Drought is a natural threat to human societies that is not restricted to specific areas and affects many people around the world. Drought is a gradual threat that roots from a shortage of rainfall that is expected, and when the water shortage lasts more than a season, meeting the human and environmental demands become a challenging issue. Given the importance of drought as well as its spatial and temporal range, drought characteristics (start, end, duration and intensity) can be evaluated. In this study, the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR) is used and compared with measured data to derive drought indices. Two single variable indicators, including standardized precipitation index, (SPI) standard runoff index (SRI) and a combined multivariate index (NMSDI), were developed in order to propose the most effective indices according to their performance. PERSIANN-CDR precipitation satellite data were used to develop the SPI index. The SRI Index was also developed from the runoff resulted from the rainfall-runoff model (input for the PERSIANN-CDR precipitation data model). To study and consider the various hydrological characteristics, rainfall and generated runoff variables were integrated to generate a combined multivariate drought index (NMSDI). In this research, the standardization procedure is recommended with the help of non-parametric functions to introduce a new compatible index. The semi-distributed model of SWAT in this study was used to simulate the outflow runoff from Cheljord sub-basin in Zayanderoud basin, Iran. The studied area has a high-altitude difference at low distances. This difference caused a variety of simultaneous precipitation quantity in different parts of the basin. In this situation, due to scatter and unevenly distributed rain gauge stations, satellite data was used for runoff production. The coefficient of NS = 0.78 obtained from comparison of simulated and measured runoff which represents the well-behaved performance of the model. The performance of the single-variable and multivariate indices was studied and evaluated. The results showed that the SPI index developed using PERSIANN-CDR data accurately detected meteorological droughts with an average probability of 75% at the threshold of -0.5. The NMSDI's multivariate index also showed a threshold of -0.5 drought with an average probability of 84%. The comparison of probabilities of SPI standardized rainfall indices, SRI standardized runoff and NMSDI combined index, indicated that for the diagnosis of the drought by using satellite data PERSIANN-CDR, the combination of the precipitation and runoff modeling had a significant impact on drought detection in comparison to the single-variable index for drought analysis alone. SPI and NMSDI in the early detection of drought compared to the SRI index which i an index of hydrological drought is assessed and the results showed that the SPI and NMSDI indices could determine the beginning of the drought with a probability of 27.27% and 42.85% ahead of the SRI index, respectively. Keywords : Rainfall-runoff simulation, SWAT Hydrological Model, Evaluation of drought characteristics, Early Drought Analysis, Multivariate Index NMSDI, SPI, SRI, PERSIANN-CDR