TL;DR
This paper investigates levee seepage using a real-scale experimental setup and an optical-fiber distributed temperature system (DTS). The study highlights the challenges of interpreting raw DTS data due to fluctuations and external influences, demonstrating that appropriate data smoothing techniques are crucial for effective levee monitoring.
What Problem Does This Paper Address?
Levee failures, often caused by seepage, pose a significant threat to communities during floods. Traditional monitoring methods like visual inspection or localized pore water pressure measurements have limitations in providing comprehensive, real-time spatial data on seepage. While optical-fiber DTS offers a promising solution for continuous, distributed monitoring, the raw temperature data can be noisy and difficult to interpret due to external factors like ambient temperature. This paper addresses the need for a robust method to analyze DTS data for effective levee seepage monitoring, bridging the gap between raw sensor readings and actionable insights into levee stability.
Methods
The study employed a real-scale experimental setup with two constructed levees within a concrete box to control environmental factors. Water was supplied to induce seepage, and water pressure was monitored using digital pressure sensors and water level loggers. Crucially, an optical-fiber DTS was installed within the levees to continuously measure temperature at 1-meter intervals.
The core methodological contribution lies in the post-processing of the DTS data. Recognizing the inherent noise and fluctuations in the raw temperature readings, the researchers explored five different smoothing techniques:
- Moving Average: Replaces each data point with the average of its neighbors.
- Savitzky-Golay: A generalized moving average that fits polynomials to data segments.
- LOESS (Locally Estimated Scatterplot Smoothing): Uses locally weighted regression.
- LOWESS (Locally Weighted Scatterplot Smoothing): Similar to LOESS but with a simpler computational approach.
- Exponential Smoothing: Assigns exponentially decreasing weights to past observations.
These methods were applied at various time intervals (1 min, 1 hr, 3 hr, 6 hr) to identify the most effective approach for reducing noise, highlighting temperature trends, and enabling comparisons across different levee locations.
Datasets
The study utilized data collected from a real-scale levee experiment. This involved:
- Water Pressure Data: Collected every second from 14 pressure sensors (two Series 26Y and twelve OTT Orpheus mini water level loggers).
- Temperature Data: Collected every 14 seconds from distributed optical-fiber cables installed at multiple heights within the levees.
This experimental dataset, generated specifically for this study, serves as the primary "dataset" for evaluating the proposed monitoring and data processing methods.
Key Insights and Ideas
- DTS is Sensitive but Noisy: The optical-fiber DTS successfully captured temperature changes within the levees, correlating with seepage events. However, the raw data was highly susceptible to external temperature fluctuations and exhibited significant noise, making direct interpretation challenging.
- Smoothing is Essential: The study unequivocally demonstrates that raw DTS temperature data is insufficient for reliable levee monitoring. Data processing, specifically smoothing, is a critical prerequisite for extracting meaningful seepage information.
- Time Interval Matters: The choice of smoothing interval significantly impacts the results. Short intervals (e.g., 1 minute) fail to adequately reduce noise, while very long intervals can oversimplify the data, obscuring important trends. A 3-hour interval was found to be a good balance for capturing temperature change tendencies.
- Moving Average Shines for this Application: Among the tested smoothing techniques, the moving average method, particularly with a 3-hour interval, proved to be the most effective for this specific levee seepage scenario. This is attributed to its ability to remove outliers without being overly influenced by seasonality (which was absent in this experimental setup) and its computational efficiency.
- Spatial and Temporal Analysis is Key: Seepage effects manifest differently across the levee's height and length. Understanding these variations requires analyzing temperature data both spatially (at different sensor locations) and temporally (over time).
Key Results
- The optical-fiber DTS successfully detected temperature changes associated with seepage, with saturation leading to a decrease in internal levee temperature.
- Raw temperature data exhibited severe fluctuations, making direct interpretation difficult.
- The moving average smoothing technique, applied at 3-hour intervals, was identified as the most suitable method for analyzing the temperature data, effectively reducing noise while preserving key trends.
- The study confirmed that seepage occurred more rapidly in levee 1 (with coarser particles) compared to levee 2 (with more silt), as indicated by faster water pressure changes and temperature responses.
- The analysis of temperature differences across the foundation layer revealed that seepage primarily occurred in the bottom layers, especially in levee 2, suggesting potential piping or foundation leakage.
Limitations and Future Work
- External Temperature Influence: The study acknowledges that external temperature significantly influences the internal levee temperature, complicating the direct identification of seepage onset.
- Limited Scope of Smoothing Techniques: While five smoothing techniques were evaluated, the paper suggests that further research into other data-processing techniques could be beneficial for different monitoring scenarios.
- Long-Term Monitoring: The current study was based on a relatively short experimental period. Future work should focus on long-term measurements to understand the system's performance under various environmental conditions and over extended periods.
- Comprehensive Environmental Understanding: For practical implementation, a deeper understanding of how various external environmental factors interact with levee seepage and DTS readings is necessary.
- Spatial Monitoring Strategy: The paper emphasizes the need to monitor both the entire levee area and specific regions susceptible to damage, suggesting that optimized sensor placement and data analysis strategies are crucial.
Summary updated