Explore the vast range of titles available.

Global climate model (GCM) output typically needs to be bias corrected before it can be used for climate change impact studies. Three existing bias correction methods, and a new one developed here, are applied to daily maximum temperature and precipitation from 21 GCMs to investigate how different methods alter the climate change signal of the GCM. The quantile mapping (QM) and cumulative distribution function transform (CDF-t) bias correction methods can significantly alter the GCM’s mean climate change signal, with differences of up to 2°C and 30% points for monthly mean temperature and precipitation, respectively. Equidistant quantile matching (EDCDFm) bias correction preserves GCM changes in mean daily maximum temperature but not precipitation. An extension to EDCDFm termed PresRat is introduced, which generally preserves the GCM changes in mean precipitation. Another problem is that GCMs can have difficulty simulating variance as a function of frequency. To address this, a frequency-dependent bias correction method is introduced that is twice as effective as standard bias correction in reducing errors in the models’ simulation of variance as a function of frequency, and it does so without making any locations worse, unlike standard bias correction. Last, a preconditioning technique is introduced that improves the simulation of the annual cycle while still allowing the bias correction to take account of an entire season’s values at once.

Precipitation estimation over the Tibetan Plateau is a critical but challenging task due to sparse gauges and high altitudes. Traditional statistic methods are often insufficient to characterize the nonlinear relationship between different precipitation information, while machine learning techniques, particularly deep learning algorithms, offer a novel and powerful approach to improve the merging accuracy of multi‐source precipitation data by efficiently capturing their spatiotemporal dynamics features. This study introduced a novel strategy called Double Machine Learning (DML), which integrates meteorological information, satellite retrievals, and reanalysis data to produce a high‐precision multi‐source merging precipitation product at 0.1° × 0.1°, daily resolution for the Tibetan Plateau. The quantitative evaluation of DML was accomplished using both auto‐meteorological gauges and independent observations. Statistical scores indicate that the new DML‐based merging product apparently outperforms three widely‐used precipitation datasets (IMERG‐Final, GSMaP‐Gauge and ERA5) over the Tibetan Plateau. The proposed DML strategy effectively integrates the advantages of traditional machine learning and deep learning, significantly enhancing the algorithmic robustness and merging accuracy, particularly at medium‐high rain rates in summer. Furthermore, the contributions of multi‐source inputs to the final merging effect was systematically analyzed. It is found that meteorological information, as an auxiliary variable in DML, plays a crucial role in identifying rainy events and adjusting the bias of precipitation estimates, especially over those ungauged regions. This study affirms the call for improving the multi‐source precipitation estimates by combining different machine learning approaches. The new merging precipitation product reported here is recommended for hydrometeorological users of the Tibetan Plateau science community. Key Points A novel strategy is developed to produce a high‐precision precipitation product (0.1°/daily, 2015‐present) for the Tibetan Plateau The new merged product apparently outperforms three widely‐used precipitation datasets, especially at medium‐high rain rates The contributions of meteorological information, satellite retrievals and reanalysis data to merging effect were systematically evaluated

Rainfall–runoff modelling is complicated due to numerous complex interactions and feedback in the water cycle among precipitation and evapotranspiration processes, and also geophysical characteristics. Consequently, the lack of geophysical characteristics such as soil properties leads to difficulties in developing physical and analytical models when traditional statistical methods cannot simulate rainfall–runoff accurately. Machine learning techniques with data-driven methods, which can capture the nonlinear relationship between prediction and predictors, have been rapidly developed in the last decades and have many applications in the field of water resources. This study attempts to develop a novel 1D convolutional neural network (CNN), a deep learning technique, with a ReLU activation function for rainfall–runoff modelling. The modelling paradigm includes applying two convolutional filters in parallel to separate time series, which allows for the fast processing of data and the exploitation of the correlation structure between the multivariate time series. The developed modelling framework is evaluated with measured data at Chau Doc and Can Tho hydro-meteorological stations in the Vietnamese Mekong Delta. The proposed model results are compared with simulations of long short-term memory (LSTM) and traditional models. Both CNN and LSTM have better performance than the traditional models, and the statistical performance of the CNN model is slightly better than the LSTM results. We demonstrate that the convolutional network is suitable for regression-type problems and can effectively learn dependencies in and between the series without the need for a long historical time series, is a time-efficient and easy to implement alternative to recurrent-type networks and tends to outperform linear and recurrent models.

