Acquisition of Rainfall in Ungauged Basins: a Study of Rainfall Distribution Heterogeneity Based on a New Method

High-density precipitation data is always desired to capture the heterogeneity of precipitation to accurately describe the components of the hydrological cycle. However, equipping and maintaining a high-density rain gauge network involves high costs, and the existing rain gauges are often unable to meet the density requirements. The objective of this study is to provide a new method to analyze the spatiotemporal variability of the precipitation field and to solve the problem of insufficient site density. To this end, the Proper Orthogonal Decomposition (POD) method is proposed, which can analyze the spatial distribution characteristics of rainfall fields to solve data shortages. To demonstrate the feasibility and advantages of the proposed methodology, four districts and counties (Hongshan District, Jianli County, Sui County, and Xuanen County) in Hubei province in China were selected as case studies. The principal results are as follows. (1) The proposed method is effective in analyzing the spatiotemporal variability of the rainfall field to reconstruct rainfall data in ungauged basins. (2) Compared with the commonly used Thiessen Polygon method, the Inverse Distance Weighting method, and the Kriging method, POD is more accurate and convenient, and the root mean squared error is reduced from 3.22, 1.83, 2.19 to 2.09; the correlation coefficients are improved from 0.60, 0.85, 0.79 to 0.89, respectively. (3) The POD method performs particularly well in simulating the peak value and the peak time and can offer a meaningful reference for analyzing the spatial distribution of rainfall.

Addressing Spatial Heterogeneity in the Discrete Generalized Nash Model for Flood Routing

River flood routing is one of the key components of hydrologic modeling and the topographic heterogeneity of rivers has great effects on it. It is beneficial to take into consideration such spatial heterogeneity, especially for hydrologic routing models. The discrete generalized Nash model (DGNM) based on the Nash cascade model has the potential to address spatial heterogeneity by replacing the equal linear reservoirs into unequal ones. However, it seems impossible to obtain the solution of this complex high order differential equation directly. Alternatively, the strict mathematical derivation is combined with the deeper conceptual interpretation of the DGNM to obtain the heterogeneous DGNM (HDGNM). In this work, the HDGNM is explicitly expressed as a linear combination of the inflows and outflows, whose weight coefficients are calculated by the heterogeneous S curve. Parameters in HDGNM can be obtained in two different ways: optimization by intelligent algorithm or estimation based on physical characteristics, thus available to perform well in both gauged and ungauged basins. The HDGNM expands the application scope, and becomes more applicable, especially in river reaches where the river slopes and cross-sections change greatly. Moreover, most traditional routing models are lumped, whereas the HDGNM can be developed to be semidistributed. The middle Hanjiang River in China is selected as a case study to test the model performance. The results show that the HDGNM outperforms the DGNM in terms of model efficiency and smaller relative errors and can be used also for ungauged basins.

Estimating Baseflow and Baseflow Index in Ungauged Basins Using Spatial Interpolation Techniques: A Case Study of the Southern River Basin of Thailand

This research aims to estimate baseflow (BF) and baseflow index (BFI) in ungauged basins in the southern part of Thailand. Three spatial interpolation methods (namely, inverse distance weighting (IDW), kriging, and spline) were utilized and compared in regard to their performance. Two baseflow separation methods, i.e., the local minimum method (LM) and the Eckhardt filter method (EF), were investigated. Runoff data were collected from 65 runoff stations. These runoff stations were randomly selected and divided into two parts: 75% and 25% for the calibration and validation stages, respectively, with a total of 36 study cases. Four statistical indices including mean absolute error (MAE), root mean squared error (RMSE), correlation coefficient (r), and combined accuracy (CA), were applied for the performance evaluation. The findings revealed that monthly and annual BF and BFI calculated by EF were mostly lower than those calculated by LM. Furthermore, IDW gave the best performance among the three spatial interpolation techniques by providing the highest r-value and the lowest MAE, RMSE, and CA values for both the calibration and validation stages, followed by kriging and spline, respectively. We also provided monthly and annual BF and BFI maps to benefit water resource management.

