Representing and communicating deep uncertainty in climate-change assessments

IPCC reports provide a synthesis of the state of the science in order to inform the international policy process. This task is made difficult by the presence of deep uncertainty in the climate problem that results from long time scales and complexity. This paper focuses on how deep uncertainty can be effectively communicated. We argue that existing schemes do an inadequate job of communicating deep uncertainty and propose a simple approach that distinguishes between various levels of subjective understanding in a systematic manner. We illustrate our approach with two examples. To cite this article: M. Kandlikar et al., C. R. Geoscience 337 (2005).     

Effects of temporal scales and space mismatches on the TRMM 3B42 v7 precipitation product in a remote mountainous area

The accurate measurement of precipitation is essential to understanding regional hydrological processes and hydrological cycling. Quantification of precipitation over remote regions such as the Tibetan Plateau is highly unreliable because of the scarcity of rain gauges. The objective of this study is to evaluate the performance of the satellite precipitation product of tropical rainfall measuring mission (TRMM) 3B42 v7 at daily, weekly, monthly, and seasonal scales. Comparison between TRMM grid precipitation and point‐based rain gauge precipitation was conducted using nearest neighbour and bilinear weighted interpolation methods. The results showed that the TRMM product could not capture daily precipitation well due to some rainfall events being missed at short time scales but provided reasonably good precipitation data at weekly, monthly, and seasonal scales. TRMM tended to underestimate the precipitation of small rainfall events (less than 1 mm/day), while it overestimated the precipitation of large rainfall events (greater than 20 mm/day). Consequently, TRMM showed better performance in the summer monsoon season than in the winter season. Through comparison, it was also found that the bilinear weighted interpolation method performs better than the nearest neighbour method in TRMM precipitation extraction.     

Efficient estimation of flood forecast prediction intervals via single‐ and multi‐objective versions of the LUBE method

Prediction intervals (PIs) are commonly used to quantify the accuracy and precision of a forecast. However, traditional ways to construct PIs typically require strong assumptions about data distribution and involve a large computational burden. Here, we improve upon the recent proposed Lower Upper Bound Estimation method and extend it to a multi‐objective framework. The proposed methods are demonstrated using a real‐world flood forecasting case study for the upper Yangtze River Watershed. Results indicate that the proposed methods are able to efficiently construct appropriate PIs, while outperforming other methods including the widely used Generalized Likelihood Uncertainty Estimation approach. Copyright © 2016 John Wiley & Sons, Ltd.     

A GLUE‐based uncertainty assessment framework for tritium‐inferred transit time estimations under baseflow conditions

The last decade has seen major technical and scientific improvements in the study of water transfer time through catchments. Nevertheless, it has been argued that most of these developments used conservative tracers that may disregard the oldest component of water transfer, which often has transit times greater than 5 years. Indeed, although the analytical reproducibility of tracers limits the detection of the older flow components associated with the most dampened seasonal fluctuations, this is very rarely taken into account in modelling applications. Tritium is the only environmental tracer at hand to investigate transfer times in the 5‐ to 50‐year range in surface waters, as dissolved gases are not suitable due to the degassing process. Water dating with tritium has often been difficult because of the complex history of its atmospheric concentration, but its current stabilization together with recent analytical improvements open promising perspectives. In this context, the innovative contribution of this study lies in the development of a generalized likelihood uncertainty estimation‐based approach for analysing the uncertainties associated with the modelling of transit time due to both parameter identification and tracer analytical precision issues. A coupled resampling procedure allows assessment of the statistical significance of the transfer time differences found in diverse waters. This approach was developed for tritium and the exponential‐piston model but can be implemented for virtually any tracer and model. Stream baseflow, spring and shallow aquifer waters from the Vallcebre research catchments, analysed for tritium in different years with different analytical precisions, were investigated by using this approach and taking into account other sources of uncertainty. The results showed three groups of waters of different mean transit times, with all the stream baseflow and spring waters older than the 5‐year threshold needing tritium. Low sensitivity of the results to the model structure was also demonstrated. Dual solutions were found for the waters sampled in 2013, but these results may be disambiguated when additional analyses will be made in a few years. Copyright © 2016 John Wiley & Sons, Ltd.     

