The descriptions and analyses of monthly naturalized stream flows were presented as anomalies (departures from average) at the hy-drometric stations and used to characterize the drought conditions at stations along the Alberta-Saskatchewan (A-S) and the Saskatch-ewan- Manitoba (S-M) borders. As noted earlier, these stations were chosen because of their relevance to the apportionment of water between provinces which is particularly critical during periods of drought. Figure 1 shows the location of these stations.
Figure 2. Monthly naturalized flows for the North Saskatchewan River near Blindloss.
Station #9: North Saskatchewan River near Deer Creek
Figure 3. The relationship between the actual flows and the naturalized flows for the North Saskatchewan River at the A-S border. The straight line represents the points where actual streamflow equals naturalized streamflow.
The naturalized flow values are used by provincial and federal water managers when assessing whether commitments to transfer 50% of the annual water available in the upstream province across the bor-der to the downstream province have been achieved. This analysis was completed for 8 stations (Assiniboine River at Kamsack, Church-ill River at Saskatchewan-Manitoba border, North Saskatchewan at Prince Albert, North Saskatchewan near Deer Creek, Red River near Blindloss, Saskatchewan River at the Pas, South Saskatchewan River above Red Deer River, and South Saskatchewan River at Medicine Hat) for the period of 1977-2006 (30 years) and supplemented by 5 more stations (Middle Creek near the Saskatchewan Boundary, Lodge Creek near the Alberta Boundary, Battle Creek at the Alberta Boundary, Red Deer River near Erwood, Pipestone Creek near the Saskatchewan Boundary) for the 1985-2006 period (22 years).
The descriptions and analyses of monthly actual streamflows were also presented as anomalies (departures from average) at the same hydrometric stations and used to characterize the drought con-ditions. The results from this analysis were also compared to the findings of Hanesiak et al. (2011) who analyzed the drought from a number of perspectives, including considerations of meteorological and land surface conditions.
Analysis of naturalized (apportionment) flows and compari-sons to actual flows
Single station analysis:
Figure 2 shows the normalized flows on the North Saskatchewan near Deer Creek. Peak flows in the North Saskatchewan occur in the
summer (June to August). This reflects the distance that water travels from the source regions in the Rocky Mountains along the river to reach the A-S border. In addition, the snowmelt and glacier melt in the Rocky Mountains continues well into the summer in the case of snow and into the fall in the case of glaciers contributing to peak flows in the June to July period. From these results it is clear that 1999 was not a hydrologic drought year along the Saskatchewan River at the A-S border. However drier conditions occurred in 2000 with most months below average. Other very dry seasons included the summers of 2001 and 2002 and the fall of 2003.
During the period from 2001 to 2004 the naturalized flows in the North Saskatchewan were below average on a number of occasions.
The spring flows in 2001 and 2002 were lower than average with the
Figure 4. Monthly naturalized flows for the Red Deer River at Blindloss.
Figure 5. The relationship between the actual flows and the naturalized flows for the Red Deer River at Bindloss (on Alberta-Saskatchewan border). The straight line shows the one-to-one match between actual flow and naturalized flow.
lowest relative flows tending to come in September. The departures from average for the naturalized flow were significantly larger that the departures from average in the actual flows. At this station both actual and naturalized flows began to recover in the last half of 2004 and returned to near normal in 2005.
One issue affecting the interpretation of naturalized flows involves their relation to actual flows. Scatter plots were developed for border stations to assess the relationships at different stations between the naturalized flows and the measured flows during the summer and winter months. Figure 3 shows the distribution of naturalized flows versus actual flows for the North Saskatchewan River near the A-S border. The results of this analysis indicate that these monthly aver-age flows are relatively well correlated especially when the actual flow exceeds 400 m3/s although the naturalized flows are systemati-cally higher than the actual flows. During periods when monthly av-erage flows are less than 400 m3/s they vary widely. Since drought is a low-flow event, naturalized flows during droughts are likely less than actual flows especially when the actual flow is less than 135 m3/s.
The naturalized flows and actual flows for the Red Deer River, which drains a large area in central and eastern Alberta, are more consist-ent than they are at some of the other border stations. At Blindloss, the natural flows and actual flows on the Red Deer River (see Figure 4 for naturalized streamflow and the scatterplot for naturalized ver-sus actual flows in Figure 5) are very similar indicating that human interventions on the river are less likely to have an impact on the measurements and that naturalized flows are a more reliable indica-tor of the severity of a hydrologic drought.
At Blindloss where the river enters Saskatchewan from Alberta, flows tend to have a relatively small annual cycle because the spring fresh-et is supplied by snowmelt on the prairies in central and eastern Alberta rather than the foothills. Beginning in the summer of 2000 the flows were below normal and continued to be below normal until April 2003 when they rose above average for two months only to decrease to below average values by June. According to the actual streamflow values the most significant period of streamflow drought
occurred in August 2003. Figure 5 shows that there is good agree-ment between the naturalized and the actual flows at Bindloss.
