The Surface Water and Ocean Topography (SWOT) mission provides an unparalleled observation system for monitoring global surface water resources. We compared different SWOT level-2 high-rate products and in situ data for rivers, lakes and reservoirs in Australia, utilising quality flags and uncertainty indicators present in these rich data products. Water heights derived from the raster product have a weighted root-mean-square error of $\sim$5cm but the product fails to sample small water bodies. The use of LakeSP and RiverSP spatial definitions of water bodies yields accuracies typically between 20 and 30 cm but often do not include data for Australian water bodies and/or small river sections. Our novel approach using pixel cloud data achieves an accuracy of $\sim$5cm in measuring water heights over rivers as narrow as $\sim$40m wide and reservoirs as small as $\sim$100$\times$100m, well below the mission requirements of 100 m river width and 250$\times$250m lake area.