Finding the most appropriate precipitation probability distribution for stochastic weather generation and hydrological modelling in Nordic watersheds

Six precipitation probability distributions (exponential, Gamma, Weibull, skewed normal, mixed exponential and hybrid exponential/Pareto distributions) are evaluated on their ability to reproduce the statistics of the original observed time series. Each probability distribution is also indirectly assessed by looking at its ability to reproduce key hydrological variables after being used as inputs to a lumped hydrological model. Data from 24 weather stations and two watersheds (Chute‐du‐Diable and Yamaska watersheds) in the province of Quebec (Canada) were used for this assessment. Various indices or statistics, such as the mean, variance, frequency distribution and extreme values are used to quantify the performance in simulating the precipitation and discharge. Performance in reproducing key statistics of the precipitation time series is well correlated to the number of parameters of the distribution function, and the three‐parameter precipitation models outperform the other models, with the mixed exponential distribution being the best at simulating daily precipitation. The advantage of using more complex precipitation distributions is not as clear‐cut when the simulated time series are used to drive a hydrological model. Although the advantage of using functions with more parameters is not nearly as obvious, the mixed exponential distribution appears nonetheless as the best candidate for hydrological modelling. The implications of choosing a distribution function with respect to hydrological modelling and climate change impact studies are also discussed. Copyright © 2012 John Wiley & Sons, Ltd.     

Spatio-temporal stochastic modelling (METMAVI)

Editorial letter for the Special Issue dedicated to the VI International Workshop on Spatio-temporal Modelling (METMAVI), which took place in Guimarães, Portugal, from 12 to 14 September 2012. This SI summarizes the main contributions made at METMAVI, related to spatio-temporal methodology illustrated with environmental applications.     

Spatio-temporal interpolation of precipitation during monsoon periods in Pakistan

Spatio-temporal estimation of precipitation over a region is essential to the modeling of hydrologic processes for water resources management. The changes of magnitude and space–time heterogeneity of rainfall observations make space–time estimation of precipitation a challenging task. In this paper we propose a Box–Cox transformed hierarchical Bayesian multivariate spatio-temporal interpolation method for the skewed response variable. The proposed method is applied to estimate space–time monthly precipitation in the monsoon periods during 1974–2000, and 27-year monthly average precipitation data are obtained from 51 stations in Pakistan. The results of transformed hierarchical Bayesian multivariate spatio-temporal interpolation are compared to those of non-transformed hierarchical Bayesian interpolation by using cross-validation. The software developed by [11] is used for Bayesian non-stationary multivariate space–time interpolation. It is observed that the transformed hierarchical Bayesian method provides more accuracy than the non-transformed hierarchical Bayesian method.     

Forecasting monthly precipitation using sequential modelling

In the hydrological cycle, rainfall is a major component and plays a vital role in planning and managing water resources. In this study, new generation deep learning models, recurrent neural network (RNN) and long short-term memory (LSTM), were applied for forecasting monthly rainfall, using long sequential raw data for time series analysis. “All-India” monthly average precipitation data for the period 1871–2016 were taken to build the models and they were tested on different homogeneous regions of India to check their robustness. From the results, it is evident that both the trained models (RNN and LSTM) performed well for different homogeneous regions of India based on the raw data. The study shows that a deep learning network can be applied successfully for time series analysis in the field of hydrology and allied fields to mitigate the risks of climatic extremes.     

Spatio-temporal evaluation of global gridded precipitation datasets across Iran

ABSTRACT

In this work, the accuracy of four gridded precipitation datasets – Climatic Research Unit (CRU), Global Precipitation Climatology Centre (GPCC), PERSIANN-Climate Data Record (PCDR) and University of Delaware (UDEL) – is evaluated across Iran to find an alternative source of precipitation data. Monthly, seasonal and annual precipitation data from 85 synoptic stations for the period 1984–2013 were used as the basis for the evaluations. Our results indicate that all datasets underestimate and overestimate precipitation in stations with annual precipitation greater than 600 and less than 100 mm, respectively. However, all datasets correctly recognize regimes of precipitation, but with a bias in amount of precipitation. Our spatio-temporal assessments show that GPCC is the most suitable dataset to be used over Iran. Both UDEL and CRU can be considered as the second and third most suitable datasets, while PCDR showed the weakest performance among the studied datasets.     

