Popular Features. New Releases. The aim is to understand how all of these sectors will interact with one another and with climate change and then ultimately, try to understand what effect these interactions will have on societies and individual people. It is these cross-sectoral interactions between multiple drivers that are really significant.
As a result, single sector models will generally produce results that inadequately represent these complexities, leading to serious over or under estimations of the effects of climate change. When attempting to model future climate change impact scenarios, it is therefore essential to consider multiple drivers and the interactions between these drivers. Speaking about his work, Professor Rounsevell, who is the Kinloch Michie Chair of Rural Economy and Environmental Sustainability, talks of the importance of impact modelling. Professor Rounsevell and his colleagues help to inform policy makers about the future consequences of climate change, and those policy makers in turn help to inform the socio-economic scenarios that are built into the impact models.
Understanding how changes will impact on the economy and the social welfare of individuals and society is really important in terms of informing the response strategies to the effects of climate change. LUC4C - A project researching into the impacts of land use and land cover change on the climate. The model was run using the numerical procedure of the Runge—Kutta 4th order discretization at daily time steps. The calibration period varies from a maximum of ten years to a minimum of five years for some river sites with exception of the HMR river site where only four years of water temperature observations were available.
Since the model calibration was conducted by minimizing water temperature differences against the observations, the simulated water temperature is not independent of the observed water temperature during the calibration time period. Once the model was calibrated and validated, we integrated the model from — to estimate long-term daily time series for each river site and to evaluate changes in the simulated water temperatures.
Overall, the main advantage of using ANUSPLIN is data homogeneity and lack of gaps, which are commonly reported as major issues with station-based data. We therefore extracted mean air temperature data from gridded observations for all 17 sites rather than using available limited station-based data. Simulated mean summer water temperatures were first quantified for all river sites for — Monotonic trends in simulated water temperatures were then computed using the non-parametric Mann-Kendall Test MKT 36 , We summarised results using the temporal mean and interannual variability standard deviation of simulations at each site along with the multi-site mean and inter-site standard deviation spread of summer water temperature.
The standard deviation at each site was used to compute signal to noise ratios SNRs , i.
The quantitative assessment of the impact of climate change on water availability and water resources management requires knowledge of. The quantitative assessment of the impact of climate change on water availability and water resources management requires knowledge of climate.
The t-test statistic was used to compute the significance of the composite differences. Here it was assumed that the underlying process behaved as an independent and identically distributed random variable. For the PDO association with water temperatures, we estimated simulated water temperature differences between the positive — and negative — phases of the PDO The t-test statistic was used for differences in conditions between positive and negative PDO phases.
Two year time periods, namely — referred hereafter as the s and — referred hereafter as the s revealed climate change impacts on simulated water temperatures from the hindcast simulations. We used the frequency of exceeding different critical temperatures related to Pacific salmon physiology 19 , The STMP regulates water temperatures through additional releases of flows from the Nechako reservoir via the Skins Lake Spillway into the Nechako River for the benefit of salmon migration Frequency distributions of water temperature for summer days utilized 50 equal bins with 0.
To better understand the contributions of air temperature and river discharge to water temperatures, we used a multivariate linear regression MLR analysis, following a similar approach by Islam et al.
The MLR was fitted for each site for mean summer and individual months independently using standardised and detrended monthly anomalies from — We standardized the time series to get zero mean and unit standard deviation. This was estimated by removing the mean and dividing the data by their standard deviation. The MLR explained variance R 2 and standardized partial regression coefficients were computed along with their significance using t-test statistics.
We only considered the lag-0 correlation between the driving and response variables in the MLR analysis considering that monthly time resolution should have encompassed any lags. Model results show that for all 17 study sites, NSE values range from 0. While the mean daily BIAS is small 0.
Higher RMSE scores are mainly due to higher day-to-day variability in the observed data, which remains challenging for the Air2Stream model to simulate accurately. This may be due to the gap-filling and spatial smoothing procedures applied to generate continuous and homogenous daily air temperature in the ANUSPLIN dataset that may have suppressed rapid air temperature fluctuations. The resulting Air2Stream simulations therefore show dampened daily variability in simulated water temperatures when compared to observations.
For the summer season on an interannual time scale, however, the model performs much better, particularly with respect to RMSE. During the calibration period summer RMSE values average 0. For the Air2Stream model validation, similar metrics cannot be computed for all sites due to limited water temperature observations. For example, the HMR observed water temperature data were available only for — and used to calibrate the Air2Stream model for this site; there were no data beyond this period for model validation.
