Ч а с т ь I i I главный редактор

164
«Молодой учёный» . № 28 (370) . Июль 2021 г.
География
Correlation analysis was done for the snow data and monthly 
river flow for the months October to September. The correlation 
result (using the both linear and polynomial trendlines) is 
plotted in Figure 3. The small improvement in correlation 
given by polynomial is not considered to be justified given the 
uncertainty in the data.
Linear is used for the trendline
Polynomial is used for the trendline
Figure 3. 
Linear and polynomial trendlines
The linear correlation between 4 month shifted snow cover 
and average monthly flow can be used to predict the flow. 
The snow data (X values) and flow data (Y values) allows a 
prediction of flow to be made 4 months in advance. A satellite 
image of snow cover, available in near real-time, in February 
can be analyzed and used to forecast the flow expected in June, 
using the equation given in Table 3. Based on the correlation 
coefficient (r2 = 0.78), the accuracy of such a forecast could be 
quite reasonable.
Table 3. 
Forecast flow (Y) using linear correlation with snow cover (X)
Water forecasting method 2: percentiles
To overcome the snow data limitations, a second method 
was developed for a short-term (3-month) forecast. This method 
uses only the recorded/observed flow data. The forecast is based 
on the variability of monthly flow over the period of record. The 
differences in the monthly percentile flows for the current and 
future months, from 10 to 90 percent exceedance, when added 
to the recorded flow for the current month, gives the probability 
of flow for future months, thus predicting the future river flows. 
This simple method is especially useful as it can be effectively 
used in the critical month of April to estimate the expected peak 
of annual flow in June. The simplicity of the method also allows 
monthly updates to give greater confidence in the forecast.
The percentile exceedance of the flow data for Chogha gauge 
station from 2010 to 2018 were calculated and are shown in a 
Table 4. These percentiles are the flows which will be exceeded 
in x% of years. The analysis is based on a water year, which starts 
in October and ends in September of the following year. The 
recorded flow for 2015 and 2017, shown in red, was found from 
an examination of the daily recorded flows to contain errors. 
For this reason, the data for these two years was not included 
in estimating the percentiles.
It can be seen from Figure 4 that the percentile exceedance 
flow curves converge at the annual peak flow month of June. 
We understand that this result is a figment of the recorded data 
which underestimates high flows. The river gauging station at 
Chogha has a sand and gravel bed which mobilizes during high 
flows, increasing the cross-section of the river and hence its 
carrying capacity with little change in the water level. Although 
this gauging station cross-section is equipped with a cable way, 
there is great difficulty in measuring the cross-section’s depths 
during high flow because of the very high velocities. It may be 
possible to increase the accuracy of cross-section measurement 
during high flows by using a heavier weight on sounding line, or 
perhaps some other measurement system could be used, such 
as an echo-sounder.
We believe that in reality the annual peak flows for 30 %, 20 % 
and 10 % are under-estimated.
The percentile data is used with the observed river flow for 
the current month to give a forecast, as described earlier, by 

---