Spatio-temporal heterogeneity and changes in extreme precipitation over eastern Himalayan catchments India

This study draws attention on the extreme precipitation changes over the eastern Himalayan region of the Teesta river catchment. To explore the precipitation variability and heterogeneity, observed (1979–2005) and statistically downscaled (2006–2100) Coupled Model Intercomparison Project Phase Five earth system model global circulation model daily precipitation datasets are used. The trend analysis is performed to analyze the long-term changes in precipitation scenarios utilizing non-parametric Mann–Kendall (MK) test, Kendall Tau test, and Sen’s slope estimation. A quantile regression (QR) method has been applied to assess the lower and upper tails changes in precipitation scenarios. Precipitation extreme indices were generated to quantify the extremity of precipitation in observed and projected time domains. To portrait the spatial heterogeneity, the standard deviation and skewness are computed for precipitation extreme indices. The results show that the overall precipitation amount will be increased in the future over the Himalayan region. The monthly time series trend analysis based results reflect an interannual variability in precipitation. The QR analysis results showed significant increments in precipitation amount in the upper and lower quantiles. The extreme precipitation events are increased during October to June months; whereas, it decreases from July to September months. The representative concentration pathway (RCP) 8.5 based experiments showed extreme changes in precipitation compared to RCP2.6 and RCP4.5. The precipitation extreme indices results reveal that the intensity of precipitation events will be enhanced in future time. The spatial standard deviation and skewness based observations showed a significant variability in precipitation over the selected Himalayan catchment.     

Trends in the chemistry of precipitation and surface water in a national network of small watersheds

Trends in precipitation and surface water chemistry at a network of 15 small watersheds (< 10 km²) in the USA were evaluated using a statistical test for monotonic trends (the seasonal Kendall test) and a graphical smoothing technique for the visual identification of trends. Composite precipitation samples were collected weekly and surface water samples were collected at least monthly. Concentrations were adjusted before trend analysis, by volume for precipitation samples and by flow for surface water samples. A relation between precipitation and surface water trends was not evident either for individual inorganic solutes or for solute combinations, such as ionic strength, at most sites. The only exception was chloride, for which there was a similar trend at 60% of the sites. The smoothing technique indicated that short-term patterns in precipitation chemistry were not reflected in surface waters. The magnitude of the short-term variations in surface water concentration was generally larger than the overall long-term trend, possibly because flow adjustment did not adequately correct for climatic variability. Detecting the relation between precipitation and surface water chemistry trends may be improved by using a more powerful sampling strategy and by developing better methods of concentration adjustment to remove the effects of natural variation in surface waters.     

Hydrological modelling using proxies for gauged precipitation and temperature

Precipitation and temperature time series suffer from many problems, such as short time, inadequate spatial coverage, missing data, and biases from various causes, which are particularly critical in remote areas such as Northern Canada. The development of alternative datasets for using as proxies for inadequate/missing weather data represents a key research area. In this paper, the performance of 6 alternative datasets is evaluated for hydrological modelling over 12 watersheds located across Canada and the contiguous United States. The datasets can be classified into 3 distinct categories: (a) interpolated gridded data, (b) reanalysis data, and (c) climate model outputs. Hydrological simulations were carried out using a lumped conceptual hydrological model calibrated using standard weather data and compared against results using a calibration specific to each alternative dataset. Prior to the hydrological simulations, the alternative datasets were all evaluated with respect to their ability to reproduce gridded daily precipitation and temperature characteristics over North America. The results show that both the reanalysis data and climate model data adequately represent the spatial pattern of daily precipitation and temperature over North America. The North American Regional Reanalysis (NARR) dataset consistently shows the best performance. With respect to hydrological modelling, the observed discharges are accurately represented by both the gridded and NARR datasets, and more so for the NARR data. The National Centers for Environmental Prediction dataset consistently performs worst as it is unable to even capture the seasonal pattern of observed streamflow for 3 out of the 12 watersheds. These results indicate that the NARR dataset could be used as a proxy for gauged precipitation and temperature for hydrological modelling over watersheds where observational datasets are deficient. The results also illustrate the ability of climate model data to be used for performing hydrological modelling when driven by reanalysis data at their boundaries, and especially so for high‐resolution models.     

