| Popis: |
The Gravity Recovery and Climate Experiment (GRACE) satellite has been widely used to monitor changes in terrestrial water storage anomalies (TWSA), which vertically integrate water storage changes from the land surface to the deepest aquifers. Isolation of groundwater storage anomalies (GWA) from TWSA requires information of other water budget components from auxiliary datasets, e.g., Land Surface Model (LSM) output, or in-situ/remotely sensed based data. Using auxiliary datasets to account for water budget components have revealed large biases and uncertainties, especially over regions with complicated hydrological processes, leading to accumulating errors in GRACE-GWA estimates. Comparisons of GRACE-GWA with in-situ observations permit evaluating how accurately we can isolate groundwater storage signals from TWSA. Goodness-of-fit (GOF) indices e.g., Spearman correlation, Nash-Sutcliffe Efficiency (NSE), and the Kling-Gupta Efficiency (KGE), are commonly applied hydrologic fit metrics that express similarity of time series. Such metrics are used in our study to compare GRACE-GWA estimations and in-situ observations. Our results showed that GOF indices failed to capture the different characteristics of GRACE-GWA timeseries. Spearman correlation requires a monotonic relationship, an assumption violated given the seasonal amplitudes of GRACE-GWA. Using a Pearson correlation is ill-advised given serial correlation and non-normality. NSE is biased and influenced by skewness and periodicity, which is given since the data is seasonal in nature. KGE is based on the assumptions of data linearity and data normality. Non-normality in GRACE-GW time series violates the implicit assumptions underlying KGE. The goal of this work is to improve interpretation and use of GOF metrics to validate GRACE-GWA estimates, highlighting the importance of assessing multiple GOF criteria beyond simply correlation often applied in GRACE studies. We show that a rigorous assessment of GOF enhances our ability to interpret GRACE-GWA.Acknowledgement:The researcher, Mohamed Akl, is funded by a full PhD scholarship from the Ministry of Higher Education of the Arab Republic of Egypt. |
