Basins 4.0 download

Nevertheless, we used CanRCM4-LE because of the limited availability of higher-resolution simulations covering the entire region. Furthermore, since our analyses are based on basin-averaged responses drainage areas: 57, km 2 to , km 2 ; SM Table S1 , the grid-scale uncertainty is assumed less critical.

As emphasized by Deser et al. We also compared the CanRCM4 projections with 0. We considered SWE, temperature, and precipitation changes under the policy-relevant thresholds of 1.

Given that GCMs have different levels of climate sensitivities and biases, and respond differently to the same radiative forcing scenario Schleussner et al.

Since CanRCM4 simulations are not available for the PI period, basin-scale comparisons of temperature, precipitation, and SWE changes at different GMT levels were carried out with respect to the more recent baseline period of — Following previous regression-based studies, which found high sensitivities of snow to seasonal precipitation and temperature e. We used the random forest RF ensemble machine learning method for the predictor-response variable importance VI analysis because of its ability to handle complex nonlinear problems including interactions between variables Breiman , which is an important consideration given that the snow response is a result of temperature and precipitation interactions.

RFs have also been found to be efficient on large databases Wang et al. We used the RF model to evaluate the basin-averaged annual maximum SWE SWE max response as a function of snow season air temperature herein referred to as temperature and precipitation, i.

The RF model inputs included four variables, i. The predictability of the response variable was assessed using the Nash-Sutcliffe coefficient of efficiency NSE. We used a categorical framework, consisting of a combination of temperature, precipitation, and SWE max quantiles at the above-normal, near-normal, and below-normal classes for evaluating the SD conditions.

The classification is analogous to the probabilistic climate forecasting in which shifts in forecasts e. The 33rd percentile threshold lies between the climatological mean threshold in Dierauer et al. We propose a generalized SD definition, applicable to primary mechanisms of SD occurrences in a range of hydro-climatic regimes, and in both historical and future climates SM Fig.

Given that SD could occur under both precipitation and temperature increases in a warmer climate, we expanded the SD classifications by Harpold et al. Accordingly, three main SD classes were identified:. Besides, these three SD classes decrease in SWE max could occur under other conditions, especially below-normal precipitation e. However, SD occurrences under these conditions were found to be rare in our evaluation of future response and were not considered.

Projected precipitation changes are highly heterogeneous, with larger increases in the northern region and southern interior, and progressively larger changes at higher GMTs see SM Figs.

S2 and S3. The signal-to-noise ratios of temperature and precipitation changes—as represented by the ensemble mean divided by standard deviation of the member model simulations—are greater than one in most areas, with the exception of precipitation changes at 1. This also suggests that the strengths of temperature and precipitation increase signals are robust across the domain, even at low GMT increases. In the case of SWE max , the responses are spatially heterogeneous, with a general pattern of increases in the northern region and declines in the southern and coastal regions Fig.

Besides this north to south gradient, there is also an interior region to coast gradient of decreasing SWE max. With the GMT increases from 1. Also relevant to SWE max changes are the basin-scale winter temperature states. In this study, the SWE max changes in the three northern basins Yukon, Peel, and Liard are small, because the basins will remain cold even under more intense warming Fig.

For the three interior basins Peace, Athabasca, and Saskatchewan , such balance between basin temperature and precipitation increases seem to mostly maintain the historical SWE max levels up to 2. However, at 3. Rapid loss was also projected by Fyfe et al.

Further, these results are in general agreement with previous studies that found relatively larger snowpack declines at lower elevations than higher elevations, and by extension larger declines at higher basin temperatures e.

However, the changes cannot be fully explained in terms of mean basin temperatures. For instance, although the basin temperatures are similar in Skeena, Fraser and Saskatchewan, the SWE max reduction in the former two is more amplified between 1. The results depict year means from member CanRCM4-LE, and the changes are considered relative to the mean of — baseline period. The controls vary greatly across the NWNA region, characterized by decreasing influence of precipitation and increasing influence of temperature from northern colder to southern warmer basins.