An evaluation of approaches for modelling hydrological processes in high‐elevation,glacierized Andean watersheds

We use two hydrological models of varying complexity to study the Juncal River Basin in the Central Andes of Chile with the aim to understand the degree of conceptualization and the spatial structure that are needed to model present and future streamflows. We use a conceptual semi‐distributed model based on elevation bands [Water Evaluation and Planning (WEAP)], frequently used for water management, and a physically oriented, fully distributed model [Topographic Kinematic Wave Approximation and Integration ETH Zurich (TOPKAPI‐ETH)] developed for research purposes mainly. We evaluate the ability of the two models to reproduce the key hydrological processes in the basin with emphasis on snow accumulation and melt, streamflow and the relationships between internal processes. Both models are capable of reproducing observed runoff and the evolution of Moderate‐resolution Imaging Spectroradiometer snow cover adequately. In spite of WEAP's simple and conceptual approach for modelling snowmelt and its lack of glacier representation and snow gravitational redistribution as well as a proper routing algorithm, this model can reproduce historical data with a similar goodness of fit as the more complex TOPKAPI‐ETH. We show that the performance of both models can be improved by using measured precipitation gradients of higher temporal resolution. In contrast to the good performance of the conceptual model for the present climate, however, we demonstrate that the simplifications in WEAP lead to error compensation, which results in different predictions in simulated melt and runoff for a potentially warmer future climate. TOPKAPI‐ETH, using a more physical representation of processes, depends less on calibration and thus is less subject to a compensation of errors through different model components. Our results show that data obtained locally in ad hoc short‐term field campaigns are needed to complement data extrapolated from long‐term records for simulating changes in the water cycle of high‐elevation catchments but that these data can only be efficiently used by a model applying a spatially distributed physical representation of hydrological processes. Copyright © 2013 John Wiley & Sons, Ltd.     

Hydrological modelling uncertainty analysis for different flow quantiles: a case study in two hydro-geographically different watersheds

This study focuses on analysis of hydrological model parameter uncertainty at varying sub-basin spatial scales. It was found that the variation in sub-basin spatial scale had little influence on the entire flow simulations. However, the different sub-basin spatial scales had a significant impact on the reproduction of the flow quantiles. The coarser sub-basin spatial scale provided a better coverage of most prediction uncertainty in observations. However, the finer sub-basin spatial scale produced the best single simulation output closer to the observations. In general, the optimal sub-basin spatial scales (ratio to the entire watershed size) in the two test watersheds were found to be in the ranges 14–19% and 2–4% for good simulation of high and low flows, respectively. It is therefore worthwhile to put more effort into reproducing different flow quantiles by investigating an appropriate sub-basin spatial scale.     

A Lagrangian approach to modelling stable isotopes in precipitation over mountainous terrain

A Lagrangian (Rayleigh) distillation model is used to track the evolution of stable isotopes in precipitation over mountainous terrain from the Pacific Coast of Canada to two alpine field sites in the Canadian Rocky Mountains. Precipitation δ¹⁸O at Vancouver constrains the model and air–mass back trajectories provide the water vapour pathway for 10 winter storm events. Isotopic values along storm pathways are modelled with a classical Rayleigh model that prescribes a linear decrease in temperature and pressure from initial to final conditions, and two models that account directly for orographic precipitation processes by: (i) applying an orographic rainfall model and (ii) using North American Regional Reanalysis data to calculate the change in vapour content along storm pathways. All models are significant predictors of snowpack δ¹⁸O, but the orographic model provides the best fit to precipitation‐weighted δ¹⁸O for each storm. The improvement in modelled δ¹⁸O by accounting for terrain along storm trajectories illustrates the need to account for orographically driven moisture loss when modelling vapour transport to ice core sites with mountainous upwind terrain. This finding is also applicable to isotopic studies of paleoaltimetry and source areas of groundwater recharge. Copyright © 2011 John Wiley & Sons, Ltd.     