For river sites with extended observed water temperature records e. NSE scores for the validation period for these sites range from 0. During validation, daily RMSE averages 1. Errors lie within the acceptable range of water temperature model reliability reported in other studies 21 , 22 , Observed and simulated water temperature climatologies during the calibration periods overlap each other for most of sites except FHE Supplementary Fig.
Furthermore, the Air2Stream model shows robust performance in simulating the observed interannual variability in summer water temperature for most river sites Supplementary Fig. At NFF and NVH where observed daily water temperature records exceed 30 years — , the model reproduces well simulated short-term summer water temperature trends Supplementary Fig. At FHG, the year mean of simulated summer water temperature The model, however, underestimates the FHG trend magnitude compared to observations 0.
Furthermore, the interannual variability of simulated summer water temperature is slightly lower than observed variability Supplementary Fig. The year comparison between simulated and observed water temperatures is only possible for FHG considering the limited availability of long-term observations for all remaining sites. Comparison of simulated water temperature with gridded air temperature shows summer water temperatures nearly 2. The summer trend of 1. The interannual variability of summer water temperature ranges between 0. The signal to noise ratio exceeds unity for the sites with significant warming trends during summer suggesting this secular pattern dominates the interannual variation.
Simulated averaged responses of the summer water temperature climatology and frequency distribution in the upper, middle and lower FRB show climate change impacts on river water temperature from the s to s Fig.
In the upper and middle FRB, the mean summer water temperatures warm by nearly 1. In addition, the timing of the peak summer water temperature changes consistently from the s to s in each section of the FRB. The averaged response of water temperatures in the upper and middle FRB shows a day shift in the timing of maximum summer water temperature from 2 August in the s to 14 August in the s, with a more modest 5-day delay in the lower FRB from 12 August in the s to 17 August in the s. The timing of peak summer water temperature at individual sites closely resembles their mean response in the upper, middle and lower FRB.
The temporal changes of maximum summer water temperature and corresponding day of occurrence further reveals the increasing frequency of warmer water temperature days and delay in the timing of river sites especially in the upper and middle Fraser Supplementary Fig. Climatology a,c,e and frequency distributions b,d,f of daily simulated summer water temperatures for the s and s.
Dashed lines represent individual responses at each site. The arrows in the left column indicate days of the summer with maximum water temperature.
See Supplementary Fig. S6 for frequency distributions at each site. The frequency distributions gradually shift toward higher summer water temperature in the recent past i. Frequency distributions at each site show a similar pattern of a climate shift Supplementary Fig. The changes in extreme water temperature are quite variable along the Nechako River but more modest at other FRB sites. For SFJ, days exceeding T c20 increase markedly from 8 days in the s to 71 days in the s reflecting an amplifying frequency of extreme water temperatures with climate change.
In contrast, at the highly regulated NVH, the largest count for days exceeding T c18 is in the s at days of which 15 days surpassed T c The response to ENSO is more prominent during August coinciding with the peak up-river salmon migration period Overall, most of the river sites within the FRB show significant correlation between far field teleconnections and simulated river water temperatures. Triangles represent the magnitude of differences in summer water temperature during the positive — and negative — phases of the PDO.
The MLR analysis allows quantification of the contribution of mean air temperature and discharge to the simulated summer water temperature in the FRB. The variables used in the MLR analysis Eq. Multivariate linear regression MLR analysis of simulated summer water temperatures. Bars show partial regression coefficients b 1 and b 2 associated with air temperature and discharge, respectively.
All time series are detrended prior to MLR analysis. Mean air temperature remains a key driver of changes in simulated summer water temperatures for all sites except NFF and SGN where discharge imposes a dominant control Fig. Comparison of regression coefficients b 1 and b 2 for individual months shows, for most sites, air temperature contributions to water temperature increase after mid-summer whereas discharge contributions decrease in late summer due to flow recession Supplementary Fig. Overall the averaged response of partial regression coefficient b 1 during summer is 0.
