Regionalization of hydrological model parameters using gradient boosting machine

Regionalization of hydrological model parameters is key to hydrological predictions in ungauged basins. The commonly used multiple linear regression (MLR) method may not be applicable in complex and nonlinear relationships between model parameters and watershed properties. Moreover, most regionalization methods assume lumped parameters for each catchment without considering within-catchment heterogeneity. Here we incorporated the Penman-Monteith-Leuning (PML) equation into the Distributed Time-Variant Gain Model (DTVGM) to improve the mechanistic representation of the evapotranspiration process. We calibrated six key model parameters grid-by-grid across China using a multivariable calibration strategy, which incorporates spatiotemporal runoff and evapotranspiration (ET) datasets (0.25°, monthly) as reference. In addition, we used the gradient boosting machine (GBM), a machine learning technique, to portray the dependence of model parameters on soil and terrain attributes in four distinct climatic zones across China. We show that the modified DTVGM could reasonably estimate the runoff and ET over China using the calibrated parameters, but performed better in humid than arid regions for the validation period. The regionalized parameters by the GBM method exhibited better spatial coherence relative to the calibrated grid-by-grid parameters. In addition, GBM outperformed the stepwise MLR method in both parameter regionalization and gridded runoff simulations at national scale, though the improvement is not significant pertaining to watershed streamflow validation due to most of the watersheds being located in humid regions. We also revealed that the slope, saturated soil moisture content, and elevation are the most important explanatory variables to inform model parameters based on the GBM approach. The machine-learning-based regionalization approach provides an effective alternative to deriving hydrological model parameters by using watershed properties in ungauged regions.

Similarity-based approaches in hydrogeology: proposal of a new concept for data-scarce groundwater resource characterization and prediction

A new concept is proposed for describing, analysing and predicting the dynamic behaviour of groundwater resources based on classification and similarity. The concept makes use of the ideas put forward by the “PUB” (predictions in ungauged basins) initiative in surface-water hydrology. One of the approaches developed in PUB uses the principle that similar catchments, exposed to similar weather conditions, will generate a similar discharge response at the catchment outlet. This way, models developed for well-observed catchments can be used to make predictions for ungauged catchments with similar properties (topography, land use, etc.). The concept proposed here applies the same idea to groundwater systems, with the goal to make predictions of the dynamic behaviour of groundwater in poorly observed systems using similarities to well-observed and understood systems. This paper gives an overview of the main ideas, the methodological background, the progress so far, and the challenges that the authors regard as most crucial for further development. One of the main goals of this article is thus to raise interest for this new concept within the groundwater community. There are a multitude of highly interesting aspects to investigate, and a community effort, as with PUB, is required. A second goal is to foster and exchange ideas between the groundwater and surface water research communities who, while often working on similar problems, have often missed the opportunity to learn from each other.

Applying Non-Parametric Bayesian Network to estimate monthly maximum river discharge: potential and challenges

Non-Parametric Bayesian Networks (NPBNs) are graphical tools for statistical inference widely used for reliability analysis and risk assessment. However, few hydrological applications can be found in the literature. We therefore explore here the potential of NPBNs to reproduce catchment-scale hydrological dynamics by investigating 240 catchments with contrasting climate across the United States from the CAMELS dataset. First, two networks, one unsaturated (UN-1) and one saturated network (SN-1) based on hydro-meteorological variables are used to generate monthly maximum river discharge considering the catchment as a single element. Then, the saturated network SN-C, based on SN-1 but additionally including physical catchments attributes, is used to model a group of catchments and infer monthly maximum river discharge in ungauged basins based on the attributes similarity. The results indicate that the UN-1 model is suitable for catchments with a positive dependence between precipitation and river discharge, while the SN-1 model can reproduce discharge also in catchments with negative dependence. Furthermore, in ~40 % of the catchments analysed the SN-1 model can reproduce statistical characteristics of discharge, tested via the Kolmogorov-Smirnov (KS) statistic, and Nash-Sutcliffe Efficiencies (NSE) ≥ 0.5. Such catchments receive precipitation mainly in winter and are located in energy-limited regions at low to moderate elevation. Further, the SN-C model, in which the inference process benefits from catchment similarity, can reproduce river discharge statistics in ~10 % of the catchments analysed. However, in these catchments a common dominant physical attribute was not identified. In this study, we show that, once a NPBNs is defined, it is straightforward to infer discharge, when the remaining variables are known. We also show that it is possible to extend the network itself with additional variables, i.e. going from SN-1 to SN-C. Despite these advantages, the results also suggest that there are considerable challenges in defining a suitable NPBN, in particular for predictions in ungauged basins. These are mainly due to the discrepancies in the time scale of the different physical processes generating discharge, the presence of a “memory” in the system, and the Gaussian-copula assumption used by NPBNs for modelling multivariate dependence.

