Testing the Hydrological Coherence of High‐Resolution Gridded Precipitation and Temperature Data Sets

Abstract

Assessing the accuracy of gridded climate data sets is highly relevant to climate change impact studies, since evaluation, bias correction, and statistical downscaling of climate models commonly use these products as reference. Among all impact studies those addressing hydrological fluxes are the most affected by errors and biases plaguing these data. This paper introduces a framework, coined Hydrological Coherence Test (HyCoT), for assessing the hydrological coherence of gridded data sets with hydrological observations. HyCoT provides a framework for excluding meteorological forcing data sets not complying with observations, as function of the particular goal at hand. The proposed methodology allows falsifying the hypothesis that a given data set is coherent with hydrological observations on the basis of the performance of hydrological modeling measured by a metric selected by the modeler. HyCoT is demonstrated in the Adige catchment (southeastern Alps, Italy) for streamflow analysis, using a distributed hydrological model. The comparison covers the period 1989–2008 and includes five gridded daily meteorological data sets: E‐OBS, MSWEP, MESAN, APGD, and ADIGE. The analysis highlights that APGD and ADIGE, the data sets with highest effective resolution, display similar spatiotemporal precipitation patterns and produce the largest hydrological efficiency indices. Lower performances are observed for E‐OBS, MESAN, and MSWEP, especially in small catchments. HyCoT reveals deficiencies in the representation of spatiotemporal patterns of gridded climate data sets, which cannot be corrected by simply rescaling the meteorological forcing fields, as often done in bias correction of climate model outputs. We recommend this framework to assess the hydrological coherence of gridded data sets to be used in large‐scale hydroclimatic studies.

Identification of factors influencing hydrologic model performance using a top‐down approach in a large number of U.S. catchments

“…Although Atkinson et al () relied on a visual assessment for determining which models had an “acceptable performance,” it was necessary to define a threshold for automatizing this task in the present study. The choice of a threshold will always be subjective (Diskin & Simon, ), but it should be goal‐oriented and take into account the available data (Guthke, ; Laiti et al, ). As this study aimed at comparing model performance at different timescales and for different model complexities, the threshold needs to take into account the abilities of the considered models at different timescales.…”

TIER version 1.0: an open-source Topographically InformEd Regression (TIER) model to estimate spatial meteorological fields

“…Given the ability to perform parameter sensitivity experiments in TIER, we reemphasize the need for novel evaluation methods including out-of-sample station networks (e.g., Daly, 2006;Daly et al, 2017;Newman et al, 2019) that are as independent from the input networks as possible and integrated validation methods using ancillary observations such as streamflow and other modeling tools such as hydrologic models (Beck et al, 2017;Henn et al, 2018;Laiti et al, 2018).…”

“…Gridded near-surface meteorological products (specifically precipitation and temperature) are a foundational product for many applications including weather and climate model validation, hydrologic modeling, climate model downscal-ing, among others (Day, 1985;Franklin, 1995;USBR, 2012;Pierce et al, 2014;Liu et al, 2017). It is often challenging to develop realistic estimates of these variables, particularly when complex terrain or large spatial climate gradients are present in the domain of interest.…”

TIER Version 1.0: An open-source Topographically InformEd Regression (TIER) model to estimate spatial meteorological fields

“…This again emphasizes the need for novel evaluation methods including true out of sample station networks (e.g. Daly 2006;Daly et al 2017;Newman et al 2019) and integrated validation methods using ancillary observations such as streamflow and other modeling tools such as hydrologic models (Beck et al 2017;Henn et al 2018;Laiti et al 2018).…”