Karst spring recession curve analysis: efficient, accurate methods for both fast and slow flow components

Analysis of karst spring recession hydrographs is essential for determining hydraulic parameters, geometric characteristics and transfer mechanisms that describe the dynamic nature of karst aquifer systems. The extraction and separation of different fast and slow flow components constituting karst spring recession hydrograph typically involve manual and subjective procedures. This subjectivity introduces bias, while manual procedures can introduce errors to the derived parameters representing the system. To provide an alternative recession extraction procedure that is automated, fully objective and easy to apply, we modified traditional streamflow extraction methods to identify components relevant for karst spring recession analysis. Mangin’s karst-specific recession analysis model was fitted to individual extracted recession segments to determine matrix and conduit recession parameters. We introduced different parameters optimisation approaches of the Mangin’s model to increase degree of freedom thereby allowing for more parameters interaction. The modified recession extraction and parameters optimisation approaches were tested on 3 karst springs in different climate conditions. The results show that the modified extraction methods are capable of distinguishing different recession components and derived parameters reasonably represent the analysed karst systems. We recorded an average KGE > 0.7 among all recession events simulated by recession parameters derived from all combinations of recession extraction methods and parameters optimisation approaches. While there are variability among parameters estimated by different combinations of extraction methods and optimisation approaches, we find even much higher variability among individual recession events. We provide suggestions to reduce the uncertainty among individual recession events and to create a more robust analysis by using multiple pairs of recession extraction method and parameters optimisation approach.

Quantifying Mountain Aquifer Recharge Rates Using Storage-Discharge Functions in the Sierra Nevada, California

<p>Mountain System Recharge (MSR) is one of the main components of recharge in many arid and semi-arid aquifers, yet the mechanisms of MSR in high-elevation mountain ranges are poorly understood. The complexity of recharge processes and the lack of groundwater observations in mountain catchments contribute to this problem. MSR consists of two distinct pathways: 1) mountain bedrock aquifer recharge (MAR) consists of snowmelt or rainfall derived infiltration into the mountain bedrock, which either discharges to streams as baseflow or reaches an alluvial aquifer in an adjacent valley via lateral subsurface flow referred to as mountain block recharge (MBR), and 2) Mountain front recharge (MFR) consists of streamflow infiltration at the mountain front. Here, we apply streamflow recession analysis across 11 anthropogenically unaffected catchments in the Sierra Nevada to derive seasonally distinct storage-discharge functions and quantify MAR in response to changes in precipitation. Median annual recharge efficiencies (ratio of annual MAR to precipitation) range from 4 to 28% and can reach up to 60% during the wettest years on record. We implement a global sensitivity analysis to identify parameters that significantly impact MAR rates. Results illustrate that MAR estimates are mostly sensitive to the filter parameters for streamflow data selection used during the recession analysis, and the number of dry days after a rain event where streamflow data are excluded has the greatest impact. Our results demonstrate that storage-discharge functions are useful for quantifying groundwater recharge in mountainous catchments under perennial flow conditions. However, estimated MAR rates are impacted by the uncertainty in streamflow data, filtering of streamflow time series and model structure. Future work will be focused on quantifying uncertainty in MAR estimates caused from various sources.</p><p> </p>

Evaluate recession characteristics from baseflow and separate baseflow via the knowledge guided machine learning