Paciorek and Wehner raise important questions around our use of the Mann‐Kendall nonparametric trend test on smoothed data for analyzing long‐term hydrometeorological trends in Zhang et al. (2021, https://doi.org/10.1029/2020gl092293). We thank them for initiating this important conversation and their gracious cooperation in exploring the issues addressed in their comment. In this reply we confirm the inflation of significant p‐values by our choice to smooth, illustrate the relatively minor impacts on the main conclusions of our paper, and add our voices to those of Paciorek and Wehner in highlighting the lack of methodology for hypothesis testing across multiple stations that have spatial structure (i.e., testing for regionally consistent trends). Plain Language Summary Our colleagues Drs. Paciorek and Wehner have raised concerns about our paper (Zhang et al., 2021, https://doi.org/10.1029/2020gl092293), which showed widespread increases in the duration of drought events over the last five decades in the western United States. They point out that our decision to smooth the data using a moving average inflated the number of weather stations at which the trends toward longer droughts were deemed significant by a statistical test. We agree with them on this point, and we have recomputed all our results using unsmoothed data to determine the impacts. We find that for most stations and regions, trend magnitudes remained largely unchanged, with many stations nearby one another often suggesting similar trends. Finally, we agree with Paciorek and Wehner that there is a lack of statistical methods to test such coherent regional patterns, and we caution that over‐reliance on p‐values limits the power of regional data to identify important climate trends. Key Points We agree that smoothing to 5‐year moving windows introduced serial correlation into time series of annual statistics of daily rainfall data, inflating the number of weather stations individually showing significant trends (p < 0.05) with the Mann‐Kendall test Recomputation with unsmoothed values produced substantially the same dry intervals trend magnitude and direction at most stations individually and had only minimal impacts on dry interval trends computed for National Ecological Observatory Network domains using the Regional Kendall test No perfect statistical approach leverages the capacity of coherent regional patterns among spatially correlated weather stations, and an over‐reliance on p‐values as a binary (significant vs. insignificant) determinant of trends limits the power of regional data

The skill of North American multimodel ensemble (NMME) seasonal forecasts in East Africa (EA), which encompasses one of the most food and water insecure areas of the world, is evaluated using deterministic, categorical, and probabilistic evaluation methods. The skill is estimated for all three primary growing seasons: March–May (MAM), July–September (JAS), and October–December (OND). It is found that the precipitation forecast skill in this region is generally limited and statistically significant over only a small part of the domain. In the case of MAM (JAS) [OND] season it exceeds the skill of climatological forecasts in parts of equatorial EA (Northern Ethiopia) [equatorial EA] for up to 2 (5) [5] months lead. Temperature forecast skill is generally much higher than precipitation forecast skill (in terms of deterministic and probabilistic skill scores) and statistically significant over a majority of the region. Over the region as a whole, temperature forecasts also exhibit greater reliability than the precipitation forecasts. The NMME ensemble forecasts are found to be more skillful and reliable than the forecast from any individual model. The results also demonstrate that for some seasons (e.g. JAS), the predictability of precipitation signals varies and is higher during certain climate events (e.g. ENSO). Finally, potential room for improvement in forecast skill is identified in some models by comparing homogeneous predictability in individual NMME models with their respective forecast skill.

Coastal areas epitomize the notion of ‘at-risk’ territory in the context of climate change and sea level rise (SLR). Knowledge of the water level changes at the coast resulting from the mean sea level variability, tide, atmospheric surge and wave setup is critical for coastal flooding assessment. This study investigates how coastal water level can be altered by interactions between SLR, tides, storm surges, waves and flooding. The main mechanisms of interaction are identified, mainly by analyzing the shallow water equations. Based on a literature review, the orders of magnitude of these interactions are estimated in different environments. The investigated interactions exhibit a strong spatiotemporal variability. Depending on the type of environments (e.g., morphology, hydrometeorological context), they can reach several tens of centimeters (positive or negative). As a consequence, probabilistic projections of future coastal water levels and flooding should identify whether interaction processes are of leading order, and, where appropriate, projections should account for these interactions through modeling or statistical methods.

