Identification and uncertainty estimation of vertical reflectivity profiles using a Lagrangian approach to support quantitative precipitation measurements by weather radar and its impact on the catchment hydrological response during cold season precipitation

Radars are known for their ability to obtain a wealth of information about the spatial and temporal characteristics of rainfall fields. Unfortunately, precipitation estimates obtained by weather radar are known to be affected by multiple sources of error, which have to be corrected for in order to provide realistic rainfall estimates. A new Lagrangian approach is presented to estimate the vertical profile of reflectivity (VPR) from volumetric weather radar data. Contrary to previous contributions, our newly proposed approach focuses specifically on individual precipitation regions as observed by the radar. For this purpose, we apply the recently developed Rotational Carpenter Square Cluster Algorithm (RoCaSCA), which is able to identify and discriminate between precipitation regions at different reflectivity levels. For each region, pixel belonging to either stratiform or non-convective type of precipitation are VPR-corrected using a new piecewise linear correction approach. For the winter half-year of study, results show that in case all sources of error associated with weather radar rainfall estimation are properly accounted for, precipitation estimates are found to be of similar quality as those obtained from in situ rain gauge measurements up to distances of 150 km. Even without bias correction or other forms of radar-gauge adjustment. It is recognized that after correcting for errors considerable differences between the measurements of both devices remain, either originating from unaccounted error sources, small-scale precipitation variability, or from scale issues when comparing weather radar and rain gauge measurements. However, instead of applying a bias correction procedure to account for these differences, a novel approach is presented which is able to account for the impact of VPR uncertainty on the estimated radar rainfall variability. Once the uncertainty originating from VPR variability is taken into account, a large part of the difference between weather radar and rain gauge measurements can be accounted for. Overall the total uncertainty in the weather radar estimates due to VPR variability is shown to be about 40%. The possibilities of using error-corrected weather radar precipitation data are used to simulate the hydrological response of a medium-sized catchment using a semi-distributed model. Next, the impact of uncertainty in the precipitation input data originating from VPR variability is compared to impact of uncertainty in hydrological model parameter values. Results show that using the corrected radar data the observed discharge response is simulated well. However, when focusing on the impact of uncertainty on the discharge estimates, results show that hydrological model parameter variability has a much larger effect than precipitation uncertainty. The latter only adds about 20% of uncertainty, respectively. This shows the effectiveness of catchments of this size to filter the large uncertainty of the intermittent input signal.