In this paper, first, a distribution system state estimation (DSSE) algorithm based on Distflow model is presented. The model is designed for radial low voltage (LV) distribution grids considering traversal components such as the cable capacitances. Next, to obtain the pseudo-measurements required for the DSSE algorithm, we developed and compared three intraday nodal load forecast (INLF) methods that take; i) near real-time stream of data from the grid measurement devices (MDs), i.e., voltage and current magnitudes at limited number of lines/nodes, and ii) the batch data set from smart meters (SMs) available for previous days. The accuracy associated with the INLF methods is quantified in terms of statistical characteristics such as average of symmetric mean absolute percentage error (sMAPE) over all nodes of the distribution grid. Afterwards, the impact of this accuracy on the DSSE results is investigated using real data available for a distribution grid in city of Geneva, Switzerland.