Inference and uncertainty of snow depth spatial distribution at the kilometre scale in the Colorado Rocky Mountains: the effects of sample size,random sampling,predictor quality,and validation procedures

Historically, observing snow depth over large areas has been difficult. When snow depth observations are sparse, regression models can be used to infer the snow depth over a given area. Data sparsity has also left many important questions about such inference unexamined. Improved inference, or estimation, of snow depth and its spatial distribution from a given set of observations can benefit a wide range of applications from water resource management, to ecological studies, to validation of satellite estimates of snow pack. The development of Light Detection and Ranging (LiDAR) technology has provided non‐sparse snow depth measurements, which we use in this study, to address fundamental questions about snow depth inference using both sparse and non‐sparse observations. For example, when are more data needed and when are data redundant? Results apply to both traditional and manual snow depth measurements and to LiDAR observations. Through sampling experiments on high‐resolution LiDAR snow depth observations at six separate 1.17‐km² sites in the Colorado Rocky Mountains, we provide novel perspectives on a variety of issues affecting the regression estimation of snow depth from sparse observations. We measure the effects of observation count, random selection of observations, quality of predictor variables, and cross‐validation procedures using three skill metrics: percent error in total snow volume, root mean squared error (RMSE), and R². Extremes of predictor quality are used to understand the range of its effect; how do predictors downloaded from internet perform against more accurate predictors measured by LiDAR? Whereas cross validation remains the only option for validating inference from sparse observations, in our experiments, the full set of LiDAR‐measured snow depths can be considered the ‘true’ spatial distribution and used to understand cross‐validation bias at the spatial scale of inference. We model at the 30‐m resolution of readily available predictors, which is a popular spatial resolution in the literature. Three regression models are also compared, and we briefly examine how sampling design affects model skill. Results quantify the primary dependence of each skill metric on observation count that ranges over three orders of magnitude, doubling at each step from 25 up to 3200. Whereas uncertainty (resulting from random selection of observations) in percent error of true total snow volume is typically well constrained by 100–200 observations, there is considerable uncertainty in the inferred spatial distribution (R²) even at medium observation counts (200–800). We show that percent error in total snow volume is not sensitive to predictor quality, although RMSE and R² (measures of spatial distribution) often depend critically on it. Inaccuracies of downloaded predictors (most often the vegetation predictors) can easily require a quadrupling of observation count to match RMSE and R² scores obtained by LiDAR‐measured predictors. Under cross validation, the RMSE and R² skill measures are consistently biased towards poorer results than their true validations. This is primarily a result of greater variance at the spatial scales of point observations used for cross validation than at the 30‐m resolution of the model. The magnitude of this bias depends on individual site characteristics, observation count (for our experimental design), and sampling design. Sampling designs that maximize independent information maximize cross‐validation bias but also maximize true R². The bagging tree model is found to generally outperform the other regression models in the study on several criteria. Finally, we discuss and recommend use of LiDAR in conjunction with regression modelling to advance understanding of snow depth spatial distribution at spatial scales of thousands of square kilometres. Copyright © 2012 John Wiley & Sons, Ltd.     

Measurement of stream cross section using ground penetration radar with Hilbert–Huang transform

This study presents a new method to measure stream cross section without having contact with water. Compared with conventional measurement methods which apply instruments such as sounding weight, ground penetration radar (GPR), used in this study, is a non‐contact measurement method. This non‐contact measurement method can reduce the risk to hydrologists when they are conducting measurements, particularly in high flow period. However, the original signals obtained by using GPR are very complex, different from studies in the past where the measured data were mostly interpreted by experts with special skill or knowledge of GPR so that the results obtained were less objective. This study employs Hilbert–Huang transform (HHT) to process GPR signals which are difficult to interpret by hydrologists. HHT is a newly developed signal processing method that can not only process the nonlinear and non‐stationary complex signals, but also maintain the physical significance of the signal itself. Using GPR with HHT, this study establishes a non‐contact stream cross‐section measurement method with the ability to measure stream cross‐sectional areas precisely and quickly. Also, in comparison with the conventional method, no significant difference in results is found to exist between the two methods, but the new method can considerably reduce risk, measurement time, and manpower. It is proven that the non‐contact method combining GPR with HHT is applicable to quickly and accurately measure stream cross section. Copyright © 2013 John Wiley & Sons, Ltd.     

