Accuracy Of Flood Forecasting Research Articles

Hydrologic Model Prediction Improvement in Karst Watersheds through Available Reservoir Capacity of Karst

This study aimed to enhance flood forecasting accuracy in the Liangfeng River basin, a small karst watershed in Southern China, by incorporating the Available Reservoir Capacity of Karst (ARCK) into the HEC-HMS model. This region is often threatened by floods during the rainy season, so an accurate flood forecast can help decision-makers better manage rivers. As a crucial influencing factor on karstic runoff, ARCK is often overlooked in hydrological models. The seasonal and volatile nature of ARCK makes the direct computation of its specific values challenging. In this study, a virtual reservoir for each sub-basin (total of 17) was introduced into the model to simulate the storage and release of ARCK-induced runoff phenomena. Simulations via the enhanced model for rainfall events with significant fluctuations in water levels during 2021–2022 revealed that the Nash–Sutcliffe efficiency coefficient (NSE) of the average simulation accuracy was improved by more than 34%. Normally, rainfalls (even heavy precipitations) during the dry season either do not generate runoff or cause negligible fluctuations in flow rates due to long intervals. Conversely, relatively frequent rainfall events (even light ones) during the wet season result in substantial runoff. Based on this observation, three distinct types of karstic reservoirs with different retaining/releasing capacities were defined, reflecting variations in both the frequency and volume of runoff during both seasons. As a real-time environmental variable, ARCK exhibits higher and lower values during the dry and rainy seasons, respectively, and we can better avoid the risk of flooding according to its special effects.

State updating in Xin’anjiang model by Asynchronous Ensemble Kalman filtering with enhanced error models

For flood simulation in humid catchments, the utilization of discharge observations to update the states of hydrological models may enhance performance. Asynchronous Ensemble Kalman Filter (AEnKF), an asynchronous variant of the Ensemble Kalman Filter (EnKF), holds substantial application potential in hydrological assimilation due to its ability to utilize more observations with almost no additional computational time. This study employs AEnKF to update the state variables of the Xin’anjiang model, necessitating the use of error models to perturb both model states and observations to generate ensemble spread. The Bias-corrected Gaussian Error Model (BGEM) is used to mitigate the systematic bias brought by perturbating soil moisture, and the Maximum a Posteriori Estimation Method (MAP) is employed for the estimation of hyperparameters of error models. Through synthetic and real-world data testing, it has been validated that the rectification of soil moisture perturbations using the BGEM significantly reduces the systematic bias induced by Gaussian perturbations. Moreover, the assimilation scheme introduced in this study, based on AEnKF with enhanced error models, outperforms the EnKF with those models. It substantially reduces the accumulation of past errors in the initial conditions at the start of the forecast, thereby aiding in elevating the accuracy of flood forecasting.

Flood Frequency Analysis Using Gumbel Distribution Method: A Case of Robigumero River, Abay Basin, Ethiopia

Hydraulic structures like weirs, dams, spillways, and bridges require precise estimation of flood peaks at the intended return period in order to be planned, built, and maintained. In this paper, the findings of a study conducted on the Robigumoro River and the flow measurements taken are presented. The flood frequency analysis of the Robigumoro River was performed using the Gumbel distribution, which is a probability distribution commonly used for modeling river flows. This analysis is crucial as it aims to safeguard the lives and properties located downstream from the catchment area. The Gumbel distribution was employed to model the highest annual river flow over a span of 20 years (1990-2009). The investigation was carried out by the Ethiopian water and energy office, Abay Basin Development Authority. The Robigumoro River's maximum annual discharge over a 20-year period (1990–2009) was modeled using Gumbel distribution technique. From the trend line equation, R<sup>2</sup> value of 0.935 which shows that Gumbel’s distribution is suitable for predicting expected flow in the river. It can be concluded that the Gumbel distribution can accurately forecast expected river flow. The flood peak values were calculated using the same procedure for various return times. This helps with storm management in the research region. The estimated discharges obtained using the Gumbel's distribution and return periods (T) of 2 years, 10 years, 50 years, 100 years, 150 years, 200 years, 300 years and 400 years are 177.327m<sup>3</sup>/s, 320.784m<sup>3</sup>/s, 446.553m<sup>3</sup>/s, 499.722m<sup>3</sup>/s, 530.727m<sup>3</sup>/s, 552.698m<sup>3</sup>/s, 583.38m<sup>3</sup>/s, and 605.577m<sup>3</sup>/s respectively. The accuracy of flood forecasts in the basin indicates their potential use in various applications such as the design of crucial hydraulic structures, river reach planning, construction of bridges, and conservation efforts for Robigumoro watershed.

