Solar generation resources are being deployed at an unprecedented rate in modern power grids. The unpredictability of solar energy demands systematic investigation of impact of solar penetration on power system operation and planning. Modern power grids have tackled solar uncertainty by building solar farms in locations with consistent solar irradiance, and less successfully by solar forecasting. Solar forecasting has remained a challenging process in modern power grids. This paper aims to understand the impact that solar forecasting can have on power system operation and planning. A variety of probabilistic models that can estimate solar irradiance and identify their performance using real-world data are investigated. Results presented in this paper show that the heavily used normal or log-normal distributions are not suited for solar irradiance prediction. These probabilistic estimates are utilized to compute probabilistic power flows (PPF). PPF result show how sensitive calculated line flows are to forecast method, in particular where lines are closer to photovoltaic installations. Also by using the PPF of accurate forecast, we identify the conditions under which the modern power grid can operate reliability or in need of mitigations.