Scientific Background ===================== This page provides detailed scientific context for the methods implemented in HyGCS. Hysteresis in C-Q Relationships -------------------------------- Hysteresis occurs when the relationship between concentration (C) and discharge (Q) differs between rising and falling flow limbs, creating a loop in C-Q space. Direction and Interpretation ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ **Clockwise Hysteresis** Concentration peaks before discharge peaks. Indicates: - Flushing of readily available material - Transport-limited export - Proximal sources - High connectivity during rising limb **Counter-Clockwise Hysteresis** Concentration peaks after discharge peaks. Indicates: - Progressive mobilization - Source-limited export - Distal sources - Delayed connectivity **Figure-8 or Complex Patterns** Multiple loops or mixed patterns. Indicates: - Multiple source contributions - Changing transport pathways - Complex hydrological response - Need for multi-method analysis Implemented Hysteresis Methods ------------------------------- HyGCS implements three complementary methods for hysteresis analysis. HARP Method (Roberts et al., 2023) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ **Hysteresis Analysis of Rising and falling Peaks** Empirical classification based on: - Peak timing difference (ΔT = T_C - T_Q) - Loop area - Residual (end-state deviation) **Strengths:** - Intuitive interpretation - Clear process identification - Named classification system **Limitations:** - Requires clear peaks - Qualitative rather than quantitative - May struggle with complex patterns **Reference:** Roberts, M.E. et al. (2023). Hysteresis Analysis of Rising and falling Peaks (HARP): A new method to identify changing sediment and hydrological connectivity. Hydrological Processes. Zuecco Index (Zuecco et al., 2016) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ **Integration-Based Hysteresis Index** Calculates h-index by integrating differential areas between rising and falling limbs: .. math:: h = \sum_{i} (A_{rise,i} - A_{fall,i}) where areas are computed between Q percentiles. **9-Class System:** ======== ============================ ================== Class h-index Range Description ======== ============================ ================== 0 Near zero Linear/no hysteresis 1-4 Positive (varying magnitude) Clockwise variants 5-8 Negative (varying magnitude) Counter-clockwise variants ======== ============================ ================== **Strengths:** - Quantitative magnitude assessment - Detects complex/mixed patterns - Robust to noise **Limitations:** - Requires interpolation - Classification thresholds somewhat arbitrary **Reference:** Zuecco, G. et al. (2016). A versatile index to characterize hysteresis between hydrological variables at the runoff event timescale. Hydrological Processes, 30(9), 1449-1466. Lloyd/Lawler Methods (2016, 2006) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ **Percentile-Based Indices** Samples C at 9 Q percentiles (0.1, 0.2, ..., 0.9) on rising and falling limbs. **HInew (Lloyd 2016) - Recommended:** .. math:: HI_{new} = \frac{C_{rise} - C_{fall}}{C_{mid}} - Symmetric range: [-1, 1] - C_mid = (C_rise + C_fall) / 2 **HIL (Lawler 2006) - Original:** .. math:: HI_L = \begin{cases} (C_{rise} / C_{fall}) - 1 & \text{if } C_{rise} > C_{fall} \\ (-1 / (C_{rise} / C_{fall})) + 1 & \text{otherwise} \end{cases} - Asymmetric range - More sensitive at extremes **Strengths:** - Standard in literature - Percentile-based (robust to outliers) - HInew facilitates comparison **Limitations:** - Requires both rising and falling limbs - May miss complex patterns **References:** - Lloyd, C.E.M. et al. (2016). Using hysteresis analysis of high-resolution water quality monitoring data. Hydrology and Earth System Sciences, 20, 2705-2719. - Lawler, D.M. et al. (2006). Turbidity dynamics during spring storm events. Science of the Total Environment, 360, 109-126. CVc/CVq Framework ----------------- Coefficient of Variation Approach (Musolff et al., 2015) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Distinguishes chemostatic from chemodynamic behavior using variability ratios: .. math:: \frac{CV_c}{CV_q} = \frac{\sigma_C / \mu_C}{\sigma_Q / \mu_Q} **Interpretation:** **CVc/CVq > 1** (Chemodynamic) Concentration varies more than flow. Indicates: - Variable source contributions - Transport-dependent export - Event-driven dynamics - Hysteretic behavior likely **CVc/CVq < 1** (Chemostatic) Concentration buffered relative to flow. Indicates: - Consistent source strength - Concentration-discharge equilibrium - Hysteresis less pronounced - Stable biogeochemical processes **Implementation in HyGCS:** Computed on rolling windows (typically 5-10 samples) to capture temporal dynamics. **Reference:** Musolff, A. et al. (2015). Catchment controls on solute export. Advances in Water Resources, 86, 133-146. C-Q Relationships ----------------- Power-Law Model (Thompson et al., 2011) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The C-Q relationship is commonly modeled as: .. math:: C = aQ^b Taking logarithms: .. math:: \log(C) = \log(a) + b \cdot \log(Q) where: - **a** = intercept (baseline concentration) - **b** = slope (mechanistic indicator) **Slope Interpretation:** **b > 0.15** (Dilution/Flushing) Concentration increases with flow. Indicates: - Transport-limited export - Flushing of accumulated material - Proximal sources activated - Increasing connectivity **b < -0.15** (Enrichment/Loading) Concentration decreases with flow. Indicates: - Dilution of point sources - Source-limited export - Groundwater contribution dominant - Decreasing connectivity **|b| < 0.1** (Chemostatic) Weak C-Q relationship. Indicates: - Buffered system - Consistent source strength - Concentration-discharge equilibrium - Weak flow dependency **Reference:** Thompson, S.E. et al. (2011). Comparative hydrology across AmeriFlux sites: The variable roles of climate, vegetation, and groundwater. Water Resources Research, 47(10). Geochemical Phase