Overall, the temperature control on the southernmost basin Columbia is about three times higher than the northernmost basin Peel , with progressively higher sensitivities from north to south.

Furthermore, there are differences in the seasonal temperature controls on SWE max , e. Possible reasons for these differences include the temperature influence on the start of snow accumulation, snowmelt initiation, and distribution of seasonal snowfall. In general, the warmer the basin temperature state, the higher is the temperature control on SWE max , and the smaller is the influence of precipitation on SWE max change.

Also shown are the model fits in terms of NSE. Comparing the colder northern basins to warmer southern basins, the NSE values are higher with the predominant temperature controls in the south.

The NSE values are lower for the northern basin, especially Peel and Liard, thus, the seasonal temperature and precipitation changes do not sufficiently explain the SWE max change. The NSE-based performance also suggests a potential for developing predictive snowpack models using RFs, with further decomposition of driving variables e.

This was not explored in this study since we are only focused on VI of precipitation and temperature on SWE max response. Further, we analyzed the sensitivities of SWE max in relationship to October to March temperature and precipitation changes, and at 1.

The results depict contrasting responses to temperature and precipitation changes, characterized by increasing sensitivities of SWE max from north to south. Specifically, while SWE max generally declines with GMT increases, the magnitude of changes varies across the region, i.

As outlined earlier, the large SWE max reductions in the southern basins correspond to higher temperature sensitivities Fig. Further, it is evident that the precipitation increase is not able to offset the warming-induced decline in snow accumulation. In contrast, in the relatively colder northern basins, the higher precipitation increases mostly sustain the baseline SWE max levels even with higher temperature increases.

The importance of the basin-scale temperature on future SWE max response is illustrated by contrasting responses between b Peel and f Saskatchewan basins. While the precipitation increases for the two basins are in the similar ranges, there are more pronounced declines in SWE max in the warmer Saskatchewan basin even with smaller temperature increases. Nevertheless, the basin-scale precipitation change does have a role in SWE max change.

For instance, comparing the f Saskatchewan and g Skeena basins, where temperature changes and historical temperature ranges are similar, there is a smaller SWE max decline in Saskatchewan due to larger precipitation increase.

Overall, under the warming scenarios, the basins with higher temperature sensitivities will experience more substantive snow declines than the basins with higher precipitation sensitivities, with the magnitude of precipitation increase determining the magnitude of SWE max change. Furthermore, the results of this study align with the northern hemisphere snowpack variability assessment by Sospedra-Alfonso and Merryfield The threshold also aligns well with the start of the rapid snowpack decline in this study Fig.

Thus, with the twenty-first century warming, sensitivity of snowpack response to temperature in warmer regions increases and even dominates, likely due to temperature-induced reduction of the snowfall fraction, which is not compensated by precipitation increases.

Additionally, the regions currently with higher precipitation sensitivity would be expected to become more sensitive to temperature. Corresponding to the SD definition, the SWE max values were normalized by the 33rd percentiles of the baseline period — The results generally reveal a consistent pattern in terms of the direction of change, i.

An exception is the Peel basin, where small shifts toward higher SWE max values are projected under higher warming thresholds.

The difference can be mainly attributed to the predominant precipitation control for this basin Fig. In the case of Yukon and Liard basins, there is some evidence of the counteracting influence of precipitation increases, as depicted by almost identical quantile values at 1. However, there are distinct shifts in the quantile values from 2. Consequently, the levels of warming at which majority of years are under SD conditions vary across the region.

The levels of warming at which majority of years are under SD conditions are 1. The SWE max values are normalized by the 33rd percentile of the — baseline period shown by a vertical line.

Second vertical line depict the 67th percentile. Classifying SWE max quantiles into three categories—above-normal, near-normal, and below-normal—depict highly varied responses across the region Fig. For rest of the region, however, the SD conditions are greater than expected for all GMT changes including 1. The bars depict mean frequencies for the three SWE max classes and whiskers depict maximum-minimum range over member ensemble. Conversely, the lower temperature controls on SWE max in the colder northern basins will make the SD conditions less likely except for 3.