Appropriate temporal resolution of precipitation data for discharge modelling in pre-alpine catchments

Precipitation time series with high temporal resolution are desired for hydrological modelling and flood studies. Yet the choice of an appropriate resolution is not straightforward because the use of too high a temporal resolution increases the data requirements, computational costs and, presumably, associated uncertainty, while performance improvement may be indiscernible. In this study, the effect of averaging hourly precipitation on model performance and associated uncertainty is investigated using two data sources: station network precipitation (SNP) and radar-based precipitation (RBP). From these datasets, time series of different temporal resolutions were generated, and runoff was simulated for 13 pre-alpine catchments with a bucket-type model. Our results revealed that different temporal resolutions were required for an acceptable model performance depending on the catchment size and data source. These were 1–12 h for small (16–59 km²), 3-21 h for medium (60–200 km²), and 24 h for large (200–939 km²) catchments.     

Spatio-temporal analysis of daily,seasonal and annual precipitation concentration in Jharkhand state,India

Nowadays, climate change and global warming have led to changes in the distribution of precipitation, which affect on the availability of water resources. Therefore, investigating the temporal and spatial variations of precipitation in the previous period is highly important in the future planning for flood control and local management of water resources. Considering the importance of this issue, in the present study, the precipitation concentration indices have been used for analysing precipitation changes at daily, seasonal, and annual time scales in the period of 1971 to 2011 over the Jharkhand state, India. Also, Modified Mann–Kendall test has used to study the trend of precipitation concentration indices in annual and seasonal time scales. The result shows a highly irregular and non-uniform distribution in the annual scale. For the seasonal scale an irregular and non-uniform distribution has been also observed, although the summer had a better situation than other seasons. For daily scale, none of the stations had a regular concentration and in the northeast and southern parts of the study area, there have been more irregularities. Furthermore, the results of investigating annual precipitation trend showed a combination of increasing and decreasing trend over the study area. The results of this study can be applied to manage water supplies, drainage projects, construct collection structures of urban flood, develop plans to prevent soil erosion, and designing appropriate plans to cope with drought conditions.     

A remote sensing contribution to hydrologic modelling in arid and inaccessible watersheds,Pishin Lora basin,Pakistan

The lack of adequate field measurements often hampers the construction and calibration of rainfall‐runoff models over many of the world's watersheds. We adopted methodologies that rely heavily on readily available remote sensing datasets as viable alternatives for assessing, managing, and modelling of such remote and inadequately gauged regions. The Soil and Water Assessment Tool was selected for continuous (1998–2005) rainfall‐runoff modelling of one such area, the northeast part of the Pishin Lora basin (NEPL). Input to the model included satellite‐based Tropical Rainfall Measuring Mission precipitation data, and modelled runoff was calibrated against satellite‐based observations, the latter included: (i) monthly estimates of the water volumes impounded by the Khushdil Khan (latitude 30°40′N, longitude 67°40′E), and the Kara Lora (latitude 30°34′N, longitude 66°52′E) reservoirs, and (ii) inferred wet versus dry conditions in streams across the NEPL. Calibrations were also conducted against observed flow reported from the Burj Aziz Khan station at the NEPL outlet (latitude 30°20′N; longitude 66°35′E). Model simulations indicate that (i) average annual precipitation (1998–2005), runoff and recharge in the NEPL are 1300 × 10⁶ m³, 148 × 10⁶ m³, and 361 × 10⁶ m³, respectively; (ii) within the NEPL watershed, precipitation and runoff are high for the northeast (precipitation: 194 mm/year; runoff: 38 × 10⁶ m³/year) and northwest (134 mm/year; 26 × 10⁶ m³/year) basins compared to the southern basin (124 mm/year; 8 × 10⁶ m³/year); and (3) construction of delay action dams in the northeast and northwest basins could increase recharge from 361 × 10⁶ m³/year up to 432 × 10⁶ m³/year and achieve sustainable extraction. The adopted methodologies are not a substitute for traditional approaches, but they could provide first‐order estimates for rainfall, runoff, and recharge in the arid and semi‐arid parts of the world that are inaccessible and/or lack adequate coverage with field data. Copyright © 2011 John Wiley & Sons, Ltd.     