Classification --------------------------------- HyGCS implements a hierarchical 6-phase classification system developed by Sanchez et al. (2025, in review) that integrates: 1. Window-scale hysteresis indices 2. C-Q slope (power-law exponent) 3. CVc/CVq variability ratios 4. Flow dynamics (rising/falling, peaks) 5. Temporal context (phase transitions) The 6 Phases ~~~~~~~~~~~~ **F - Flushing** Characteristics: - Steep concentration decline during high flow - Positive C-Q slope (b > 0.15) - High CVc/CVq (chemodynamic) - Clockwise hysteresis common - Concentration in high percentile, declining Process interpretation: Rapid mobilization of accumulated material during high-flow events. Transport-limited export with strong connectivity. **L - Loading** Characteristics: - Concentration rising to maximum - Negative C-Q slope (b < -0.15) - Rising concentration trajectory - Counter-clockwise hysteresis possible - Concentration increasing with or before flow Process interpretation: Accumulation phase with progressive source mobilization. Enrichment before peak flow arrival. **C - Chemostatic** Characteristics: - Low hysteresis magnitude - Flat C-Q slope (|b| < 0.1) - Low CVc/CVq (< 1) - Stable concentration - Low variability Process interpretation: Buffered system with consistent source strength. Concentration-discharge equilibrium maintained. **D - Dilution** Characteristics: - Post-flush recovery - Declining flow - Declining concentration - Follows flushing phase - Medium to low connectivity Process interpretation: Recovery after flushing event. Sources depleted, system returning to baseflow conditions. **R - Recession** Characteristics: - Late cycle, both flow and concentration declining - Low CVc/CVq - Low hysteresis - Days since peak > threshold - Low connectivity Process interpretation: Baseflow-dominated conditions. Limited source availability and weak connectivity. **V - Variable** Characteristics: - Ambiguous patterns - Mixed signatures - Low confidence classification - Transitional behavior Process interpretation: Complex or mixed processes not fitting other categories. May indicate multiple overlapping processes or insufficient data. Percentile-Based Thresholds ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The classification uses percentile-based thresholds rather than absolute values, making it compound-agnostic and adaptable to different concentration ranges. **Advantages:** - Works across different compounds (metals, nutrients, etc.) - Adapts to site-specific conditions - Robust to concentration scale differences - Reduces need for parameter tuning **Thresholds Computed:** - Flow percentiles (33rd, 67th for low/medium/high) - Concentration change percentiles (25th, 75th) - C-Q slope absolute thresholds (±0.15, ±0.1) - CVc/CVq ratio (typically 1.0 threshold) Window-Scale Hysteresis ~~~~~~~~~~~~~~~~~~~~~~~ Unlike traditional event-scale hysteresis, HyGCS computes hysteresis indices on moving windows (typically 10-20 points) around each classified segment. **Why window-scale?** - Captures local temporal dynamics - Avoids artifacts from full time series loops - More appropriate for long-term monitoring data - Detects changing patterns over time Hierarchical Rule System ~~~~~~~~~~~~~~~~~~~~~~~~~ Rules are checked in priority order to avoid ambiguity: 1. **Strong signatures** (F, L) - checked first 2. **Moderate signatures** (C, D, R) - checked next 3. **Default to V** if no clear pattern Confidence scoring based on: - Number of rules triggered - Agreement between indicators (hysteresis, slope, CVc/CVq) - Data quality (sufficient points, valid metrics) - Consistency with temporal context (previous phases) Critical Perspective -------------------- Knapp & Musolff (2024) - Critical Assessment ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Important considerations for C-Q analysis: **Multi-Method Validation** No single method is perfect. Convergent evidence from multiple methods increases confidence. When methods disagree, investigate further. **Contextual Interpretation** Hysteresis and C-Q slopes reveal patterns but not mechanisms directly. Must combine with: - Site knowledge (geology, land use, hydrology) - Process understanding (biogeochemistry, transport) - Additional data (tracers, high-frequency monitoring) **Data Requirements** Quality and quantity of data matter: - Temporal resolution affects pattern detection - Sampling bias can create artifacts - Outliers and measurement errors propagate - Need adequate coverage of flow range **Limitations of Classification** Automated classification is a tool, not truth: - Low confidence classifications need investigation - Phase boundaries are fuzzy, not discrete - Complex systems may not fit simple categories - Always validate against known reference periods **Reference:** Knapp, J.L.A. & Musolff, A. (2024). Mind the gap: A critical perspective on concentration-discharge relationships. Hydrological Processes. https://doi.org/10.1002/hyp.15328 Best Practices -------------- 1. **Use Multiple Methods** Always compare HARP, Zuecco, and Lloyd/Lawler results. Agreement → confidence. 2. **Validate with Known Events** Test on reference events with known behavior to calibrate interpretation. 3. **Check Data Quality** - Remove obvious outliers - Ensure Q and C are synchronized - Verify sufficient temporal coverage - Examine gaps and missing data 4. **Consider Temporal Context** - Seasonal patterns - Antecedent conditions - Long-term trends - Event sequencing 5. **Integrate Process Knowledge** - Site characteristics (geology, land use) - Known sources and pathways - Historical behavior - Expert knowledge 6. **Report Uncertainty** - Confidence scores - Method agreement/disagreement - Data limitations - Alternative interpretations 7. **Avoid Over-Interpretation** Hysteresis reveals patterns, not mechanisms directly. Use as hypothesis-generating tool, not definitive proof. See Also -------- - :doc:`api_core` - Implementation details - :doc:`api_classification` - Classification algorithm - :doc:`quickstart` - Practical usage examples