The long-term effects projected for future climate in the basin were mainly related to a substantial decrease in rainfall, and an increase in temperature and wind speed. The temperature was generally predicted to increase more significantly during the summer, late spring, and winter, while a decrease was expected in March. All of the terms of the energy balance depend on the land surface temperature LST , which allows the energy balance equation to be solved by finding the thermodynamic equilibrium temperature which closes the equation using the Newton-Raphson method: 7 where LST n is the actual value, LST n- 1 is the value at the previous iteration, f t LST n- 1 is the energy balance function, and f t ' LST n- 1 is its derivative. The validation involved a graphical and statistical analysis. Int J Climatol — Nat Hazard Earth Sys 8: — Advertisement Hide.
This study presents new insights on the evolution of daily water temperatures for 17 sites across the FRB using Air2Stream, a state-of-the-art, physically-based water temperature model. The simulations quantified changes in summer water temperature magnitudes, timing and extremes during the late 20 th and early 21 st centuries. Since the daily NSE scores ranged between 0. Indeed, all model performance metrics spanned the range of model reliability reported in the literature 5 , 21 , 22 , For example, Foreman et al. Simulated water temperatures warmed substantially across the FRB between — in response mainly to rising air temperatures.
Robust responses of water temperatures to changes in air temperatures e. The low gradient of elevation along the Stuart River slows the delivery of water towards the basin outlet and hence air and water temperatures remain tightly coupled. In contrast, the much steeper gradient along the Stellako and Nautley rivers decouples air and water temperatures thereby suppressing the impacts of rising air temperatures in the Air2Stream simulations at SGN and NFF.
Such association of river water temperatures changes with basin characteristics is explored in many recent studies 46 , 47 , 48 , Overall, heightened air and river water temperature trends observed generally across parts of the Nechako watershed are well supported 50 , 51 , 52 and consistent with enhanced warming effects projected for northern latitudes The physical mechanism s however, may well be different for these sites depending on regional characteristics such as the presence of large lakes, glaciers, and gradients of elevation along waterways.
For example, while discharge contributes insufficiently to simulated water temperatures at HHY, modest increases in air temperature yield lower water temperature trends relative to the neighbouring QQL site. The two branches NTR and STC of the Thompson River, a tributary in the southeastern section of the FRB, also exhibits dampened warming trends for river water temperature relative to other waterways, likely due to the contribution of glacier-melt water from the Monashee, Cariboo and Rocky Mountains that regulates downstream river temperatures in this system.
Consistent with Foreman et al. Simulation results suggest that the discharge at these two sites strongly influences water temperatures when compared to air temperature. Improving understanding of these far-field relationships with changes in water temperature is crucial for the future conservation of different aquatic species, especially salmon populations. Changing aquatic conditions have implications for migrating salmon in the FRB as the mortality increases markedly when water temperatures surpass critical temperatures e.
The frequency of critical temperatures has amplified appreciably in recent decades e. Perhaps surprisingly, the northernmost river system SFJ experiences the highest frequency of critical water temperatures due to amplified atmospheric warming in the northern FRB. Further, with a nearly 1. Our simulated increases in mean summer water and critical temperatures are consistent with past studies focusing on Pacific salmon populations in the Columbia and Fraser river systems 56 , 57 , Studies 56 , 59 have indeed reported significant increases in summer freshwater temperatures since the s during spawning migrations in the Columbia and Fraser rivers, thus posing a major threat to the future sustainability of salmon populations in these systems 40 , An exception to increasing days exceeding critical temperatures occurs in the Nechako River in the northwest portion of the FRB.
Summer flows in the Nechako River are further regulated in response to forecasted air temperatures owing to a court injunction 42 , 50 , Model results suggest that the number of days when simulated water temperatures rise above T c20 has not changed during the pre- and post-STMP time periods despite rising air temperatures, thereby revealing the effectiveness of the STMP in meeting temperature requirements near NVH.
The warming trends in the Nechako River are likely not related to the impact of regulation, given that the varying summer water temperature response to climate change in its unregulated tributaries, i. Regulation decreases water temperatures especially during the STMP time window of 20 July to 20 August whereas the long-term trend abates due to regulation. This implies that overall water temperature increases in the Nechako River are mainly due to climate change rather than regulation. Available water temperature records remain limited for many FRB river sites posing challenges in calibrating and validating the Air2Stream model.
The available water temperature records also have data gaps and random, spurious data entries which, to some extent, are rectified by quality control and analysis. Another limitation of the Air2Stream model application is its assumption of a stationary response between air temperature, daily discharge and simulated water temperature. Furthermore, the model cannot adequately account for all the complexities of river thermal properties, especially the effects of different surface radiative and energy fluxes.