Explore the vast range of titles available.

Droughts around the world have resulted in less annual streamflow relative to annual rainfall. The decrease in streamflow cannot be explained by changes in land‐use, precipitation or temperature alone. Rather, the decrease in annual streamflow has been linked to changes in annual rainfall‐partitioning. Climate change is predicted to induce similar, if not worse conditions than previous droughts. This could make the disproportionate shifts in annual streamflow more frequent or intense. Currently, most methods assume annual rainfall‐partitioning remains constant and thus are unlikely to accurately predict how climate change could impact streamflow. Our aim is to conduct a thought‐experiment to examine the various ways climate change could alter annual rainfall‐partitioning and to determine what types of changes in rainfall‐partitioning are likely to be statistically detectable in the coming decades. Using a synthetic streamflow model, we show how changing the rainfall‐runoff intercept, rainfall‐runoff slope, lag‐1 autocorrelation, standard deviation per rainfall depth and, skewness per rainfall depth alters rainfall‐partitioning. Of all the rainfall‐partitioning changes examined only the rainfall‐runoff intercept and slope are likely to be statistically detectable in the coming decades. Changes impacting lag‐1 autocorrelation of annual streamflow, standard deviation per rainfall depth and skewness per rainfall depth are unlikely to be statistically detectable in the coming decades. Although being unlikely to be statistically identified, these changes still alter both the rainfall‐runoff and streamflow‐time relationships. Rainfall‐partitioning changes that alter streamflow and are unlikely to be statistically detected in the coming decades will make prediction and allocation of water resources more challenging.

A holy grail of hydrology is to understand catchment processes well enough that models can provide detailed simulations across a variety of hydrologic settings at multiple spatiotemporal scales, and under changing environmental conditions. Clearly, this cannot be achieved only through intensive place-based investigation at a small number of heavily instrumented catchments, or by empirical methods that do not fully exploit our understanding of hydrology. In this opinion paper, we discuss the need to actively promote and pursue the use of a \"large catchment sample\" approach to modeling the rainfall–runoff process, thereby balancing depth with breadth. We examine the history of such investigations, discuss the benefits (improved process understanding resulting in robustness of prediction at ungauged locations and under change), examine some practical challenges to implementation and, finally, provide perspectives on issues that need to be taken into account as we move forward. Ultimately, our objective is to provoke further discussion and participation, and to promote a potentially important theme for the upcoming Scientific Decade of the International Association of Hydrological Sciences entitled Panta Rhei.

Torrential and long-lasting rainfall often causes long-duration floods in flat and lowland areas in data-scarce Nyaungdon Area of Myanmar, imposing large threats to local people and their livelihoods. As historical hydrological observations and surveys on the impact of floods are very limited, flood hazard assessment and mapping are still lacked in this region, making it hard to design and implement effective flood protection measures. This study mainly focuses on evaluating the predicative capability of a 2D coupled hydrology-inundation model, namely the Rainfall-Runoff-Inundation (RRI) model, using ground observations and satellite remote sensing, and applying the RRI model to produce a flood hazard map for hazard assessment in Nyaungdon Area. Topography, land cover, and precipitation are used to drive the RRI model to simulate the spatial extent of flooding. Satellite images from Moderate Resolution Imaging Spectroradiometer (MODIS) and the Phased Array type L-band Synthetic Aperture Radar-2 onboard Advanced Land Observing Satellite-2 (ALOS-2 ALOS-2/PALSAR-2) are used to validate the modeled potential inundation areas. Model validation through comparisons with the streamflow observations and satellite inundation images shows that the RRI model can realistically capture the flow processes (R2 ≥ 0.87; NSE ≥ 0.60) and associated inundated areas (success index ≥ 0.66) of the historical extreme events. The resultant flood hazard map clearly highlights the areas with high levels of risks and provides a valuable tool for the design and implementation of future flood control and mitigation measures.

The Soil Conservation Service curve number (SCS-CN) method, also known as the Natural Resources Conservation Service curve number (NRCS-CN) method, is popular for computing the volume of direct surface runoff for a given rainfall event. The performance of the SCS-CN method, based on large rainfall ( P ) and runoff ( Q ) datasets of United States watersheds, is evaluated using a large dataset of natural storm events from 27 agricultural plots in India. On the whole, the CN estimates from the National Engineering Handbook (chapter 4) tables do not match those derived from the observed P and Q datasets. As a result, the runoff prediction using former CNs was poor for the data of 22 (out of 24) plots. However, the match was little better for higher CN values, consistent with the general notion that the existing SCS-CN method performs better for high rainfall–runoff (high CN) events. Infiltration capacity (fc) was the main explanatory variable for runoff (or CN) production in study plots as it exhibited the expected inverse relationship between CN and fc. The plot-data optimization yielded initial abstraction coefficient ( λ ) values from 0 to 0.659 for the ordered dataset and 0 to 0.208 for the natural dataset (with 0 as the most frequent value). Mean and median λ values were, respectively, 0.030 and 0 for the natural rainfall–runoff dataset and 0.108 and 0 for the ordered rainfall–runoff dataset. Runoff estimation was very sensitive to λ and it improved consistently as λ changed from 0.2 to 0.03.