Flood forecasting based on radar precipitation nowcasting using U-net and its improved models

Accurate and timely short-term precipitation nowcasting is important for achieving reliable flood forecasting. The data-driven approaches have performed well in radar echo extrapolation to nowcast precipitation. In this paper, U-Net and its improved models, including SmaAt-Unet, Nested-Unet, and U-Net 3Plus were applied to perform radar echo extrapolation and precipitation nowcasting for 0.5 h, 1 h, and 2 h lead times of typical rainfall processes. The nowcasted precipitation was used as input to the HEC-HMS hydrological model for flood forecasting to compare the effect of different structural improvements to U-Net on the accuracy of flood forecasting. The results demonstrated that Nested-Unet and U-Net 3Plus aided in enhancing the accuracy of the extrapolation of moderate intensity radar echoes. With fewer discrepancies and better correlation with measured rainfall, the U-Net and U-Net 3Plus precipitation nowcasting results also produced improved flood forecasting outcome. The precipitation nowcasting and flood forecasting for SmaAt-Unet were slightly worse than other models; the relative errors of both flood peak and depth for Nested-Unet at 0.5 h lead time were less than 20 %, showing a good performance. Moreover, in a separate control experiment, the accuracy of the echo extrapolation was significantly decreased when convolutional block attention module (CBAM) was added to each model. However, all models have better extrapolation accuracy than the basic ConvLSTM. In general, Nested-Unet and U-Net 3Plus were helpful to improve the accuracy of precipitation nowcasting and flood forecasting, and the forecasted flood with 0.5 h and 1 h lead times could match the actual flood processes, but the peak discharge from nowcasting with 2 h lead time were severely underestimated, while the peak occurrence time could be forecasted correctly. These conclusions and attempts can provide effective guidelines for regional precipitation nowcasting and flood forecasting.

The Flood Simulation of the Modified Muskingum Model with a Variable Exponent Based on the Artificial Rabbit Optimization Algorithm

In order to further improve the accuracy of flood routing, this article uses the Variable Exponential Nonlinear Muskingum Model (VEP-NMM), combined with the Artificial Rabbit Optimization (ARO) algorithm for parameter calibration, to construct the ARO-VEP-NMM flood routing model. Taking Wilson’s (1974) flood as an example, the model calculation results were compared and analyzed with the Muskingum model constructed with seven optimization algorithms. At the same time, six measured floods in the Zishui Basin were selected for model applicability testing. The results show that the ARO algorithm exhibits stronger robustness and search ability compared with other optimization algorithms and can better solve the parameter optimization problem of the Muskingum model. The use of the ARO-VEP-NMM model for flood routing accurately reflects the movement patterns of floods. The Nash coefficient of the Wilson section reached 0.9983, and the average Nash coefficient during the flood validation period in the Zishui Basin was 0.9, further verifying the adaptability and feasibility of the ARO-VEP-NMM model in flood routing. The research results can provide certain references and a theoretical basis for improving the accuracy of flood forecasting.