Gridded precipitation and temperature products are inherently uncertain because of myriad factors, including interpolation from a sparse observation network, measurement representativeness, and measurement errors. Generally uncertainty is not explicitly accounted for in gridded products of precipitation or temperature; if it is represented, it is often included in an ad hoc manner. A lack of quantitative uncertainty estimates for hydrometeorological forcing fields limits the application of advanced data assimilation systems and other tools in land surface and hydrologic modeling. This study develops a gridded, observation-based ensemble of precipitation and temperature at a daily increment for the period 1980–2012 for the conterminous United States, northern Mexico, and southern Canada. This allows for the estimation of precipitation and temperature uncertainty in hydrologic modeling and data assimilation through the use of the ensemble variance. Statistical verification of the ensemble indicates that it has generally good reliability and discrimination of events of various magnitudes but has a slight wet bias for high threshold events (>50 mm). The ensemblemean is similar to other widely used hydrometeorological datasets but with some important differences. The ensemble product produces a more realistic occurrence of precipitation statistics (wet day fraction), which impacts the empirical derivation of other fields used in land surface and hydrologic modeling. In terms of applications, skill in simulations of streamflow in 671 headwater basins is similar to other coarse-resolution datasets. This is the first version, and future work will address temporal correlation of precipitation anomalies, inclusion of other data streams, and examination of topographic lapse rate choices.

Defining homogeneous precipitation regions is fundamental for hydrologic applications, yet nontrivial, particularly for regions with highly varied spatial–temporal patterns. Traditional approaches typically include aspects of subjective delineation around sparsely distributed precipitation stations. Here, hierarchical and nonhierarchical (k means) clustering techniques on a gridded dataset for objective and automatic delineation are evaluated. Using a spatial sensitivity analysis test, the k-means clustering method is found to produce much more stable cluster boundaries. To identify a reasonable optimal k, various performance indicators, including the within-cluster sum of square errors (WSS) metric, intra- and intercluster correlations, and postvisualization are evaluated. Two new objective selection metrics (difference in minimum WSS and difference in difference) are developed based on the elbow method and gap statistics, respectively, to determine k within a desired range. Consequently, eight homogenous regions are defined with relatively clear and smooth boundaries, as well as low intercluster correlations and high intracluster correlations. The underlying physical mechanisms for the regionalization outcomes not only help justify the optimal number of clusters selected, but also prove informative in understanding the local- and large-scale climate factors affecting Ethiopian summertime precipitation. A principal component linear regression model to produce cluster-level seasonal forecasts also proves skillful.

Design flood corresponding to probable maximum precipitation (PMP) is often used in risk assessment of large hydraulic structures. PMP is widely estimated using the Hershfield method (HM) when reasonably long precipitation records are available, but other hydrometeorological data are unavailable/limited. Uncertainty in HM‐based PMP estimates can arise from multiple sources, but the literature lacks methodologies to identify significant contributors to the uncertainty and determine the overall uncertainty bounds. To address this research gap, a novel framework based on imprecise probability (IP) theory is proposed. It facilitates quantifying the overall uncertainty in PMP estimates arising from multiple sources and discerning contributions from individual sources and their combinations. Uncertainties analyzed include those due to (i) the sampling effect of precipitation observations, and (ii) 24 combinations of options for (a) defining a meaningful region/zone, (b) envelope curve space preparation, and (c) the curve construction. The effectiveness of the proposed framework in determining IP bounds of PMP estimates is illustrated through case studies on two major flood‐prone river basins (Mahanadi and Godavari) in India and the Brazos River basin in the United States. Results revealed that the highest contributor to the overall uncertainty in PMP estimates is the combined/interaction component of all four uncertainty sources for the Indian river basins and the statistical sampling effect for the United States basin. Whereas the least contribution to uncertainty is consistently from envelope curve construction. Options/guidelines are provided to reduce the uncertainty (interval between IP bounds) arising from different sources in PMP estimation with HM.