Time-series of groundwater recharge, Tiber Riber Basin, Italy from 801 CE to the present day
Data files
Jan 08, 2024 version files 313.89 KB
-
Groundwater_Recharge_Tiber_River_Basin_Italy.xlsx
311.37 KB
-
README.md
2.52 KB
Abstract
Groundwater, essential for water availability, sanitation, and achieving Sustainable Development Goals, is shaped by climate dynamics and complex hydrogeological structures. Here, we provide a time-series of groundwater recharge from 801 CE to the present day in the Tiber River Basin, Italy, using historical records and hydrological modelling. Groundwater drought occurred in 36% of the Medieval Climatic Anomaly (801-1249) years, 12% of the Little Ice Age (1250-1849) years, and 26% of the Modern Warming Period (1850-2020) years. Importantly, a predominant warm phase of the Atlantic Multidecadal Oscillation, aligned with solar maxima, coincided with prolonged dry spells during both the medieval and modern periods, inducing a reduction in recharge rates due to hydrological memory effects. This study enhances understanding of climate-water interactions, offering a comprehensive view of groundwater dynamics in central Mediterranean and highlighting the importance of the past for sustainable future strategies. Leveraging this understanding can address water scarcity and enhance basin resilience.
README: Groundwater recharge, Tiber Riber Basin, Italy
https://doi.org/10.5061/dryad.zs7h44jgs
The dataset focuses on the Tiber River Basin (TRB) in central Italy and provides comprehensive information on groundwater dynamics from 801 CE to the present day. The study involves the collection of various climatic and hydrological data, including annual mean precipitation, annual mean temperature, and surface runoff, sourced from historical records and the Annals Project. The dataset employs Thornthwaite's water balance model to estimate groundwater recharge (GWR) rates in the TRB. The dataset introduces the HHydroREM historical model for groundwater recharge estimation, incorporating a non-linear multivariate regression approach. This model uses monthly storm severity index, snow severity index and Palmer Drought Severity Index data as inputs. It accounts for delayed drainage and simulates groundwater recharge patterns during dry spells and other climatic conditions.
Description of the data and file structure
The dataset is organized into five sheets within a spreadsheet file. The dataset provides a comprehensive view of the model structure and the model calibration and validation results, including the inputs, model parameters, and corresponding modeled groundwater recharge outputs in the first sheet. Additionally, it offers a reconstructed time-series of groundwater recharge values from 801 to 2020, with associated input variables starting from the year 789. This structure enables users to analyze the model's performance, understand the impact of different input variables, and explore the modeled groundwater recharge patterns over a long historical period.
The 'GWR Calibration' and 'GWR Validation' sheets compare the model estimates with the Thornthwaite groundwater recharge water balance (or Thornthwaite's surplus), while the 'GWR Extended Validation' sheet compares the model estimates with standardised mean values of groundwater level measurements made at two sites (Boschetti and Peschiera). In each sheet, each row corresponds to a specific year. Where necessary, specific months (from August to April of the following year) are represented by a column, each month with its corresponding values of the monthly storm severity index and hydrological weights in the 'GWR Modeling' and 'GWR Reconstruction' sheets. Other factors are also shown in columns. The term 'n/a' used throughout the dataset indicates that certain information is not relevant for a particular data point.
Methods
The dataset involves the collection of various climatic and hydrological data for the TRB, the application of Thornthwaite’s model for groundwater recharge estimation, and the development and validation of the HHydroREM historical model. The process includes calibration, delay consideration, and extensive statistical analysis to assess model performance.
The dataset focuses on the Tiber River Basin (TRB) in central Italy, covering an area of over 17,000 km2. Data for the period 1928–1974 were collected, including annual mean precipitation (P) and annual mean temperature (Tm) for the TRB. Surface runoff (Qon) data were obtained from the Annals Project, overseen by the Italian National Hydrological and Oceanographic Service. Soil-moisture storage capacity data were derived from statistical records (ISTAT, 2010).
Thornthwaite’s water balance model was chosen for estimating groundwater recharge (GWR) rates. The model requires mean areal precipitation (P), mean areal temperature (T), direct runoff factor (DRO), and soil-moisture storage capacity (STC) as inputs. Specific parameters for the TRB were considered, along with adjustable parameters. The Thornthwaite-revised approach was calibrated by comparing its output with water equivalent thickness data from the GRACE-NASA MEaSUREs Program.
The historical estimation of groundwater recharge (GWR(h)) was performed using a non-linear multivariate regression model (HHydroREM) tailored to the TRB. Monthly storm severity index (MSSI), snow severity index (SSI), and Palmer Drought Severity Index (PDSI) data were used as inputs. Model parameters were determined through statistical and physical evaluations, with iterative adjustments to achieve a meaningful correlation. The model was validated by comparing outputs with observed time-series data from specific locations in the TRB. To simulate the delay leading to groundwater recharge, a delayed drainage factor was estimated over the preceding three years. The model was validated using observed time-series data from the Boschetti well and Peschiera discharge in the TRB. Time-series data were normalized for statistical and graphical comparisons. Model parameterization involved an iterative method and trial-and-error approach. Criteria for goodness-of-fit and error minimization were employed during parameterization. Metrics such as R2, mean absolute error (MAE), skewness, and kurtosis were used to assess model performance. Statistical and graphical software tools, including STATGRAPHICS, WESSA, and CurveExpert Professional, were used for data analysis. The Hurst exponent was calculated to estimate the rate of chaos and cyclical-trend patterns in the time-series data.