**6. Conclusions**

In this study, WRF-3DVar data assimilation experiments were conducted, in which radar reflectivity and GTS data were assimilated with the involvement of coupled hy-

drological structures of different complexity for rainfall-runoff prediction. The performance of three atmospheric-hydrological systems, established by coupling WRF with the lumped Hebei model, the grid-based Hebei model, and the fully distributed WRF-Hydro, were compared and analyzed for storms with different temporal and spatial distribution characteristics before and after data assimilation. We further explored model potentials and limitations in the localization of flood events. Focusing on the impact of data assimilation on flood forecasting after improving different types of rainfall and coupling systems of varying complexity, we found that WRF-3DVar produces more accurate rainfall forecasts, and that the assimilated model system provides higher confidence in the flood forecasts.

When the lumped model was coupled, its input rainfall was averaged over all grid points at the catchment scale, which may conceal the potential advantages of highresolution rainfall datasets. The grid-based Hebei model obtained better flood forecasting results, but it did not provide a more comprehensive description of the spatial and temporal processes of the land-surface hydrology. The WRF-Hydro system, on the other hand, is built on the basis of water balance and heat balance in terms of the physical processes, thus clearly necessary for future flood research. The main reasons for the lack of accuracy of the WRF-Hydro predictions might be the preciseness of the input meteorological elements and the structure of the modeling system. Demonstrating the former requires further exploration of the transition from multi-source data assimilation to multi-process data assimilation. The coupling system of WRF-Hydro may differ from actual regional characteristics in its representation of the rainfall-runoff mechanisms, and thus its spatial scale and applicability need to be further explored in the future, especially in relation to the high-resolution land surface and hydrological processes that are essential for flash flood forecasting. There is also a need for future work to build on the strengths of this model and tailor atmospheric-hydrological coupling systems to the study area.

**Author Contributions:** Conceptualization, W.W. and J.L.; Methodology, W.W., J.L., and F.Y.; Software, W.W. and Y.L.; Validation, W.W. and Y.L.; Formal analysis, W.W. and J.L.; Investigation, W.W., C.L., and F.Y.; Writing-original draft preparation, W.W. and J.L.; Writing-review and editing, W.W., J.L., and C.L.; Visualization, W.W.; Funding acquisition, J.L. and C.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China (51822906), the Major Science and Technology Program for Water Pollution Control and Treatment (2018ZX07110001), the National Key Research and Development Project (2017YFC1502405), and the IWHR Research & Development Support Program (WR0145B732017).

**Conflicts of Interest:** The authors declare no conflict of interest.
