CHOSEN: A synthesis of hydrometeorological data from intensively monitored catchments and comparative analysis of hydrologic extremes

Abstract

Comparative hydrology has been hampered by limited availability of geographically extensive, intercompatible monitoring data on comprehensive water balance stores and fluxes. These limitations have, for example, restricted comprehensive assessment of multiple dimensions of wetting and drying related to climate change and hampered understanding of why widespread changes in precipitation extremes are uncorrelated with changes in streamflow extremes. Here, we address this knowledge gap and underlying data gap by developing a new data synthesis product and using that product to detect trends in the frequencies and magnitudes of a comprehensive set of hydroclimatic and hydrologic extremes. CHOSEN (Comprehensive Hydrologic Observatory Sensor Network) is a database of streamflow, soil moisture, and other hydroclimatic and hydrologic variables from 30 study areas across the United States. An accompanying data pipeline provides a reproducible, semi-automated approach for assimilating data from multiple sources, performing quality assurance and control, gap-filling and writing to a standard format. Based on the analysis of extreme events in the CHOSEN dataset, we detected hotspots, characterized by unusually large proportions of monitored variables exhibiting trends, in the Pacific Northwest, New England, Florida and Alaska. Extreme streamflow wetting and drying trends exhibited regional coherence. Drying trends in the Pacific Northwest and Southeast were often associated with trends in soil moisture and precipitation (Pacific Northwest) and evapotranspiration-related variables (Southeast). In contrast, wetting trends in the upper Midwest and the Rocky Mountains showed few univariate associations with other hydroclimatic extremes, but their latitudes and elevations suggested the importance of changing snowmelt characteristics. On the whole, observed trends are incompatible with a 'drying-in-dry, wetting-in-wet' paradigm for climate-induced hydrologic changes over land. Our analysis underscores the need for more extensive, longer-term observational data for soil moisture, snow and evapotranspiration.

HydroBench: Jupyter supported reproducible hydrological model benchmarking and diagnostic tool

“…In addition to collaboration on code developments, open-source hydrological data are also critical for communitywide model benchmarking, as they enable modelers to test their hypotheses beyond local watersheds and over a broad range of time against consistent information. Examples of large-sample open-source data in hydrology include the MOPEX, CAMELS, EMDNA, and CHOSEN datasets (Duan et al, 2006;Addor et al, 2017;Tang et al, 2021;Zhang et al, 2021).…”

Inferring causal associations in hydrological systems: A comparison of methods

“…However, it is often unfeasible or unethical in the real world, such as operating and intervening in regional or global hydrological variables. Additionally, with the development of monitoring technology, several experimental catchments have been established in recent years (Zhang et al, 2021b). However, the construction and maintenance are costly, and due to the spatial heterogeneity and scale problems of hydrological systems (Bergström and Graham, 1998;Blöschl et al, 2019), the extensibility and upscaling of discovered mechanisms are also challenging.…”