Nitrate dynamics in a rural headwater catchment: measurements and modelling

This study was designed to improve our understanding of, and mechanistically simulate, nitrate (NO₃) dynamics in a steep 9.8 ha rural headwater catchment, including its production in soil and delivery to a stream via surface and subsurface processes. A two‐dimensional modelling approach was evaluated for (1) integrating these processes at a hillslope scale annually and within storms, (2) estimating denitrification, and (3) running virtual experiments to generate insights and hypotheses about using trees in streamside management zones (SMZs) to mitigate NO₃ delivery to streams. Total flow was mathematically separated into quick‐ and slow‐flow components; the latter was routed through the HYDRUS software with a nitrogen module designed for constructed wetlands. Flow was monitored for two years. High surface‐soil NO₃ concentrations started to be delivered to the stream via preferential subsurface flow within two days of the storm commencing. Groundwater NO₃‐N concentrations decreased from 1.0 to less than 0.1 mg l^?1 from up‐slope to down‐slope water tables, respectively, which was attributed to denitrification. Measurements were consistent with the flushing of NO₃ mainly laterally from surface soil during and following each storm. The model accurately accounted for NO₃ turnover, leading to the hypotheses that denitrification was a minor flux (<3 kg N ha^?1) compared to uptake (98^?127 kg N ha^?1), and that SMZ trees would reduce denitrification if they lowered the water table. This research provides an example of the measurement and modelling of NO₃ dynamics at a small‐catchment scale with high spatial and temporal resolution. Copyright © 2013 John Wiley & Sons, Ltd.     

A study for improving precipitation occurrences modelling with a Markov chain

Daily precipitation occurrences and their monthly wet-days' sums of precipitation-measuring stations in Greece are modelled with a Markov chain. The order of the chain is taken to be seasonally varying in accordance with the precipitation station's meteorological conditions and geographical location. The modelling efficiency of the Markov chain is significantly improved when it is conjunctively used with a second-order autoregressive stochastic model fitted on the monthly wet-days' sums.     

Spatio-temporal modelling of hydro-meteorological derived risk using a Bayesian approach: a case study in Venezuela

Extreme environmental events have considerable impacts on society. Preparation to mitigate or forecast accurately these events is a growing concern for governments. In this regard, policy and decision makers require accurate tools for risk estimation in order to take informed decisions. This work proposes a Bayesian framework for a unified treatment and statistical modeling of the main components of risk: hazard, vulnerability and exposure. Risk is defined as the expected economic loss or population affected as a consequence of a hazard event. The vulnerability is interpreted as the loss experienced by an exposed population due to hazard events. The framework combines data of different spatial and temporal supports. It produces a sequence of temporal risk maps for the domain of interest including a measure of uncertainty for the hazard and vulnerability. In particular, the considered hazard (rainfall) is interpolated from point-based measured rainfall data using a hierarchical spatio-temporal Kriging model, whose parameters are estimated using the Bayesian paradigm. Vulnerability is modeled using zero-inflated distributions with parameters dependent on climatic variables at local and large scales. Exposure is defined as the total population settled in the spatial domain and is interpolated using census data. The proposed methodology was applied to the Vargas state of Venezuela to map the spatio-temporal risk for the period 1970–2006. The framework highlights both high and low risk areas given extreme rainfall events.