Drought history derived from naturalized flows:
Based on the normalized flows an analysis of the drought was carried out that was in good agreement with the much broader drought analsyis carried out by Hanesiak et. al. (2011). Single station graphs were used to assess the progress of the drought along the A-S border and the S-M border. Similar results were obtained between natural-ized and actual flows although a few shifts were noted a specific stations. The text below describes the signature of the drought found in naturalized and actual flows along the borders.
1999:
During this year the naturalized flow along the southern parts of the A-S border was below average in the latitudes south of 50oN. How-ever, these are relatively small tributaries and which are nearly dry during many summers. Further north flows appeared to be average or even above average (North Saskatchewan River). Naturalized flows at stations on the southern part of the S-M border tended to be at or above average except for a below average flows on the Saskatch-ewan River in the April to July time frame. Farther north, the Churchill River was below average for the entire year reflecting the effects of the below average precipitation in the northern prairies. It appeared that the drought did not have a significant hydrologic signal in 1999.
2000:
Along the A-S border the hydrologic drought effects appeared at stations in the south. Naturalized flows in the central part of the A-S border were below average with both the Lodge and Middle Creeks going dry. The Red Deer River showed a large decrease beginning in August. The use of upstream irrigation to compensate for the lack of precipitation may have affected the drawdown rates in this basin. Naturalized flows along the southern part of the S-M border were below average with virtually no spring melt peak although
secondary peaks in June and July appeared in both the Qu’Appelle and Assiniboine Rivers. The Saskatchewan River flow was also below normal from May on-ward. Further north, the Churchill River flow was above average for most of the summer.
2001:
During 2001 both actual and naturalized flows were very low at stations along the A-S border. These low flows indicated a much smaller snowmelt contribution than most other years. Flows in the larger basins decreased in 2001 with very low values beginning in the South Saskatchewan River in May 2001, the Red Deer River in June 2001 and the North Saskatchewan in September 2001. The Red Deer River was above average in May and then decreased rapidly to the point where June and July flows were well below average. The flows at stations further north were also lower than average for the entire year. Along the S-M border, the naturalized flows were average to slightly above average for the southern stations in the spring of 2001 but decreased to below average after June in the Qu’Appelle River. Further north, the Saskatchewan River was below average for the entire year while the Churchill River was below average with very low flows in the May to September period. These decreases were more dramatic in the naturalized flows than in the actual flows.
2002:
For the southern rivers on the A-S borders, the naturalized flows were below average but increased rapidly in May in the Red Deer River Basin and in June in the South Saskatchewan River. Flows for the other basins were below average with the exception of a peak in June in the Battle Creek and Lodge Basins. These increases may have been associated with a storm event that affected southern Alberta (Stewart et al. 2011), The drought effects also were clearly evident in the southern stations along the S-M border with naturalized flows well below average on the Qu’Appelle River and the spring peak was absent for the Assiniboine River. The Saskatchewan River naturalized flow was also below normal except for above average flows in the months of June and July, possibly associated with convective storms
in the South Saskatchewan River Basin. The Churchill River, where flows appeared to be decoupled from the southern basins in many years, remained below average until October.
2003:
The spring flows in the South Saskatchewan and Red Deer Rivers on the A-S border were above average until May and then dropped off rapidly in the summer as the drought conditions intensified. There were no significant decreases in any other stations in 2003 so it ap-peared that the hydrologic drought had ended along the A-S border at the end of 2003. On the M-S border the Assiniboine River had a significant spring peak but a strong recession quickly reduced the
flows to below average flows. However, the Qu’Appelle was above average in the spring and remained above average for most of the year. The Saskatchewan River naturalized flow was below average until July while the Churchill River was slightly below average.
2004:
Naturalized flows along the southern A-S border were generally near average for the Red Deer River and the South Saskatchewan Rivers while flows in the Lodge and Middle Creeks were below average.
Other stations along the border varied from above average to below average but only the North Saskatchewan River, which started the year below average but recovered by September, showed a
signifi-cant streamflow deficit. Along the S-M border naturalized flows on the Qu’Appelle River and Saskatchewan Rivers were below average until the summer. The Assiniboine River was average with the ex-ception of an above average spring runoff while further north the Churchill was slightly above average during the latter half of the year.
Based on these results it appeared that the hydrologic drought had ended on most of the rivers by the close of 2004.
Assessment
While the use of hydrographs from actual and naturalized flows may appear to be very basic it is a useful step in understanding the be-havior of drought at each hydrometric station. Furthermore, it does not appear that the use of naturalized flows rather than actual flows would influence conclusions from this approach except where large uncertainties exist in the data being used in the naturalized flows.
In some cases the naturalized flows departed significantly from the actual flows making interpretation of the results more complex since the factors accounting for these differences were not monitored.
However, in nearly all cases the direction of the anomalies (positive or negative) that were present in the actual streamflow also existed in the normalized streamflow estimates. A major limitation in using the naturalized flows for scientific applications arises from the reser-vation of both the data providers and experts about the unquanti-fied uncertainties in the naturalized flows. In order for naturalized flows to be most useful in water management, the authors believe that it is important to document the uncertainties and variabilities that are used in calculating these flows. Ultimately these values would be most useful if these flows were calculated with the same methodologies for each station and the values of each input to the naturalized flow calculation was made available.