Flood Forecasting Using Hybrid LSTM and GRU Models with Lag Time Preprocessing

Climate change and urbanization have increased the frequency of floods worldwide, resulting in substantial casualties and property loss. Accurate flood forecasting can offer governments early warnings about impending flood disasters, giving them a chance to evacuate and save lives. Deep learning is used in flood forecasting to improve the timeliness and accuracy of flood water level predictions. While various deep learning models similar to Long Short-Term Memory (LSTM) have achieved notable results, they have complex structures with low computational efficiency, and often lack generalizability and stability. This study applies a spatiotemporal Attention Gated Recurrent Unit (STA-GRU) model for flood prediction to increase the models’ computing efficiency. Another salient feature of our methodology is the incorporation of lag time during data preprocessing before the training of the model. Notably, for 12-h forecasting, the STA-GRU model’s R-squared (R2) value increased from 0.8125 to 0.9215. Concurrently, the model manifested reduced root mean squared error (RMSE) and mean absolute error (MAE) metrics. For a more extended 24-h forecasting, the R2 value of the STA-GRU model improved from 0.6181 to 0.7283, accompanied by diminishing RMSE and MAE values. Seven typical deep learning models—the LSTM, the Convolutional Neural Networks LSTM (CNNLSTM), the Convolutional LSTM (ConvLSTM), the spatiotemporal Attention Long Short-Term Memory (STA-LSTM), the GRU, the Convolutional Neural Networks GRU (CNNGRU), and the STA-GRU—are compared for water level prediction. Comparative analysis delineated that the use of the STA-GRU model and the application of the lag time pre-processing method significantly improved the reliability and accuracy of flood forecasting.

Flood Simulation Study of China’s Data-Deficient Mountainous Watersheds Based on CMPA-Hourly

Heavy rainfall and flood disasters are frequent in mountainous watersheds in southwest China, and forecasting runoff floods in some mountainous watersheds is difficult. In this study, a typical watershed in the southwest mountainous region, the Qingyi River (13,000 km2), was selected for the lack of precipitation observation data in the watershed, and the BTOPMC (block-wise use of the topographic-based hydrologic model (TOPMODEL)) was used, using CMPA-Hourly (China Hourly Merged Precipitation Analysis combining observations from automatic weather stations, meteorological satellite, and weather radar at 0.05° × 0.05° grid) to improve the accuracy of flood forecasting. The results show that the Nash–Sutcliffe efficiency (NSE) of the flood forecast for the verification period in the Jiajiang section of the Qingyi River using CMPA-Hourly improved from 0.66 to 0.78, the flood error reduced from 18% to 9%, and the overall accuracy reached grade B or above. The results indicate that CMPA-Hourly, which integrates ground observation–radar–satellite precipitation, effectively combined the advantages of different sources of data to improve the resolution and accuracy of precipitation data, and then CMPA-Hourly can be used to improve the accuracy of runoff and flood forecasting.

Analysis of lightning flash rate with the occurrence of flash floods and hailstorms in Peninsular Malaysia

This paper will present five flash flood cases in Johor and two hailstorms from Kuala Lumpur and Melaka reported during lightning activities to determine the relationship between lightning flash rate and rainfall that occurred from 2013 to 2017. The flash floods were observed based on lightning data obtained from Tenaga Nasional Berhad Research Sdn. Bhd. (TNBR) and rainfall data from the Department of Irrigation and Drainage (DID). On the other hand, storm-producing hails were observed based on the lightning flash rate using electromagnetic field measurements and weather radar for their reflectivity value and cloud structure. The detection of a lightning flash, especially the negative narrow bipolar event (−NBE) and positive cloud-to-ground (+CG), could be a possible indicator of a deeper convective storm that could lead to the production of hails. In the 24-h observation of selected days when a flash flood occurred, lightning strikes correlation with rainfall had r-values ranging from 0.6 to 0.9, where the lightning data may enhance flood forecasting accuracy when combined with rainfall. Meanwhile, during the first hailstorms in Kuala Lumpur, the highest −NBEs flash rate detected during the hails appeared where 16 −NBEs were detected in a span of 5 min with reflectivity value between 53 and 55 dBz and cloud top height of 20 km. During the second hailstorms in Melaka, more + CG flashes rather than −NBE were detected during the appearance of the hail, with 64 +CG flashes and five −NBEs with reflectivity values between 60 and 66 dBz and cloud top height of 18.8 km. Based on the high convective strength, cloud top height, and presence of hailstones, −NBE and +CG flash could be possible indicators of storm severity, especially in Malaysia.