Regional rainfall–runoff modeling is an old but still mostly outstanding problem in the hydrological sciences. The problem currently is that traditional hydrological models degrade significantly in performance when calibrated for multiple basins together instead of for a single basin alone. In this paper, we propose a novel, data-driven approach using Long Short-Term Memory networks (LSTMs) and demonstrate that under a “big data” paradigm, this is not necessarily the case. By training a single LSTM model on 531 basins from the CAMELS dataset using meteorological time series data and static catchment attributes, we were able to significantly improve performance compared to a set of several different hydrological benchmark models. Our proposed approach not only significantly outperforms hydrological models that were calibrated regionally, but also achieves better performance than hydrological models that were calibrated for each basin individually. Furthermore, we propose an adaption to the standard LSTM architecture, which we call an Entity-Aware-LSTM (EA-LSTM), that allows for learning catchment similarities as a feature layer in a deep learning model. We show that these learned catchment similarities correspond well to what we would expect from prior hydrological understanding.

Long short-term memory (LSTM) models are recurrent neural networks from the field of deep learning (DL) which have shown promise for time series modelling, especially in conditions when data are abundant. Previous studies have demonstrated the applicability of LSTM-based models for rainfall–runoff modelling; however, LSTMs have not been tested on catchments in Great Britain (GB). Moreover, opportunities exist to use spatial and seasonal patterns in model performances to improve our understanding of hydrological processes and to examine the advantages and disadvantages of LSTM-based models for hydrological simulation. By training two LSTM architectures across a large sample of 669 catchments in GB, we demonstrate that the LSTM and the Entity Aware LSTM (EA LSTM) models simulate discharge with median Nash–Sutcliffe efficiency (NSE) scores of 0.88 and 0.86 respectively. We find that the LSTM-based models outperform a suite of benchmark conceptual models, suggesting an opportunity to use additional data to refine conceptual models. In summary, the LSTM-based models show the largest performance improvements in the north-east of Scotland and in south-east of England. The south-east of England remained difficult to model, however, in part due to the inability of the LSTMs configured in this study to learn groundwater processes, human abstractions and complex percolation properties from the hydro-meteorological variables typically employed for hydrological modelling.

Measuring the impact of climate change on flood frequency is a complex and controversial task. Identifying hydrological changes is difficult given the factors, other than climate variability, which lead to significant variations in runoff series. The catchment filtering role is often overlooked and thus may hinder the correct identification of climate variability signatures on hydrological processes. Does climate variability necessarily imply hydrological variability? This research aims to analytically derive the flood frequency distribution based on realistic hypotheses about the rainfall process and the rainfall–runoff transformation. The annual maximum peak flow probability distribution is analytically derived to quantify the filtering effect of the rainfall–runoff process on climate change. A sensitivity analysis is performed according to typical semi-arid Mediterranean climatic and hydrological conditions, assuming a simple but common scheme for the rainfall–runoff transformation in small-size ungauged catchments, i.e. the CN-SCS model. Variability in annual maximum peak flows and its statistical significance are analysed when changes in the climatic input are introduced. Results show that depending on changes in the annual number of rainfall events, the catchment filtering role is particularly significant, especially when the event rainfall volume distribution is not strongly skewed. Results largely depend on the return period: for large return periods, peak flow variability is significantly affected by the climatic input, while for lower return periods, infiltration processes smooth out the impact of climate change.

Deep learning is becoming an increasingly important way to produce accurate hydrological predictions across a wide range of spatial and temporal scales. Uncertainty estimations are critical for actionable hydrological prediction, and while standardized community benchmarks are becoming an increasingly important part of hydrological model development and research, similar tools for benchmarking uncertainty estimation are lacking. This contribution demonstrates that accurate uncertainty predictions can be obtained with deep learning. We establish an uncertainty estimation benchmarking procedure and present four deep learning baselines. Three baselines are based on mixture density networks, and one is based on Monte Carlo dropout. The results indicate that these approaches constitute strong baselines, especially the former ones. Additionally, we provide a post hoc model analysis to put forward some qualitative understanding of the resulting models. The analysis extends the notion of performance and shows that the model learns nuanced behaviors to account for different situations.

Machine learning (ML) has played an increasing role in the hydrological sciences. In particular, Long Short-Term Memory (LSTM) networks are popular for rainfall–runoff modeling. A large majority of studies that use this type of model do not follow best practices, and there is one mistake in particular that is common: training deep learning models on small, homogeneous data sets, typically data from only a single hydrological basin. In this position paper, we show that LSTM rainfall–runoff models are best when trained with data from a large number of basins.

The rainfall and runoff relationship becomes an intriguing issue as urbanization continues to evolve worldwide. In this paper, we developed a simulation model based on the soil conservation service curve number (SCS-CN) method to analyze the rainfall-runoff relationship in Guangzhou, a rapid growing metropolitan area in southern China. The SCS-CN method was initially developed by the Natural Resources Conservation Service (NRCS) of the United States Department of Agriculture (USDA), and is one of the most enduring methods for estimating direct runoff volume in ungauged catchments. In this model, the curve number (CN) is a key variable which is usually obtained by the look-up table of TR-55. Due to the limitations of TR-55 in characterizing complex urban environments and in classifying land use/cover types, the SCS-CN model cannot provide more detailed runoff information. Thus, this paper develops a method to calculate CN by using remote sensing variables, including vegetation, impervious surface, and soil (V-I-S). The specific objectives of this paper are: (1) To extract the V-I-S fraction images using Linear Spectral Mixture Analysis; (2) To obtain composite CN by incorporating vegetation types, soil types, and V-I-S fraction images; and (3) To simulate direct runoff under the scenarios with precipitation of 57mm (occurred once every five years by average) and 81mm (occurred once every ten years). Our experiment shows that the proposed method is easy to use and can derive composite CN effectively.