Conventional Probable Maximum Precipitation (PMP) methods face several limitations, including the lack of statistical uncertainty characterization, subjectivity in storm maximization, and the assumption of a stationary climate. To address these limitations, we propose a nonstationary PMP approach that combines the novel stochastic spatiotemporal rainfall generation model StormLab with a nonstationary Generalized Extreme Value (GEV) model. We applied the new approach to the Upper Red River Basin in the south-central United States. StormLab provided 10,000 years of high-resolution (6-hour, 0.03°) precipitation fields from 1901 to 2100, based on 50 ensembles of a global climate model (GCM). A nonstationary GEV model was fitted to the simulated precipitation annual maxima, providing PMP estimates under different climate periods with an associated annual exceedance probability (AEP). The simulated precipitation was then integrated with a hydrologic model to generate annual peak discharge in major tributaries and to estimate the probable maximum flood (PMF). Our approach produces PMP estimates for areas ranging from 10-20,000 mi2 and durations from 6 to 360 hours. Results show a 16-25% increase in PMP with an AEP of 10-4 from 2020 to 2100 at different spatial and temporal scales. Higher increases of 35% and 37% are projected in PMF with the same AEP in two major tributaries. The PMP and PMF results were further compared with previous PMP/PMF estimates. This study demonstrates the value of utilizing stochastic rainfall models and GCM large ensembles to inform PMP and PMF analysis in a changing climate.

In global applications and data sparse regions, which comprise most of the earth, hydrologic model-based flood monitoring relies on precipitation data from satellite multisensor precipitation products or numerical weather forecasts. However, these products often exhibit substantial errors during the meteorological conditions that lead to flooding, including extreme rainfall. The propagation of precipitation forcing errors to predicted runoff and streamflow is scale-dependent and requires an understanding of the autocorrelation structure of precipitation errors, since error autocorrelation impacts the accumulation of precipitation errors over space and time in hydrologic models. Previous efforts to account for satellite precipitation uncertainty in hydrologic models have demonstrated the potential for improving streamflow estimates; however, these efforts use satellite precipitation error models that rely heavily on ground reference data such as rain gages or weather radar and do not characterize the nonstationarity of precipitation error autocorrelation structures. This work evaluates a new method, the Space-Time Rainfall Error and Autocorrelation Model (STREAM), which stochastically generates possible true precipitation fields, as input to the Hillslope Link Model to generate ensemble streamflow estimates. Unlike previous error models, STREAM represents the nonstationary and anisotropic autocorrelation structure of satellite 2 precipitation error and does not use any ground reference to do so. Ensemble streamflow predictions are compared with streamflow generated using satellite precipitation fields as well as a radar-gage precipitation dataset during peak flow events. Results demonstrate that this approach to accounting for precipitation uncertainty effectively characterizes the uncertainty in streamflow estimates and reduces the error of predicted streamflow. Streamflow ensembles forced by STREAM improve streamflow prediction nearly to the level obtained using ground-reference forcing data across basin sizes.

In this study, the authors explore simple concepts of persistence in streamflow forecasting based on the real-time streamflow observations. The authors use 15-minute streamflow observations from the year 2002 to 2018 at 140 U.S. Geological Survey (USGS) streamflow gauges monitoring the streams and rivers over the State of Iowa. The spatial scale of the basins ranges from about 7 km2 to 37,000 km2. Motivated by the need for evaluating the skill of real-time streamflow forecasting systems, the authors perform quantitative skill assessment of different persistence schemes across spatial scales and lead-times. They show that temporal persistence forecasts skill has strong dependence on basin size and weaker, but non-negligible, dependence on geometric properties of the river network of the basin. The authors show that anomaly persistence forecasting can serve as a good reference for the evaluation of real-time streamflow forecasts at scales of order 100 km2. Building on results from this temporal persistence, they extend the streamflow persistence to space through flow-connected river network. It simply assumes that streamflow at a station in space will persist to another station which is flow-connected, and refer to it as pure spatial persistence forecasts (PSPF). The authors show that skill of PSPF derived streamflow forecasts is strongly dependent on basin area-ratio and lead-times, and weakly related to the downstream flow distance between stations. They show that the skill depicted in terms of Kling-Gupta efficiency (KGE) > 0.5 can be achieved for basin area ratio > 0.5 and lead-time up to three days. Adding complexities of hydrologic routing and rainfall QPF to the PSPF further improves the skill. The authors discuss the implications of their findings for improvements of rainfall-runoff models as well as data assimilation schemes.

Floods claim a high toll in fatalities and economic impacts. Despite their societal relevance, there is much more to learn about the projected changes in discharge and flooding. Here we force an operational hydrologic model over the state of Iowa with high-resolution convection-permitting climate-model precipitation to evaluate the response of 140 watersheds to climate change. At the end of the century, under the most aggressive scenario in terms of fossil fuel use, we show that the transition from snow to rainfall and a ~30% increase in extreme precipitation rates lead to a doubling of maximum discharge during the spring and extended the flood season into the fall. Total discharge volumes are also expected to increase. Our results suggest that flood projections based on extreme precipitation increases alone substantially underestimate future risk due to the nonlinearity of the hydrologic response explained by long-term soil moisture memory and its feedbacks with precipitation.