Comparison of Three Radar-Based Precipitation Nowcasts for the Extreme July 2021 Flooding Event in Germany

Abstract Quantitative precipitation nowcasts (QPN) can improve the accuracy of flood forecasts, especially for lead times up to 12 h, but their evaluation depends on a variety of factors, namely, the choice of the hydrological model and the benchmark. We tested three precipitation nowcasting techniques based on radar observations for the disastrous mid-July 2021 event in seven German catchments (140–1670 km2). Two deterministic [advection-based and spectral prognosis (S-PROG)] and one probabilistic [Short-Term Ensemble Prediction System (STEPS)] QPN with a maximum lead time of 3 h were used as input to two hydrological models: a physically based, 3D-distributed model (ParFlowCLM) and a conceptual, lumped model (GR4H). We quantified the hydrological added value of QPN compared with hydrological persistence and zero-precipitation nowcasts as benchmarks. For the 14 July 2021 event, we obtained the following key results. 1) According to the quality of the forecasted hydrographs, exploiting QPN improved the lead times by up to 4 h (8 h) compared with adopting zero-precipitation nowcasts (hydrological persistence) as a benchmark. Using a skill-based approach, obtained improvements were up to 7–12 h depending on the benchmark. 2) The three QPN techniques obtained similar performances regardless of the applied hydrological model. 3) Using zero-precipitation nowcasts instead of hydrological persistence as benchmark reduced the added value of QPN. These results highlight the need for combining a skill-based approach with an analysis of the quality of forecasted hydrographs to rigorously estimate the added value of QPN.

A distributed hydrological model for semi-humid watersheds with a thick unsaturated zone under strong anthropogenic impacts: A case study in Haihe River Basin

In the past decades, human activities, including hydraulic structure construction and operation, and groundwater overexploitation have largely changed the land surface conditions and correspondingly altered the natural rainfall-runoff processes. These changes make the traditional operational flood forecasting models without considering these anthropogenic impacts fail to capture these changes, leading to a downgraded forecasting capability. To improve the accuracy of flood forecasting in semi-humid watersheds under strong aboveground and underground anthropogenic impacts, the widely used operational flood forecasting model, the Xin’anjiang model, was chosen as a basis to build a new distributed hydrological model accounting for these anthropogenic impacts. To characterize the runoff generation process under the condition of a thick unsaturated zone caused by groundwater overexploitation, this study developed a runoff generation algorithm based on an idea that conceptualizes the soil free-water storage into two virtual reservoirs. To characterize the impacts of manmade reservoirs on the runoff concentration process, we modified the Muskingum method. In addition, we further developed a priori estimation method for deriving the spatial distribution of the tension water storage capacity by considering the anthropogenic impacts and geographical conditions. Based on the above work, a Grid Xin’anjiang-Haihe model (GXH) was developed. In this study, the Qingshui River, which is a tributary of the Haihe River Basin and a typical semi-humid medium-sized watershed in China with poor flood forecasting accuracy, was selected as the study area to test the effectiveness of the model. The results show that the GXH model can simulate the flood peak flow and peak time with higher accuracy than the current operational flood forecasting models and several other models. In addition, the parameter estimation method developed in this study has also proven to be superior in theory and practice.

Exploring a similarity search-based data-driven framework for multi-step-ahead flood forecasting

Due to a small proportion of observations, reliable and accurate flood forecasts for large floods present a fundamental challenge to artificial neural network models, especially when the forecast horizons exceed the flood concentration time of a river basin. This study proposed for the first time a Similarity search-based data-driven framework, and takes the advanced Temporal Convolutional Network based Encoder-Decoder model (S-TCNED) as an example for multi-step-ahead flood forecasting. A total of 5232 hourly hydrological data were divided into two datasets for model training and testing. The input sequence of the model included hourly flood flows of a hydrological station and rainfall data (traced back to the previous 32 h) of 15 gauge stations, and the output sequence stepped into 1- up to 16-hour-ahead flood forecasts. A conventional TCNED model was also built for comparison purposes. The results demonstrated that both TCNED and S-TCNED could make suitable multi-step-ahead flood forecasts, while the proposed S-TCNED model not only could effectively mimic the long-term rainfall-runoff relationship but also could provide more reliable and accurate forecasts of large floods than the TCNED model even in extreme weather conditions. There is a significant positive correlation between the mean sample label density improvement and the mean Nash-Sutcliffe Efficiency (NSE) improvement of the S-TCNED over the TCNED at the long forecast horizons (13 h up to 16 h). Based on the analysis of the sample label density, it is found that the similarity search largely improves the model performance by enabling the S-TCNED model to learn the development process of similar historical floods in a targeted manner. We conclude that the proposed S-TCNED model that converts and associates the previous rainfall-runoff sequence with the forecasting runoff sequence under a similar scenario can enhance the reliability and accuracy of flood forecasts while extending the length of forecast horizons.

