Journal of Hydrologic Engineering

Flood Insurance Rate Maps (FIRMs) managed by FEMA have been providing ongoing flood information to most communities in the United States over the past half-century. However, the uncertainty associated with the modeling of FIRMs, some of which are created by using a single Hydrologic Engineering Center River Analysis System (HEC-RAS) one-dimensional (1D) steady-flow model, may have adverse effects on the reliability of flood stage and inundation extent predictions. Therefore, a systematic understanding of the uncertainty in the modeling process of FIRMs is necessary. Bayesian model averaging (BMA), which is a statistical approach that can combine estimations from multiple models and produce reliable probabilistic predictions, was applied to evaluating the uncertainty associated with FIRMs. In this study, both the BMA and hierarchical BMA (HBMA) approaches were used to quantify the uncertainty within the detailed FEMA models of the Deep River and the Saint Marys River in the state of Indiana based on water stage predictions from 150 HEC-RAS 1D unsteady-flow model configurations that incorporate four uncertainty sources including bridges, channel roughness, floodplain roughness, and upstream flow input. Given the ensemble predictions and the observed water stage data in the training period, the BMA weight and the variance for each model member were obtained, and then the BMA prediction ability was validated for the observed data from the later period. The results indicate that the BMA prediction is more robust than both the original FEMA model and the ensemble mean. Furthermore, the HBMA framework explicitly shows the propagation of various uncertainty sources, and both the channel roughness and the upstream flow input have a larger impact on prediction variance than bridges. Hence, it provides insights for modelers into the relative impact of individual uncertainty sources in the flood modeling process. The results show that the probabilistic flood maps developed based on the BMA analysis could provide more reliable predictions than the deterministic FIRMs.

Abstract

Flood seasonality (FS) analysis is a key to delineating hydrologically homogenous regions and better understanding flood risk, especially for arid regions. Although Iran has undergone heavy floods in current decades, a quantitative spatially distributed overview of the flood seasonality and the its affected mechanisms is currently unavailable. This study aimed in seasonality analysis of annual maximum flood with average recurrence of 2 to 500 years in 291 continuous hydrometric stations across Iran having sufficient recorded flood data ($>20\mathrm{years}$, total of 9,565 events) using a circular statistics approach. Moreover, the relationships of recurrence season of floods with eight natural and human-induced factors including catchment area, catchment perimeter, mean catchment slope, normalized difference vegetation index (NDVI), elevation of station, mean slope of river, and latitude were examined. Findings showed that flooding in Iran presents strong seasonality, with 32% and 41% of the total events occurring during winter and spring, and mostly during April (24%). Although summer floods ($\sim 11\%$) are more frequent in the northern parts, winter floods present higher percentages in the south. A significant difference in flood seasonality (2 to 500 years) was identified between the northern and southeastern regions. This study suggests that the geographical location (especially latitude) of the hydrometric station in Iran seems to be more significant than other considered factors in shaping flood seasonality. Results can serve as a foundation for enhancing the scientific understanding of the flood seasonality across Iran, which can provide useful insights for flood prevention planning and future changes projected by models.

Practical Applications

Flooding is one of most damaging phenomena around the globe that affect humans’ life and economy. So, in order to decrease the dire consequences of flooding, especially in arid zones like Iran, we should increase our knowledge about flood risk by using advanced techniques like flood seasonality analysis. This research is a specific survey at a nationwide scale for enhancing the scientific understanding of the flood timing across Iran, which can provide useful insights for future planning and practical engineering designs. The purpose of this research is the study of seasonality analysis of annual maximum flood in 291 continuous hydrometric stations across Iran with a total of 9,565 events using a circular statistics approach. Owing to the prominent role of humanity and nature on flooding, eight natural and human-induced factors have been surveyed. The results of this study illustrated that 32% and 41% of the total events occurred during winter and spring, mostly during April (24%). What is further noticeable is that a significant difference in flood seasonality was identified between the northern and southeastern regions. Among considered factors in shaping seasonality of flooding, the geographical location (especially latitude) of the hydrometric station in Iran seems to be more significant than other factors.

Soil heterogeneity is a key factor affecting horizontal water flow and solute transport in unsaturated soil, in which particle distribution, pore structure, and pore connectivity in the soil all present fractal characteristics. The initial water content of soil has significant influence in the resulting heterogeneity of the soil structure, which can affect the water flow and solute transport. In this study, through a series of horizontal transport laboratory simulation experiments in soil columns, the initial water content dependency of transport was elucidated using the fractal Richards’ equation (FRE) model. We first observed that water flows faster in soil columns with higher initial water content, but the law is opposite in the initial process of water flow. By analyzing the variation of parameter values in the FRE model, we speculated that the abnormal phenomenon of early diffusion is caused by the heterogeneity of the soil structure. To elucidate the mechanisms of the water content’s dependency of the solute transport process and to identify the circumstance of altering of flow regime, a series of dye tracer experiments were conducted. Results show that in soil columns with low water content, the inflow of external fluid leads to the agglomeration of soil particles. Combined with matric suction, the heterogeneity of the soil structure is enhanced, which affects the formation of water flow and solute transport paths. It even leads to preferential flow in unsaturated soil columns with low water content.

The constant-discharge pumping test is a widely used aquifer characterization tool for estimating hydrogeological parameters. Various graphical and numerical solution techniques have been devised for interpretation of pumping test results. This study proposes an alternative regression-based solution approach for analyzing pumping tests conducted in nonleaky, confined aquifers. A regression model was developed by making use of the similarity solution that describes the temporal drawdown behavior at a monitoring well within a pumping period. The field curve obtained from a pumping test (drawdown versus time data) and the theoretical type curve (dimensionless drawdown versus dimensionless time) exhibit identical behavior in logarithmic scale. Correspondingly, based on the equivalence of the area above these two curves, the proposed regression equation establishes a link between a selected drawdown value and its dimensionless counterpart. Once this drawdown value is associated with the corresponding dimensionless drawdown, the explored aquifer transmissivity and storativity can be retrieved easily with the similarity solution. The proposed fit equation enables users to retrieve hydrogeological parameters effortlessly by eradicating the difficulties encountered while implementing conventional graphical or optimization-based methods. The method was tested for a large number of hypothetical noise-free and noise-perturbed time drawdown data sets, each synthesized with different aquifer parameters within a wide range, and in all these attempts the proposed algorithm achieved providing highly accurate parameter estimates. In addition, two published test cases were employed to assess the estimation performance of the method in comparison with the existing techniques. The developed solution scheme was revealed to yield results in very good agreement with those reported in previous works. Considering the accuracy of the method and the ease of its implementation, the proposed approach can be a practical tool for analyzing constant-discharge tests performed in confined aquifers.