Search

One way to adapt to and mitigate current and future water scarcity is to manage and store water more efficiently. Reservoirs act as critical buffers to ensure agricultural and municipal [...]

Extreme daily precipitation contributes to flooding that can cause significant economic damages, and so is important to properly capture in gridded meteorological datasets. This work examines precipitation extremes, the mean precipitation on wet days, and fraction of wet days in two widely used gridded datasets over the conterminous United States. Compared to the underlying station observations, the gridded data show a 27% reduction in annual 1-day maximum precipitation, 25% increase in wet day fraction, 1.5–2.5 day increase in mean wet spell length, 30% low bias in 20-yr return values of daily precipitation, and 25% decrease in mean precipitation on wet days. It is shown these changes arise primarily from the time adjustment applied to put the precipitation gauge observations into a uniform time frame, with the gridding process playing a lesser role. A new daily precipitation dataset is developed that omits the time adjustment (as well as extending the gridded data by 7 years) and is shown to perform significantly better in reproducing extreme precipitation metrics. When the new dataset is used to force a land surface model, annually averaged 1-day maximum runoff increases 38% compared to the original data, annual mean runoff increases 17%, evapotranspiration drops 2.3%, and fewer wet days leads to a 3.3% increase in estimated solar insolation. These changes are large enough to affect portrayals of flood risk and water balance components important for ecological and climate change applications across the CONUS. [ABSTRACT FROM AUTHOR]

Physically based models facilitate understanding of seasonal snow processes but require meteorological forcing data beyond air temperature and precipitation (e.g., wind, humidity, shortwave radiation, and longwave radiation) that are typically unavailable at automatic weather stations (AWSs) and instead are often represented with empirical estimates. Research is needed to understand which forcings (after temperature and precipitation) would most benefit snow modeling through expanded observation or improved estimation techniques. Here, the impact of forcing data availability on snow model output is assessed with data-withholding experiments using 3-yr datasets at well-instrumented sites in four climates. The interplay between forcing availability and model complexity is examined among the Utah Energy Balance (UEB), the Distributed Hydrology Soil Vegetation Model (DHSVM) snow submodel, and the snow thermal model (SNTHERM). Sixty-four unique forcing scenarios were evaluated, with different assumptions regarding availability of hourly meteorological observations at each site. Modeled snow water equivalent (SWE) and snow surface temperature Tsurf diverged most often because of availability of longwave radiation, which is the least frequently measured forcing in cold regions in the western United States. Availability of longwave radiation (i.e., observed vs empirically estimated) caused maximum SWE differences up to 234 mm (57% of peak SWE), mean differences up to 6.2°C in Tsurf, and up to 32 days difference in snow disappearance timing. From a model data perspective, more common observations of longwave radiation at AWSs could benefit snow model development and applications, but other aspects (e.g., costs, site access, and maintenance) need consideration. [ABSTRACT FROM AUTHOR]

The current California drought has cast a heavy burden on statewide agriculture and water resources, further exacerbated by concurrent extreme high temperatures. Furthermore, industrial-era global radiative forcing brings into question the role of long-term climate change with regard to California drought. How has human-induced climate change affected California drought risk? Here, observations and model experimentation are applied to characterize this drought employing metrics that synthesize drought duration, cumulative precipitation deficit, and soil moisture depletion. The model simulations show that increases in radiative forcing since the late nineteenth century induce both increased annual precipitation and increased surface temperature over California, consistent with prior model studies and with observed long-term change. As a result, there is no material difference in the frequency of droughts defined using bivariate indicators of precipitation and near-surface (10 cm) soil moisture, because shallow soil moisture responds most sensitively to increased evaporation driven by warming, which compensates the increase in the precipitation. However, when using soil moisture within a deep root zone layer (1 m) as covariate, droughts become less frequent because deep soil moisture responds most sensitively to increased precipitation. The results illustrate the different land surface responses to anthropogenic forcing that are relevant for near-surface moisture exchange and for root zone moisture availability. The latter is especially relevant for agricultural impacts as the deep layer dictates moisture availability for plants, trees, and many crops. The results thus indicate that the net effect of climate change has made agricultural drought less likely and that the current severe impacts of drought on California's agriculture have not been substantially caused by long-term climate changes. [ABSTRACT FROM AUTHOR]

Continental-domain assessments of climate change impacts on water resources typically rely on statistically downscaled climate model outputs to force hydrologic models at a finer spatial resolution. This study examines the effects of four statistical downscaling methods [bias-corrected constructed analog (BCCA), bias-corrected spatial disaggregation applied at daily (BCSDd) and monthly scales (BCSDm), and asynchronous regression (AR)] on retrospective hydrologic simulations using three hydrologic models with their default parameters (the Community Land Model, version 4.0; the Variable Infiltration Capacity model, version 4.1.2; and the Precipitation-Runoff Modeling System, version 3.0.4) over the contiguous United States (CONUS). Biases of hydrologic simulations forced by statistically downscaled climate data relative to the simulation with observation-based gridded data are presented. Each statistical downscaling method produces different meteorological portrayals including precipitation amount, wet-day frequency, and the energy input (i.e., shortwave radiation), and their interplay affects estimations of precipitation partitioning between evapotranspiration and runoff, extreme runoff, and hydrologic states (i.e., snow and soil moisture). The analyses show that BCCA underestimates annual precipitation by as much as −250 mm, leading to unreasonable hydrologic portrayals over the CONUS for all models. Although the other three statistical downscaling methods produce a comparable precipitation bias ranging from −10 to 8 mm across the CONUS, BCSDd severely overestimates the wet-day fraction by up to 0.25, leading to different precipitation partitioning compared to the simulations with other downscaled data. Overall, the choice of downscaling method contributes to less spread in runoff estimates (by a factor of 1.5-3) than the choice of hydrologic model with use of the default parameters if BCCA is excluded. [ABSTRACT FROM AUTHOR]