Research on Rain Pattern Classification Based on Machine Learning: A Case Study in Pi River Basin

For the purpose of improving the scientific nature, reliability, and accuracy of flood forecasting, it is an effective and practical way to construct a flood forecasting scheme and carry out real-time forecasting with consideration of different rain patterns. The technique for rain pattern classification is of great significance in the above-mentioned technical roadmap. With the rapid development of artificial intelligence technologies such as machine learning, it is possible and necessary to apply these new methods to assist rain classification applications. In this research, multiple machine learning methods were adopted to study the time-history distribution characteristics and conduct rain pattern classification from observed rainfall time series data. Firstly, the hourly rainfall data between 2003 and 2021 of 37 rain gauge stations in the Pi River Basin were collected to classify rain patterns based on the universally acknowledged dynamic time warping (DTW) algorithm, and the classifications were treated as the benchmark result. After that, four other machine learning methods, including the Decision Tree (DT), Long- and Short-Term Memory (LSTM) neural network, Light Gradient Boosting Machine (LightGBM), and Support Vector Machine (SVM), were specifically selected to establish classification models and the model performances were compared. By adjusting the sampling size, the influence of different sizes on the classification was analyzed. Intercomparison results indicated that LightGBM achieved the highest accuracy and the fastest training speed, the accuracy and F1 score were 98.95% and 98.58%, respectively, and the loss function and accuracy converged quickly after only 20 iterations. LSTM and SVM have satisfactory accuracy but relatively low training efficiency, and DT has fast classification speed but relatively low accuracy. With the increase in the sampling size, classification results became stable and more accurate. Besides the higher accuracy, the training efficiency of the four methods was also improved.

Spatial and Temporal Pattern of Rainstorms Based on Manifold Learning Algorithm

Identifying the patterns of rainstorms is essential for improving the precision and accuracy of flood forecasts and constructing flood disaster prevention systems. In this study, we used a manifold learning algorithm method of machine learning to analyze rainstorm patterns. We analyzed the spatial–temporal characteristics of heavy rain in Beijing and Shenzhen. The results showed a strong correlation between the spatial–temporal pattern of rainstorms and underlying topography in Beijing. However, in Shenzhen, the spatial–temporal distribution characteristics of rainstorms were more closely related to the source of water vapor causing the rainfall, and the variation in characteristics was more complex and diverse. This method may be used to quantitatively describe the development and dynamic spatial–temporal patterns of rainfall. In this study, we found that spatial–temporal rainfall distribution characteristics, extracted by machine learning technology could be explained by physical mechanisms consistent with the climatic characteristics and topographic conditions of the region.

Research on particle swarm optimization in LSTM neural networks for rainfall-runoff simulation

Flood forecasting is an essential non-engineering measure for flood prevention and disaster reduction. Many models have been developed to study the complex and highly random rainfall-runoff process. In recent years, artificial intelligence methods, such as the artificial neural network (ANN), have attempted to construct rainfall-runoff models. The more advanced deep learning methods of long short-term memory (LSTM) network have been proved to better predict hydrological time series. However, the selection of LSTM hyperparameters in the past mostly relied on the experience of the staff, which often led to failure to achieve the best performance. The aim of this study is to develop a method to improve flood forecast accuracy and lead time. A deep learning neural network model based on LSTM networks and particle swarm optimization (PSO) is proposed in this paper. The PSO algorithm was used to optimize the LSTM hyperparameter to improve the ability to learn data sequence features. The model focuses on the Jingle Watershed in the Fenhe River and the Lushi Watershed in the Luohe River and was used to predict flood processes using rainfall and runoff observation data from stations in the watersheds. We evaluated the performance of the model with the Nash Sutcliffe efficiency coefficient, root mean square error, and bias. The results show that the PSO-LSTM model outperforms the M-EIES, ANN, PSO-ANN, and LSTM at all stations in the watersheds. The PSO-LSTM model improves the flood forecasting accuracy at different lead times, especially for those exceeding 6 h, and has higher prediction accuracy and stability. The PSO-LSTM model could be used to improve accuracy in short-term flood forecast applications.

