Kalyl Gomes Calixto

and 2 more

Investigating watershed hydrology from a data-driven causal perspective is an attractive opportunity to characterize and understand relationships between water storages and fluxes. Here we assess integrally how the water balance components interact with themselves, aiming to find relevant time-lags or dependency patterns. Granger’s causality test and time-lagged mutual information were used in a pairwise approach to examine cause-effect relationships between precipitation, streamflow, groundwater levels under different land-covers, and evapotranspiration data (daily timescale) from 2009 to 2019 in a Brazilian watershed (52 km²), located in a recharge area of the Guarani Aquifer System. A verification assessment using synthetic datasets shows that the methods are effective to identify the underlying generating mechanisms. Statistically significant causal connections were confirmed in practically all pairs of observed data. Granger’s causality indicates that groundwater and streamflow responses are influenced by precipitation even with a lag of 1-day, while evapotranspiration can take more than 200 days to influence groundwater responses, depending on the water table depth and surrounding land-cover. Mutual information curves show dependency patterns between hydrological processes that are different from the ones obtained by cross-correlation functions. The causal analysis provides a complementary view of the hydrological system’s functioning and may lead us to develop predictive models that reproduce not only the target variables but also the diverse time-lagged dependencies observed in environmental data.

Dimaghi Schwamback

and 2 more

Inherent errors in tipping bucket flow meters may limit monitoring data reliability. In this work, we perform the static and dynamic calibration of four large tipping buckets, apply different regression curves and investigate the possible measurement error sources. The volumetric capacity (static calibration) of each piece of equipment was determined. They were tested (dynamic calibration) under ten flow intensities, ranging from low to high rainfall intensities (return period larger than 100 years). For each flow rate, the measurement was recorded during six time intervals (1, 2, 5, 10, 20 and 30 minutes) and four regression equations - linear, potential, T vs. 1/Q and quadratic - were tested. According to the static calibration, the equipment has a volumetric capacity of 11.63 mL (TB1), 64.16 mL (TB2), 139.86 mL (TB3) and 660.95 mL (TB4). When tested under different flow rates (dynamic calibration), underestimations were identified according to the size of the cavity: TB1 (3.31%), TB2 (5.75%), TB3 (9.33%) and TB4 (13.57%). Among the alternative curves, linear regression showed the best correlation (above 99%) with the monitored data. Using this method, the measurement errors were reduced to -1.35% (TB1), 0.04% (TB2), 3.18% (TB3) and 3.73% (TB4). We investigated how the different variables (tipping speed, cavity volumetric capacity and time interval of data collection) influenced the error. Errors follow a parabolic function of tipping velocity and a linear function of cavity volumetric capacity. The time interval of data collection interfered in the data sampled, however no statistical correlation was found. Among those variables, cavity size is the most important one. Given its low cost we aimed to minimize the inherent error in large tipping buckets flow meters and encourage its application, increasing in-situ collection of hydrological data.