Using machine learning methods for supporting GR2M model in runoff estimation in an ungauged basin

Estimating monthly runoff variation, especially in ungauged basins, is inevitable for water resource planning and management. The present study aimed to evaluate the regionalization methods for determining regional parameters of the rainfall-runoff model (i.e., GR2M model). Two regionalization methods (i.e., regression-based methods and distance-based methods) were investigated in this study. Three regression-based methods were selected including Multiple Linear Regression (MLR), Random Forest (RF), and M5 Model Tree (M5), and two distance-based methods included Spatial Proximity Approach and Physical Similarity Approach (PSA). Hydrological data and the basin's physical attributes were analyzed from 37 runoff stations in Thailand's southern basin. The results showed that using hydrological data for estimating the GR2M model parameters is better than using the basin's physical attributes. RF had the most accuracy in estimating regional GR2M model’s parameters by giving the lowest error, followed by M5, MLR, SPA, and PSA. Such regional parameters were then applied in estimating monthly runoff using the GR2M model. Then, their performance was evaluated using three performance criteria, i.e., Nash–Sutcliffe Efficiency (NSE), Correlation Coefficient (r), and Overall Index (OI). The regionalized monthly runoff with RF performed the best, followed by SPA, M5, MLR, and PSA. The Taylor diagram was also used to graphically evaluate the obtained results, which indicated that RF provided the products closest to GR2M's results, followed by SPA, M5, PSA, and MLR. Our finding revealed the applicability of machine learning for estimating monthly runoff in the ungauged basins. However, the SPA would be recommended in areas where lacking the basin's physical attributes and hydrological information.

Streamflow Estimation at Ungauged Basin using Modified Group Method of Data Handling

Among the foremost frequent and vital tasks for hydrologist is to deliver a high accuracy estimation on the hydrological variable, which is reliable. It is essential for flood risk evaluation project, hydropower development and for developing efficient water resource management. Presently, the approach of the Group Method of Data Handling (GMDH) has been widely applied in the hydrological modelling sector. Yet, comparatively, the same tool is not vastly used for the hydrological estimation at ungauged basins. In this study, a modified GMDH (MGMDH) model was developed to ameliorate the GMDH model performance on estimating hydrological variable at ungauged sites. The MGMDH model consists of four transfer functions that include polynomial, hyperbolic tangent, sigmoid and radial basis for hydrological estimation at ungauged basins; as well as; it incorporates the Principal Component Analysis (PCA) in the GMDH model. The purpose of PCA is to lessen the complexity of the GMDH model; meanwhile, the implementation of four transfer functions is to enhance the estimation performance of the GMDH model. In evaluating the effectiveness of the proposed model, 70 selected basins were adopted from the locations throughout Peninsular Malaysia. A comparative study on the performance was done between the MGMDH and GMDH model as well as with other extensively used models in the area of flood quantile estimation at ungauged basins known as Linear Regression (LR), Nonlinear Regression (NLR) and Artificial Neural Network (ANN). The results acquired demonstrated that the MGMDH model possessed the best estimation with the highest accuracy comparatively among all models tested. Thus, it can be deduced that MGMDH model is a robust and efficient instrument for flood quantiles estimation at ungauged basins.

Hydrological Similarity-Based Parameter Regionalization under Different Climate and Underlying Surfaces in Ungauged Basins

Hydrological similarity-based parameter regionalization is the dominant method used for runoff prediction in ungauged basin. However, the application of this approach depends on assessing hydrological similarity between basins. This study used data for runoff, climate, and the underlying surface of the Hulan River Basin and Poyang Lake Basin to construct a novel physical hydrological similarity index (HSI). The index was used to compare the efficiency of transfer of the parameters of commonly used regionalization methods and to finally apply parameters to ungauged basins. The results showed that: (1) Precipitation is the main climatic factor regulating magnitude of runoff in the Poyang Lake Basin. Spring runoff in Hulan River Basin was regulated by precipitation and temperature. (2) The GR4J and CemaNeigeGR4J models achieved reasonable simulations of runoff of Poyang Lake Basin and Hulan River Basin. Although CemaNeigeGR4J considers snowmelt, the model simulations of spring runoff in the Hulan River Basin were not accurate. (3) There was a significant correlation between climate, the underlying surface, and hydrological model parameters. There were fewer significant correlations between environmental factors and between environmental factors and hydrological model parameters in the Hulan River Basin compared to those in the Poyang Lake Basin, possibly due to less sub-basins in the Hulan River Basin. (4) The HSI based on a combination of principal component analysis and the entropy method efficiently identified the most similar gauged basin for an ungauged basin. A significant positive correlation existed between the HSI and parameter transfer efficiency. The relationship between the HSI and transfer efficiency could be represented by logistic regression and linear regression in the Poyang Lake Basin and Hulan River Basin, respectively. The HSI was better able to quantify the hydrological similarity between basins in terms of climate and underlying surface and can provide a scientific reference for the transfer of hydrological model parameters in an ungauged basin.