<p>Baseflow, referred to as the groundwater discharge, is essential to investigate the groundwater system. A common and classic approach to study baseflow is recession analysis method, but current methods confuse the concept of streamflow recessions and baseflow recessions. This confusion leads to a mixing effect of the fluxes from different storage components and theoretically inconsistent recession analysis results accordingly. Therefore, it motivates an improvement and enhanced scientific understanding of the empirically derived baseflow recession characteristics.  In addition, quantifying baseflow from streamflow is defined as the baseflow separation problem. The state-of-the-art baseflow separation tools are in lack of physical rules and have either structural limitations or are inapplicable in regions with insufficient data, which confines the generalization performance. To overcome these issues, we applied a knowledge guided machine learning (KGML) approach to separate baseflow, which embeds physically derived baseflow recession characteristics in the traditional machine learning framework.</p><p>Recession parameter, which is derived from empirical recession analysis, has been observed to exceed its theoretical range on a recession event scale. Besides many potential environmental factors, we hypothesize that this well recognized inconsistency is because the quick flow from surficial water bodies has not been successfully excluded based on the recession selection criterion. We conduct recession analysis using both streamflow and baseflow over 1,000 gages across the continental United States. The baseflow was estimated from Eckhardt two-parameter digital filter and was calibrated against the in-stream field data. It was found that for gages whose calibration performance is satisfactory, the baseflow derived recession parameter agrees more consistently with the recession characteristics, which are indicated by the Boussinesq solutions.</p><p>Traditional baseflow separation tools partition streamflow into quick flow and base flow. Those tools have data scarcity issues and structural limitations without involving physical perspectives. To introduce physical rules into baseflow separation and overcome data scarcity issues, we apply a recession-based loss function to train the machine learning model such that the recession characteristics of separated baseflow agree with their theoretical behaviors. Guided by the recession knowledge of baseflow on a catchment scale, progress is being made to finalize this KGML implementation and to improve the baseflow separation approach.</p>

Keynes's trading on Wall Street: did he follow the same behaviour when investing for himself and for King's?

In the last few years Keynes's investment activity, both as an individual trader and as a manager of institutions’ portfolios, has attracted attention in the specialised literature. Recently his investments on Wall Street, in particular – both on his own account (Cristiano, Marcuzzo and Sanfilippo 2018) and on behalf of King's College, Cambridge (Chambers and Kabiri 2016) – have been analysed, and the evident connection with his theoretical analysis of the functioning of the financial markets contained in chapter 12 of The General Theory has been duly stressed. This article aims to contribute to a more comprehensive understanding of Keynes's trading behaviour on Wall Street by providing a detailed comparison of his investment choices when he traded for himself and for King's. There are similarities, as might be expected, but also significant differences, well worth investigating. As far as the differences are concerned, one of the most striking is to be seen, for instance, in his attitude when, after a period of bull market in 1936, he had to face the spring 1937 burst of the speculative bubble and subsequent recession. Analysis of his behaviour in this specific case reveals that the event took him by surprise but his reaction differed with regard to his personal investments and the King's investments. The prevalence of a ‘buy and hold’ strategy, which, according to Chambers and Kabiri's reconstruction (2016), marked Keynes's behaviour in general (and also in this particular case) when he invested on behalf of King's, was not always his typical choice when the investments were undertaken on his own account. A tentative explanation of this result, which is also grounded on some different features characterising the two portfolios and not sufficiently investigated in previous studies, is at last provided in the article.

Extraction Method of Baseflow Recession Segments Based on Second-Order Derivative of Streamflow and Comparison with Four Conventional Methods

Baseflow recession analysis is widely used in hydrological research, water resource planning and management, and watershed hydrogeological research. The first step of baseflow recession analysis is to extract the baseflow recession segments from the hydrograph. Different extraction results lead to different analysis results. At present, the four major recession segment extraction methods applied by hydrologists are mostly based on experience, and there is no clear theoretical basis. Therefore, this study derives a second-order derivation (Sec-D) recession segment extraction method based on the power law relationship between storage and discharge. Moreover, by applying the Sec-D method and the four conventional extraction methods to four hydrological stations in the Tao’er River basin in northeastern China, the differences in the recession segment extraction, determination of basin-wide hydrogeological parameters, and groundwater balance estimation are compared. The results demonstrate that, contrary to the four conventional methods, the Sec-D method can effectively eliminate the early recession stage affected by the surface runoff or rainfall and some streamflow data with more than 1% non-sequential error. The hydraulic conductivity of the four basins estimated by the Sec-D method is between 2.3 × 10−5–4.9 × 10−5 m/s, and the aquifer thickness is between 131.2 and 202.5 m. However, the four conventional extraction methods may underestimate (by about 2.5 times) the basin-wide hydraulic conductivity and overestimate (by about 3 times) the aquifer thickness. The groundwater balance elements calculated by the Sec-D method and the four conventional methods present similar intra-annual fluctuation characteristics; the correlation coefficients of daily evapotranspiration calculated by the five methods ranged from 0.7 to 0.95, and those of daily effective groundwater recharge ranged from 0.95 to 0.99. The use of the Sec-D method in baseflow recession analyses is significant for future studies and can be combined with conventional methods.