The reliability of urban flood modelling can be largely improved if high-accuracy and fine-resolution rainfall estimates are available; however, this requires a very dense network of rainfall sensors and is usually not available due to limited budget and space. Adjustment and downscaling techniques are largely used respectively to post process the radar and rain gauge data to obtain better rainfall estimates in terms of accuracy and resolution. However, the combined application of these two types of techniques was seldom discussed in literatures, and its impact on the subsequent hydraulic modelling is unknown. This work implements a combined procedure of stochastic downscaling and gauge-based adjustment, aiming to evaluate its applicability to urban pluvial flood modelling. Unlike the adjustment process that reduces the rainfall input uncertainty (due to mean bias) through merging rainfall information from different sensors, the stochastic process of generating street-scale rainfall estimates actually causes additional uncertainty (due to spatial variability). This additional uncertainty will further propagate through hydraulic modelling and consequently affect the reliability of the resulting hydraulic outputs. The result of case study suggests that the uncertainty caused by the downscaling process could be larger than that reduced by the adjustment as the drainage area is very small.