<![CDATA[The limited availability of daily data is a major issue when working with weather datasets. It can lead to discontinuities in the data history that impact the accuracy of the model output, which is a function of the observation input. This study presents a simple technique for generating daily weather data from monthly data using synoptic observations from the North Minahasa Climatology Station (1989-2014). Our approach involves the use of logit, Fourier, and gamma functions. The generated rainfall exhibits a similar pattern to the observed rainfall, and the statistical test indicates no significant difference (p<0.000). However, the resulting correlation is low (0.33-0.49). Based on the seasonal division, both the generated and observed rainfall values are high in the December[1]January-February (DJF) season, followed by a decrease in the March-April-May (MAM) and June-July-August (JJA) and an increase in the September-October-November (SON) season. Additionally, there are discrepancies in rainfall generation compared to observations due to the use of uniform distribution random numbers, which tend to overlook temporal specificity. Moreover, other meteorological factors, such as temperature and relative humidity, generate daily values from monthly data that closely resemble observations, with a correlation coefficient greater than 0.8. This is due to the Fourier function’s lack of a field variability factor.]]>
Copyrights © 2023