Balancing trade‐off issues in land use change and the impact on streamflow and salinity management

The south‐west region of the Goulburn–Broken catchment in the south‐eastern Murray–Darling Basin in Australia faces a range of natural resource challenges. A balanced strategy is required to achieve the contrasting objectives of remediation of land salinization and reducing salt export, while maintaining water supply security to satisfy human consumption and support ecosystems. This study linked the Catchment Analysis Tool (CAT), comprising a suite of farming system models, to the catchment‐scale CATNode hydrological model to investigate the effects of land use change and climate variation on catchment streamflow and salt export. The modelling explored and contrasted the impacts of a series of different revegetation and climate scenarios. The results indicated that targeted revegetation to only satisfy biodiversity outcomes within a catchment is unlikely to have much greater impact on streamflow and salt load in comparison with simple random plantings. Additionally, the results also indicated that revegetation to achieve salt export reduction can effectively reduce salt export while having a disproportionately smaller affect on streamflows. Furthermore, streamflow declines can be minimized by targeting revegetation activities without significantly altering salt export. The study also found that climate change scenarios will have an equal if not more significant impact on these issues over the next 70 years. Uncertainty in CATNode streamflow predictions was investigated because of the effect of parameter uncertainty. Copyright © 2012 John Wiley & Sons, Ltd.     

Effect of rainfall uncertainty on the performance of physically based rainfall–runoff models

This paper analyses the effect of rain data uncertainty on the performance of two hydrological models with different spatial structures: a semidistributed and a fully distributed model. The study is performed on a small catchment of 19.6 km² located in the north‐west of Spain, where the arrival of low pressure fronts from the Atlantic Ocean causes highly variable rainfall events. The rainfall fields in this catchment during a series of storm events are estimated using rainfall point measurements. The uncertainty of the estimated fields is quantified using a conditional simulation technique. Discharge and rain data, including the uncertainty of the estimated rainfall fields, are then used to calibrate and validate both hydrological models following the generalized likelihood uncertainty estimation (GLUE) methodology. In the storm events analysed, the two models show similar performance. In all cases, results show that the calibrated distribution of the input parameters narrows when the rain uncertainty is included in the analysis. Otherwise, when rain uncertainty is not considered, the calibration of the input parameters must account for all uncertainty in the rainfall–runoff transformation process. Also, in both models, the uncertainty of the predicted discharges increase in similar magnitude when the uncertainty of rainfall input increase.     

Spatio‐temporal variability of the isotopic input signal in a partly forested catchment: Implications for hydrograph separation

The isotopic composition of precipitation (D and ¹⁸O) has been widely used as an input signal in water tracer studies. Whereas much recent effort has been put into developing methodologies to improve our understanding and modelling of hydrological processes (e.g., transit‐time distributions or young water fractions), less attention has been paid to the spatio‐temporal variability of the isotopic composition of precipitation, used as input signal in these studies. Here, we investigated the uncertainty in isotope‐based hydrograph separation due to the spatio‐temporal variability of the isotopic composition of precipitation. The study was carried out in a Mediterranean headwater catchment (0.56 km²). Rainfall and throughfall samples were collected at three locations across this relatively small catchment, and stream water samples were collected at the outlet. Results showed that throughout an event, the spatial variability of the input signal had a higher impact on hydrograph separation results than its temporal variability. However, differences in isotope‐based hydrograph separation determined preevent water due to the spatio‐temporal variability were different between events and ranged between 1 and 14%. Based on catchment‐scale isoscapes, the most representative sampling location could also be identified. This study confirms that even in small headwater catchments, spatio‐temporal variability can be significant. Therefore, it is important to characterize this variability and identify the best sampling strategy to reduce the uncertainty in our understanding of catchment hydrological processes.     

Assessing uncertainty in best management practice effectiveness under future climate scenarios

Uncertainty of best management practice (BMP) performance in future climates is an important consideration for water resources managers. The objective of this study was to quantify the level of uncertainty in performance of seven agricultural BMPs due to climate change in reducing sediment, total nitrogen, and total phosphorus loads. The Soil and Water Assessment Tool coupled with mid‐21^st century climate data from the Community Climate System Model were used to develop climate change scenarios for the Tuttle Creek Lake Watershed of Kansas and Nebraska. Uncertainty level of each BMP was determined using Latin Hypercube Sampling, a constrained Monte Carlo sampling technique. Samples were taken from distributions of several variables (monthly precipitation, temperature, CO₂, and BMP implementation parameters). Cumulative distribution functions were constructed for each BMP, pollutant, and climate scenario combination. Results demonstrated that BMP performance uncertainty is amplified in the extreme climate scenario. Among BMPs, native grass replacement generally had higher uncertainty level but also had the greatest reductions. This study highlights the importance of incorporating uncertainty analysis into mitigation strategies aiming to reduce negative impacts of climate change on water resources. Copyright © 2013 John Wiley & Sons, Ltd.