A Coupled Dynamic System Inversion Model for Higher Accuracy in Flood Forecasting

AbstractFlood routing is a critical component of real‐time flood forecasting and plays a pivotal role in reducing economic and life losses due to flooding. However, flood routing always possesses errors because it is affected by various sources of uncertainty, which are transmitted and accumulated in the routing process from upstream to downstream, particularly for long or complex river systems. Thus, a real‐time error correction is rapidly becoming a key measure for reducing errors and improving flood forecasting accuracy. In this study, a dynamic system inversion (DSI) model was developed as a new correction method to update forecasting errors for long river systems. The new method utilizes all observed, forecasted, and corrected information at upstream stations and couples the matrix‐based Muskingum routing method with an error‐inversed updating technique to form the DSI equation. Multisource spatiotemporal errors of the river system are simultaneously combined into the DSI equation, and the discharges at the downstream station are reforecasted, thereby providing error correction in flood routing. Based on a case study on the Le'an River (China), the DSI model performed better than the traditional autoregressive (AR) model based on five metrics for both small and large flood events. Moreover, with an increase in lead time, the metrics of AR deteriorated rapidly, whereas that of DSI deteriorated slightly. In conclusion, the proposed DSI model is a sophisticated and robust method for flood routing and provides a new alternative for error correction in flood routing simulations of complex river systems.

Improved Flood Forecasting in Basins With No Precipitation Stations: Constrained Runoff Correction Using Multiple Satellite Precipitation Products

AbstractSatellite‐based precipitation products (SPPs) with short latencies provide a new opportunity for flood forecasting in basins with no precipitation stations. However, the larger uncertainties associated with these near‐real‐time SPPs can influence the accuracy of flood forecasts. Here, we propose a real‐time updating method, referred to as “Constrained Runoff Correction using Multiple SPPs” (CRC‐M). The method is based on the hypothesis that the range of runoff volumes computed using different near‐real‐time SPPs provides an indication of the approximate range in which the true runoff volume lies. Accordingly, a constrained runoff correction is performed using the discharge observed at the basin outlet within this range. Evaluation using real data indicates that the new method performs well with Nash–Sutcliffe (NS) values of 0.85 and 0.87 during calibration and validation, respectively. The necessity of imposing constraints using multiple SPPs is demonstrated by comparing CRC‐M with 2 controls, referred to as “Unconstrained Runoff Correction using Single SPP” (URC‐S) and “Constrained Runoff Correction using Single SPP with perturbations” (CRC‐S). Experiments indicate that the key factors that result in good performance are (a) relatively reliable SPPs and (b) wider constraint ranges. Overall, the CRC‐M method can result in accurate and stable flood forecasts in basins with no precipitation stations, without the shortening of leading time and the need for increased model complexity (i.e., the numbers of model parameters).

Assessing the Potential of Upcoming Satellite Altimeter Missions in Operational Flood Forecasting Systems

This study investigates the potential of observations with improved frequency and latency time of upcoming altimetry missions on the accuracy of flood forecasting and early warnings. To achieve this, we assessed the skill of the forecasts of a distributed hydrological model by assimilating different historical discharge time frequencies and latencies in a framework that mimics an operational forecast system, using the European Ensemble Forecasting system as the forcing. Numerical experiments were performed in 22 sub-basins of the Tocantins-Araguaia Basin. Forecast skills were evaluated in terms of the Relative Operational Characteristics (ROC) as a function of the drainage area and the forecasts’ lead time. The results showed that increasing the frequency of data collection and reducing the latency time (especially 1 d update and low latency) had a significant impact on steep headwater sub-basins, where floods are usually more destructive. In larger basins, although the increased frequency of data collection improved the accuracy of the forecasts, the potential benefits were limited to the earlier lead times.

