Regionalization of land‐use impacts on streamflow using a network of paired catchments

Abstract

Quantifying the impact of land use and cover (LUC) change on catchment hydrological response is essential for land‐use planning and management. Yet hydrologists are often not able to present consistent and reliable evidence to support such decision‐making. The issue tends to be twofold: a scarcity of relevant observations, and the difficulty of regionalizing any existing observations. This study explores the potential of a paired catchment monitoring network to provide statistically robust, regionalized predictions of LUC change impact in an environment of high hydrological variability. We test the importance of LUC variables to explain hydrological responses and to improve regionalized predictions using 24 catchments distributed along the Tropical Andes. For this, we calculate first 50 physical catchment properties, and then select a subset based on correlation analysis. The reduced set is subsequently used to regionalize a selection of hydrological indices using multiple linear regression. Contrary to earlier studies, we find that incorporating LUC variables in the regional model structures increases significantly regression performance and predictive capacity for 66% of the indices. For the runoff ratio, baseflow index, and slope of the flow duration curve, the mean absolute error reduces by 53% and the variance of the residuals by 79%, on average. We attribute the explanatory capacity of LUC in the regional model to the pairwise monitoring setup, which increases the contrast of the land‐use signal in the data set. As such, it may be a useful strategy to optimize data collection to support watershed management practices and improve decision‐making in data‐scarce regions.

Watershed services in the humid tropics: Opportunities from recent advances in ecohydrology

“…At a larger scale, ecohydrologists can use regionalization approaches, gaining insights on the hydrological behaviour of watersheds over large areas. Regionalization approaches allow watershed services programs to benefit from insights gained in similar environmental settings (e.g., Initiative for Hydrological Monitoring of Andean Ecosystems (iMHEA) network [Célleri et al, ; Ochoa‐Tocachi, Buytaert, & De Bièvre, ]).…”

“…The development of information networks and platforms has implications for the design of monitoring and experimental strategies, which should facilitate regional comparisons and therefore generalization of local findings (Ochoa‐Tocachi, Buytaert, & De Bièvre, ). Robust and consistent methodology for data analyses is critical to interpret these datasets (Adams & Fowler, ).…”

“…Second, similar to other humid tropical systems, the annual distribution of precipitation in the region is controlled by the meridional oscillation of the intertropical convergence zone, leading to a bimodal distribution of precipitation throughout the year and an average annual precipitation above 2,000 mm·year −1 (Poveda, Waylen, & Pulwarty, 2006). Ecosystems such as the páramo regulate the water resources of communities at lower elevations, yet they remain poorly characterized and managed in an ad hoc way (Ochoa-Tocachi, Buytaert, De Bièvre, Célleri et al, 2016;Ponette-González et al, 2014), and studies that systematically combine direct observations, modelling, and citizen science are virtually not existent. We believe that it is feasible to use techniques that are already at hand-and widely applied in temperate regions-to examine both the hydroclimatic responses and the coupling of human-natural dynamics of this unique ecosystem.…”

“…After a calibration period, one of the catchments is subjected to a treatment (e.g., deforestation or afforestation) and the other remains as a control. Pairedwatershed studies have been useful to "substitute space for time" in hydrological monitoring, for example, to understand the effects of forest regeneration(Bren, Lane, & Hepworth, 2010) or watershed development on water and sediment fluxes(Neary, 2016;Ochoa- Tocachi, Buytaert, De Bièvre, Célleri et al, 2016;Wemple, Shanley, Denner, Ross, & Mills, 2007), the function of riparian buffers…”

Can climate variability information constrain a hydrological model for an ungauged Costa Rican catchment?

“…Shafii and Tolson (2015), for example, use hydrological signatures to define goodness-of-fit measures in an optimization procedure used for calibration. For ungauged catchments, where observed signature values are not available, hydrological signatures have been regionalized (Ochoa-Tocachi, Buytaert, & De Bièvre, 2016) and then used to constrain the model-parameter space (Kapangaziwiri, Hughes, & Wagener, 2012;Yadav, Wagener, & Gupta, 2007). Signatures are also used as diagnostic tools to detect-and to improve-weaknesses in model representations (e.g., Fenicia, McDonnell, & Savenije, 2008;Fovet, Ruiz, Hrachowitz, Faucheux, & Gascuel-Odoux, 2015; Gupta, Wagener, & Liu, 2008;Guse et al, 2016).…”