Development of Variogram Models Retrieved from PERSIANN Family and TRMM 3B47 V. 7 Satellite-Derived Datasets in Fars Province

<p style="text-align: left;"><strong>Introduction:</strong> Precipitation data plays a crucial role in hydrological models, and it is important to have a good understanding of its spatial and temporal distribution before incorporating it into these models. Access to sufficient statistics on precipitation events is necessary to address this issue. However, due to the cost and limited availability of ground-based rain monitoring statistics in various locations, satellite-derived datasets can be a highly effective alternative.</p> <p style="text-align: left;"><strong>Methods:</strong> In the current study four satellite-derived datasets (PERSIANN, PERSIANN-CDR, PERSIANN-CCS, and TRMM 3B43 V.7) were compared to assess and enhance the variogram curves of average annual precipitation. Ground-based observations from 23 stations in the area were utilized to evaluate the datasets.</p> <p style="text-align: left;"><strong>Findings:</strong> The regression coefficient of the employed PERSIANN and TRMM families' satellite-derived datasets with ground-based observations were found to be 0.35 and 0.65, respectively. These datasets were found to be anisotropic, meaning that their characteristics vary directionally, and the variogram curves obtained from them were unbounded. These factors make their use challenging in most hydrological applications. To mitigate these issues, the trend of 1st or 2nd order polynomials was removed from the datasets in order to make them isotropic and separate the non-random component. After trend removal, the resulting two datasets prepared based on PERSIANN-CCS and TRMM 3B43 V.7 exhibited acceptable characteristics and isotropy. The bound indices of the variograms reached approximately 0.85 and 0.31, respectively. Among various models of theoretical variogram, the Gaussian model was selected as the most suitable model to express the variogram of the satellite-derived precipitation datasets.</p>