Abstract:
The forecasting of river flow has been an important goal for scientists and engineers for
centuries. This research attempts to explore the nature and strength of possible
teleconnections between the riverflows of Jamuna and Surma and El Nino Southern
Oscillation (ENSO) variability over the equatorial Pacific Ocean. Hence using the ENSO
information efforts have been put to develop flood forecast model for each of the rivers
which can capture, at least in part, the natural variability of flows with a reasonable lead
time. Traditional hydrologic forecasts of the basin through rainfall-runoff modeling could
provide a lead time on the order of basin response time, which may be several days or so.
Such a short time is inadequate for water resources planning and to hedge against the
extreme events in large river basins. Moreover absence of adequate and authentic data
from upper catchments has made this type of forecasting less appreciable in our country.
So attempts have been made to develop a long lead forecast model using infonnation like
ENSO.
This research demonstrates a notewOlihy relationship between natural variability of
average flood flows of the month July-August September (JAS) of the Jamuna and Sunna
rivers with ENSO index of the conesponding months. Here sea surface temperature
(SST) has been used as ENSO index. The cOlTelation analysis between SST and
riverflows of Jamuna for different time period shows a notable improvement in
relationship from the beginning of 1980s. However for Surma all the available data from
1969 to 2003 signifies this conelation. Then discriminant analyses have been perfornled
to evaluate the possibilities of long lead forecast. From discriminant analyses it has been
found that high flood events are mostly associated with La Nina i.e. the cold episode of
ENSO and low floods are linked with El Nino which is the warm episode of ENSO. For
Jamuna, possibility of high flow in a cold event is 100% and for Surma, it is 86%. These
links prompted the development of a statistical model which will allow a forecast lead
time up to one year. Through investigation it has been found that ENSO index and its
gradients are statistically related to wet season flows of those selected rivers.
Subsequently using the previous flow record, predicted ENSO data and its gradient,
separate flood forecast model has been developed for each of these two rivers.Verifications of the models have been done for the years 2000 to 2004 and the flow for
the year 2005 has been forecasted. The flow forecast model for Jamuna shows a
maximum error of 15% at 3 month lead and 20% at 6 month lead in the verification
period. While for Sunna, it is within 15% for both of three and six month lead. The
appreciable perfonnances of the models in the verification period assure the potentiality
of the proposed approach for long-tenn planning of water resource management,
agricultural practices and disaster relief preparation.
