Human amplified changes in precipitation-runoff patterns in large river basins of the Midwestern United States

Complete transformations of land cover from prairie, wetlands, and hardwood forests to homogenous row crop agriculture scattered with urban centers are thought to have caused profound changes in hydrology in the Upper Midwestern US since the 1800s. Continued intensification of land use and drainage practices combined with increased precipitation have caused many Midwest watersheds to exhibit higher streamflows today than in the historical past. While changes in crop type and farming practices have been well documented over the past few decades, changes in artificial surface (ditch) and subsurface (tile) drainage systems have not. This makes it difficult to quantitatively disentangle the effects of climate change and artificial drainage intensification on the observed hydrologic change, often spurring controversial interpretations with significant implications for management actions. In this study, we investigate four large (23,000–69,000 km2) Midwest river basins that span climate and land use gradients to understand how climate and agricultural drainage have influenced basin hydrology over the last 79 years. We use daily, monthly, and annual flow metrics to document streamflow changes and discuss those changes in the context of climate and land use change. While we detect similar timing of precipitation and streamflow changes in each basin, overall the magnitude and significance of precipitation changes are much less than we detect for streamflows. Of the basins containing greater than 20 % area drained by tile and ditches, we observe 2 to 4 fold increases in low flows and 1.5 to 3 fold increases in high and extreme flows. Monthly precipitation has increased slightly for some months in each basin, mostly in fall and winter months (August – March), but total monthly streamflow has increased in all months for the Minnesota River Basin (MRB), every month but April for the Red River Basin (RRB), September-December and March in the Illinois River Basin (IRB), and no months in the Chippewa River basin (CRB). Using a water budget, we determined that the soil moisture/groundwater storage term for the intensively drained and cultivated MRB, IRB, and RRB, has decreased by about 200 %, 100 %, and 30 %, respectively while increased by roughly 30 % in the largely forested CRB since 1975. We argue that agricultural land use change, through wetland removal and artificial drainage installation, has decreased watershed storage and amplified the streamflow response to precipitation increases in the Midwest. Highly managed basins with large reservoirs and urban centers, such as the Illinois River basin (IRB), may be able to buffer some of these impacts better than largely unregulated systems such as the Minnesota River (MRB) and Red River of the North (RRB) basins. The reported streamflow increases in the MRB, IRB, and RRB are large (18 %–318 %), and should have important implications for channel adjustment and sediment and nutrient transport. Acknowledging both economic benefits and apparent detrimental impacts of artificial drainage on river flows, sediments, and nutrients, we question whether any other human activity has comparably altered critical zone activities, while remaining largely unregulated and undocumented. We argue that better documentation of existing and future drain tile and ditch installation is greatly needed.

Hydroinformatics advances for operational river forecasting: using graphs for drainage network descriptions

Distributed hydrologic models provide accurate river streamflow forecasts and a multitude of spatially varied products on basin scales. The distributed elements of the basins are pieced together using drainage networks. An efficient representation of drainage networks in computer code is necessary. Graph theory has long been applied in many engineering areas to solve network problems. In this paper we demonstrate that adjacent list graph is the most efficient way of presenting the drainage network in terms of development and execution. The authors have implemented drainage networks using the adjacency-list structure in both the research and operational versions of the US National Weather Service (NWS) distributed model. A parallel routing algorithm based on Dijsktra's shortest path algorithm was also developed using the MPI library, which was tested on a cluster using the Oklahoma Illinois River basin dataset. Theoretical analysis and test results show that inter-processor communication and unbalanced workload among the processors limit the scalability of the parallel algorithm. The parallel algorithm is more applicable to computers with high inter-processor bandwidth, and to basins where the number of grid cells is large and the maximum distance of the grid cells to the outlet is short.