The importance of geology when estimating catchment-scale streamflow characteristics: Application of a new technique for hydrologic similarity and regionalization

<p>Expanding the scientific understanding of global hydrological processes is a key research area for hydrologists. Research in this area can allow hydrologists to make better predictions in ungauged basins and catchments under climate change scenarios. Though hydrological processes are largely understood at a laboratory-scale, catchment-scale processes are much more complex and unknown. Previous studies at the catchment-scale have shown catchment geology is largely irrelevant in determining components of streamflow. Laboratory-scale experiments, however, have revealed that this is unlikely. This contradiction indicates the current techniques for determining hydrological variable importance in the literature are insufficient. In this paper, we quantify the influence of the interaction amongst climatic, geological, and topographical features on a large set of hydrological signatures in snow-dominated regions across North America, using Stable Extrapolative Marginal Contribution Feature Importance. The preliminary results show that when we consider interaction effects among climatic and geophysical features, and remove the influence of correlation, geological features show considerable importance at the catchment scale. We contend that this study contributes to the scientific understanding of catchment-scale hydrological processes, especially in cold, ungauged basins.</p>

Can the Budyko framework and satellite data help improve hydrological modeling in ungauged and poorly monitored catchments? The case study of the Lurín catchment in Peru

<p>Predictions in ungauged basins still present one of the major challenges in hydrology. In many cases, the absence of a stream gauge also implies a low density of the meteorological monitoring network in these catchments and surroundings as well as little available data on water management infrastructure and agricultural consumptions. This combination creates a circle of uncertainties and thus individual influences of relevant water balance components are difficult to disentangle and quantify. </p><p>The original Budyko curve presents a very general model that yields, to first order, an estimate of the steady-state water balance of a catchment at the climatological scale, assuming its landscape and functioning has evolved naturally and free of anthropogenic interferences. Even at smaller time scales, the Budyko relationship allows approximating the water partitioning in the catchment, and thus helps correct erroneous assumptions[JW1]  or missing information about for instance unknown human-induced alterations. On the other hand, an increasing variety of global remote-sensing data products is becoming available providing spatial estimates of land surface properties such as for instance vegetation indexes or soil moisture. Even if the predictive power of such products in terms of absolute values remains questionable, it is possible to derive coarse spatial patterns or temporal dynamics to narrow down zones and orders of magnitude of interferences with the natural hydrological cycle such as reservoirs or irrigated lands. This study combines these two general approaches in order to improve hydrological modelling and system understanding of the semi-arid Lurín catchment in the Western Andes of Peru.</p>

Using remote sensing evapotranspiration solely calibrating hydrology model for predicting runoff time series in ungauged regions

<p>It is important yet challenging to predict runoff in data sparse regions or ungauged regions, majority of which belong to headwater catchments that are normally the major water source for middle and lower river reaches. There are numerous studies carried out since the launch of the Predictions in Ungauged Basins (PUB) initiative by the International Association of Hydrological Sciences (IAHS) in 2003. Most runoff prediction studies rely on modelling approaches via two steps. The first step is to calibrate the hydrological model against observed streamflow at the gauged catchments. The second step is regionalization in which the set of calibrated parameter values from a suitable donor catchment is used for predicting runoff in a targeted ungauged catchment. The major challenge of this approach is that when the gauged catchments are sparsely distributed or little available, it is hard to get sensible regionalization results. This study develops a new approach to calibrate a hydrological model purely against remote sensed actual evapotranspiration data obtained from 8-day and 500 m resolution PML-V2 products and the calibrated parameters can be directly used for runoff prediction across global land surface. This approach has been successfully used for predicting daily, monthly and annual runoff in Australia and southeastern Tibetan Plateau. This is an exciting research domain for hydrologists to pursue since remote sensing data is accumulated in a fast-increasing rate, and will provide researchers an unprecedent opportunity.</p>

Understanding dominant controls on streamflow spatial variability to set up a semi-distributed hydrological model: the case study of the Thur catchment

This study documents the development of a semi-distributed hydrological model aimed at reflecting the dominant controls on observed streamflow spatial variability. The process is presented through the case study of the Thur catchment (Switzerland, 1702 km2), an alpine and pre-alpine catchment where streamflow (measured at 10 subcatchments) has different spatial characteristics in terms of amounts, seasonal patterns, and dominance of baseflow. In order to appraise the dominant controls on streamflow spatial variability and build a model that reflects them, we follow a two-stage approach. In a first stage, we identify the main climatic or landscape properties that control the spatial variability of streamflow signatures. This stage is based on correlation analysis, complemented by expert judgement to identify the most plausible cause–effect relationships. In a second stage, the results of the previous analysis are used to develop a set of model experiments aimed at determining an appropriate model representation of the Thur catchment. These experiments confirm that only a hydrological model that accounts for the heterogeneity of precipitation, snow-related processes, and landscape features such as geology produces hydrographs that have signatures similar to the observed ones. This model provides consistent results in space–time validation, which is promising for predictions in ungauged basins. The presented methodology for model building can be transferred to other case studies, since the data used in this work (meteorological variables, streamflow, morphology, and geology maps) are available in numerous regions around the globe.

Enhancing Physical Similarity Approach to Predict Runoff in Ungauged Watersheds in Sub-Tropical Regions

Regionalization techniques have been comprehensively discussed as the solution for runoff predictions in ungauged basins (PUB). Several types of regionalization approach have been proposed during the years. Among these, the physical similarity one was demonstrated to be one of the most robust. However, this method cannot be applied in large regions characterized by highly variable climatic conditions, such as sub-tropical areas. Therefore, this study aims to develop a new regionalization approach based on an enhanced concept of physical similarity to improve the runoff prediction of ungauged basins at country scale, under highly variable-weather conditions. A clustering method assured that watersheds with different hydrologic and physical characteristics were considered. The novelty of the proposed approach is based on the relationships found between rainfall-runoff model parameters and watershed-physiographic factors. These relationships were successively exported and validated at the ungauged basins. From the overall results, it can be concluded that the runoff prediction in the ungauged basins was very satisfactory. Therefore, the proposed approach can be adopted as an alternative method for runoff prediction in ungauged basins characterized by highly variable-climatic conditions.