SD frequencies for below-normal SWE max conditions with respect to baseline period — under 1. Again, the severity of SD occurrences, e.

As expected, the reduction in SWE max , thus the severity of SD, are greater with decreased wetness, i. Hence, although SWE max change is primarily a temperature-controlled process Fig. Assessment of the impacts of climatic variability and anthropogenic stress on hydrologic resilience to warming shifts in Peninsular India.

Integrating risks of climate change into water management. Fowler, H. Using regional climate model data to simulate historical and future river flows in northwest England.

Change 80 , — Dobler, C. Assessment of climate change impacts on flood hazard potential in the Alpine Lech watershed. Uniyal, B. Assessing climate change impact on water balance components of a river basin using SWAT model. Molina-Navarro, E. Hydrological modeling and climate change impacts in an agricultural semiarid region.

Case study: Guadalupe River basin, Mexico. Water Manag. Gu, H. Projected changes in hydrological extremes in the Yangtze River Basin with an ensemble of regional climate simulations. Water 10 9 , Gulizia, C. Wilby, R. An intercomparison of a large ensemble of statistical downscaling methods over Europe: Results from the VALUE perfect predictor cross-validation experiment.

Teutschbein, C. Regional climate models for hydrological impact studies at the catchment scale: A review of recent modeling strategies. Compass 4 , — Gudmundsson, L. Technical Note: Downscaling RCM precipitation to the station scale using statistical transformations—A comparison of methods. Earth Syst. Arnold, J. Large area hydrologic modeling and assessment part I: Model development. Moriasi, D. Model evaluation guidelines for systematic quantification of accuracy in watershed simulations.

ASABE 50 , — Taye, M. Implications of climate change on hydrological extremes in the Blue Nile basin: A review. Wickham, H. Welcome to the Tidyverse. Open Source Softw. South, A. R Core Team. R Foundation for Statistical Computing, Roberts, C. Quantifying the likelihood of a continued hiatus in global warming. Her, Y. Uncertainty in hydrological analysis of climate change: multi-parameter vs. Randall, D.

Hewitson, B. Climate downscaling: Techniques and application. Shrestha, M. Bias correction of climate models for hydrological modelling—Are simple methods still useful?. Krysanova, V.

Assessment of climate and land use change impacts with SWAT. Piman, T. Assessment of hydrological changes in the lower Mekong Basin from Basin-Wide development scenarios.

Zhang, A. Veijalainen, N. National scale assessment of climate change impacts on flooding in Finland. Abbaspour, K. A continental-scale hydrology and water quality model for Europe: Calibration and uncertainty of a high-resolution large-scale SWAT model.

Shrestha, S. Neitsch, S. Book Google Scholar. The Nature Conservancy. Indicators of Hydrologic Alteration Version 7.

TNC, Download references. Authors would like to extend sincere thanks to all the organizations, especially hydro-met departments, for providing data to conduct this study. You can also search for this author in PubMed Google Scholar. Correspondence to Sangam Shrestha. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.

If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. Reprints and Permissions. Future hydrology and hydrological extremes under climate change in Asian river basins.

Sci Rep 11, Download citation. Received : 12 February Accepted : 13 August Published : 24 August Anyone you share the following link with will be able to read this content:.

Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative. By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Advanced search. Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily. Skip to main content Thank you for visiting nature. Advanced downloads Advanced downloads for. Run apps - Runtime Tooltip: Do you want to run apps? Web installer Tooltip: Downloads all required files, including localized resources, while the installer runs. An Internet connection is required during installation. No Internet connection is required after you have downloaded the installer.

Developer pack Runtime Language packs Tooltip: Adds translated error messages and other UI text to an existing installation.

Other versions. NET Framework 3.