Improved Streamflow Forecast in a Small-Medium Sized River Basin with Coupled WRF and WRF-Hydro: Effects of Radar Data Assimilation

Accurate and long leading time flood forecasting is very important for flood disaster mitigation. It is an effective method to couple the Quantitative Precipitation Forecast (QPF) products provided by Numerical Weather Prediction (NWP) models to a distributed hydrological model with the goal of extending the leading time for flood forecasting. However, the QPF products contain a certain degree of uncertainty and would affect the accuracy of flood forecasting, especially in the mountainous regions. Radar data assimilation plays an important role in improving the quality of QPF and further improves flood forecasting. In this paper, radar data assimilation was applied in order to construct a high-resolution atmospheric-hydrological coupling model based on the WRF and WRF-Hydro models. Four experiments with conventional observational and radar data assimilation were conducted to evaluate the flood forecasting capability of this coupled model in a small-medium sized basin based on eight typical flood events. The results show that the flood forecast skills are highly QPF-dependent. The QPF from the WRF model is improved by assimilating radar data and further increasing the accuracy of flood forecasting, although both precipitation and flood are slightly over-forecasted. However, the improvements by assimilating conventional observational data are not obvious. In general, radar data assimilation can improve flood forecasting effectively in a small-medium sized basin based on the atmospheric-hydrological coupling model.

Assimilation of probabilistic flood maps from SAR data into a coupled hydrologic–hydraulic forecasting model: a proof of concept

Abstract. Coupled hydrologic and hydraulic models represent powerful tools for simulating streamflow and water levels along the riverbed and in the floodplain. However, input data, model parameters, initial conditions, and model structure represent sources of uncertainty that affect the reliability and accuracy of flood forecasts. Assimilation of satellite-based synthetic aperture radar (SAR) observations into a flood forecasting model is generally used to reduce such uncertainties. In this context, we have evaluated how sequential assimilation of flood extent derived from SAR data can help improve flood forecasts. In particular, we carried out twin experiments based on a synthetically generated dataset with controlled uncertainty. To this end, two assimilation methods are explored and compared: the sequential importance sampling method (standard method) and its enhanced method where a tempering coefficient is used to inflate the posterior probability (adapted method) and reduce degeneracy. The experimental results show that the assimilation of SAR probabilistic flood maps significantly improves the predictions of streamflow and water elevation, thereby confirming the effectiveness of the data assimilation framework. In addition, the assimilation method significantly reduces the spatially averaged root mean square error of water levels with respect to the case without assimilation. The critical success index of predicted flood extent maps is significantly increased by the assimilation. While the standard method proves to be more accurate in estimating the water levels and streamflow at the assimilation time step, the adapted method enables a more persistent improvement of the forecasts. However, although the use of a tempering coefficient reduces the degeneracy problem, the accuracy of model simulation is lower than that of the standard method at the assimilation time step.

Multi-source error correction for flood forecasting based on dynamic system response curve method

Real-time correction is the key to reducing hydrological forecasting errors. Methods of real-time correction can be classified into terminal error correction (TEC) and process error correction (PEC) methods. A state-of-the-art PEC method is the dynamic system response curve (DSRC) method which has been developed to improve flood forecasting, but its underlying assumption of error source limits the correction accuracy. This study developed a multi-source error DSRC method (MSE-DSRC) to partition the total error of forecast discharge into input-caused error and model-caused error. Based on the known quantitative relationship between rain gauge density and error division ratio, two different sources of error are corrected simultaneously. A synthetic case study was done to evaluate the ability of the MSE-DSRC method to correct the variables and parameters of the hydrological model (XAJ model). The MSE-DSRC method was found to overcome the limitations of the traditional DSRC method and have a high accuracy indexed with the Nash-Sutcliffe (NS) efficiency increasing above 0.9. Then, real case studies in three basins of the Huaihe River (China) demonstrated that the MSE-DSRC method achieved higher accuracy and stability than did the traditional DSRC method and can be easily applied in practice, with the average improvement of NS above 50%. Overall, the error division of the MSE-DSRC method opens the possibility for multi-variable updating and multi-source error correction, which may further improve the accuracy of flood forecasting.