In this study, we present the improved Rangeland Hydrology and Erosion Model (RHEM V2.3), a process‐based erosion prediction tool specific for rangeland application. The article provides the mathematical formulation of the model and parameter estimation equations. Model performance is assessed against data collected from 23 runoff and sediment events in a shrub‐dominated semiarid watershed in Arizona, USA. To evaluate the model, two sets of primary model parameters were determined using the RHEM V2.3 and RHEM V1.0 parameter estimation equations. Testing of the parameters indicated that RHEM V2.3 parameter estimation equations provided a 76% improvement over RHEM V1.0 parameter estimation equations. Second, the RHEM V2.3 model was calibrated to measurements from the watershed. The parameters estimated by the new equations were within the lowest and highest values of the calibrated parameter set. These results suggest that the new parameter estimation equations can be applied for this environment to predict sediment yield at the hillslope scale. Furthermore, we also applied the RHEM V2.3 to demonstrate the response of the model as a function of foliar cover and ground cover for 124 data points across Arizona and New Mexico. The dependence of average sediment yield on surface ground cover was moderately stronger than that on foliar cover. These results demonstrate that RHEM V2.3 predicts runoff volume, peak runoff, and sediment yield with sufficient accuracy for broad application to assess and manage rangeland systems.

As the atmosphere gets warmer, rainfall intensification is expected across the planet with anticipated impacts on ecological and human systems. In the southwestern United States and northwestern Mexico, the highly variable and localized nature of rainfall during the North American Monsoon makes it difficult to detect temporal changes in rainfall intensities in response to climatic change. This study addresses this challenge by using the dense, subdaily, and daily observations from 59 rain gauges located in southeastern Arizona. We find an intensification of monsoon subdaily rainfall intensities starting in the mid‐1970s that has not been observed in previous studies or simulated with high‐resolution climate models. Our results highlight the need for long‐term, high spatiotemporal observations to detect environmental responses to a changing climate in highly variable environments and show that analyses based on limited observations or gridded data sets fail to capture temporal changes potentially leading to erroneous conclusions.

The 14 September 2015 Hildale, Utah, storm resulted in 20 flash flood fatalities, making it the most deadly natural disaster in Utah history; it is the quintessential example of the “paroxysmal precipitation of the desert”. The measured peak discharge from Maxwell Canyon at a drainage area of 5.3 km² was 266 m³/s, a value that exceeds envelope curve peaks for Utah. The 14 September 2015 flash flood reflects features common to other major flash flood events in the region, as well as unique features. The flood was produced by a hailstorm that was moving rapidly from southwest to northeast and intensified as it interacted with complex terrain. Polarimetric radar observations show that the storm exhibited striking temporal variability, with the Maxwell Canyon tributary of Short Creek and a small portion of the East Fork Virgin River basin experiencing extreme precipitation. Periods of extreme rainfall rates for the 14 September 2015 storm are characterized by K_DP signatures of extreme rainfall in polarimetric radar measurements. Similar K_DP signatures characterized multiple storms that have produced record and near‐record flood peaks in Colorado Plateau watersheds. The climatology of monsoon thunderstorms that produce flash floods exhibits striking spatial heterogeneities in storm occurrence and motion. The hydroclimatology of flash flooding in arid/semiarid watersheds of the southwestern United States exhibits relatively weak dependence on drainage basin area. Large flood peaks over a broad range of basin scales can be produced by small thunderstorms like the 14 September 2015 Hildale Storm, which pass close to the outlet.

The U.S. Department of Agriculture‐Agricultural Research Service's (ARS) Experimental Watershed Network grew from Dust Bowl era efforts of the Soil Conservation Service in the mid‐1930s with the establishment of small experimental watersheds. In the 1950s, five watershed research centers with intensively instrumented watersheds at the scale of 100 to 700 km² were established. Primary network research objectives were to quantify on‐site and downstream effects of conservation practices and develop rainfall‐runoff relationships for design of water conservation structures. With passage of the Clean Water Act in 1972, research objectives have evolved to add a variety of observations relevant to the water quality issues. Many of the watersheds within the network have served, and continue to serve, as core validation sites for satellite sensors. As a result of the network's long history and intensive monitoring, coupled with mission‐driven research, a deep knowledge base of watershed processes has been developed. This has led to the extensive development and validation of numerous watershed models that are in widespread use today. The visionary investments in building and maintaining this network and associated scientific investigations for more than half a century have not only resulted in numerous high‐impact research accomplishments but also a wide array of accomplishments that directly benefit society. The ARS Experimental Watersheds formed the core of the Conservation Effects Assessment Project (CEAP) as well as the recently established Long‐Term Agroecosystem Research (LTAR) network. LTAR will expand the mission of the ARS Watersheds Network to include agricultural intensification, maintaining or improving ecosystem services while enhancing rural prosperity.