7. Discussion

7.1 Key Findings and implications for navigation

This study shows that future low flow days are generally higher than in the historical period, although clear long term trends over time are only visible under SSP370. The observed period contains an average of 27 annual low flow days, while future periods range from 31 to 49 annual low flow days depending on the climate scenario. SSP370 shows the clearest increase in low flow days between 2025-2099 and is the only scenario approaching statistical significance, with a p value of 0,054 exceeding the 0,05 threshold. This suggests the evidence for a long term increase in low flow days becomes stronger under the most severe climate scenario.

However, the distribution of annual low flow days shows that extreme low flow years become more frequent in all climate scenarios. This is most extreme for SSP370, where 25% of years exceed 60 low flow days in a year, equal to the observed period’s most extreme year. The magnitude of extreme years with the number of low flow days also increases as the most extreme year for all climate scenarios exceeds 130 low flow days, compared to the 60 days of the observed period. This is important for navigation because impacts are stronger when low flow conditions last several weeks or months. This would negatively affect indigenous communities that depend on the river for hunting, fishing, travel and transportation of resources. The severity of these impacts is scenario dependent and becomes stronger under SSP370, where both the long term trend and frequency of extreme years increase.

The seasonal river ice break up and freeze up analysis shows that the open water season expands under all future climate scenarios. River ice break up shifts from late April in the historical period toward early to mid April in the future scenarios. Freeze up shifts from early November in the historical period toward early to mid November for SSP126 and SSP245 and from mid to late November for SSP370. An expansion of the open water season may improve transportation potential because river navigation does not require the construction and maintenance of ice roads. A longer open water period may increase navigation flexibility and accessibility. However, if current ice road days are replaced by warmer open water days with low flows, transportation may become limited entirely. An expansion of the open water season can therefore be problematic for indigenous communities.

7.2 Research Limitations

A limitation of this study is regarding conceptual HBV model errors. The LAR basin has a complex structure of which some elements are not accounted for in the model. For example, the model does not explicitly account for glacier melt or non contributing areas created by potholes in the landscape. The unpredictability of ice jams make high flow or flood timings difficult to model. Using an ensemble of hydrological models or HBV model adaptations that explicitly include glacier melt, pothole storage and ice jams may yield more reliable results.

This study only uses MPI-ESM1-2 as the earth system model within CMIP6 in generating the forcing data. A more reliable approach would use an ensemble of CMIP6 models or Canada specialised CMIP6 models, allowing uncertainty between climate models to be assessed.

Another limitation is that the model assumes hydrological behaviour to remain constant over time. The HBV parameters are calibrated over a 20 year period and validated over a 5 year period but applied to future periods up to 2099. However, processes in the river basin may change over time. As mentioned in Chapter 2, snowfall contribution to total precipitation has decreased over time and might change further as temperatures keep rising. This could influence snow storages in the model, altering timing of runoff. Soil behavior may also change, for example through changes in seasonal frozen ground. This influences soil infiltration and HBV calibration parameters such as maximum soil moisture storage, soil runoff and maximum percolation rate. Therefore, the basin may respond differently to water in the future.

Finally, the usage of a fixed open water period does not account for ice break up occurring earlier causing a shift in the hydrograph. Winter precipitation is partly stored as snow and released during spring melt when temperatures exceed 0 ℃. After this period, precipitation falls as rain instead of snow. If break up occurs substantially earlier under future scenarios, part of this snowmelt driven freshet as well as a fraction of rainfall may occur before the fixed start date of 18 May. As a result of this, the period after 18 May may appear drier when in reality, the seasonal hydrograph is shifted in time. For example, days in April that were frozen in the historical period may become high flow days instead. At the same time, high flow days occurring in July during the historical period may now become low flow days.