledged and revisited in a recent study. The later concern draws relatively few attentions. One recent study investigated the later concern by using some wind speed observations at ground level. As the sample size is limited, the effect of sample size and disjunct sampling could be mixed in their observation. Moreover, appreciation of the disjunct sampling from sampling theory was not investigated in their study. To better appreciate the impact of disjunct sampling on design wind speed calibration alone, the effect of sample size needs to be eliminated. This requires very long period wind speed records that are not available for historical records. Therefore, this study employs Monte Carlo technique to simulate long term wind speed samples. The simulated wind speeds are used to investigate the impact of disjunct sampling on the design wind speed and the combined effect of disjunct sampling and limited sample size on design wind speed. Effect of disjunct sampling appears in part of the period on the estimated return period wind speeds is also investigated.