508
G. Romanescu and I. Minea
– mathematical modelling of surface and underground water runoff;
– warnings in case of extreme hydrological phenomena (floods, droughts, frost) for
crisis prevention and management.
Forecasts always involve probability. A physical model features the values that
appear rarely, and that favours extreme events, such as droughts or tidal waves. The
forecasting of hydrological elements is only possible under the conditions of cyclic
repeatability, over a long period. Unfortunately, the hydrological information is on
a recent date. The future cannot be an replica of the past. From both a theoretical
and a practical perspective, it is impossible to regulate completely, the mean multiannual flow used. The reservoirs capable of such annual or multi-annual regulation
are situated on the rivers of Bistrita (HPP Stejaru), Arges (HPP Corbeni), Lotru (HPP
Ciunget), Somes (HPP Mariselu), and Sebes (HPP Calceag). On the Danube and on
the Olt River, storage is not possible, though, this river does hold 33% of Romania’s
hydroelectric power potential.
The most common long-term—up to 12 months—forecast method used, is the
Extended Streamflow Prediction (ESP). The ESP system is that part of the NWSRFS (National Weather Service River Forecast System) that provides the possibility of making long-term forecasts for streams. The ESP uses conceptual hydrologic/hydraulic models to forecast future streamflow, using the current snow, soil
moisture, river, and reservoir conditions with historical meteorological data. The
ESP procedure assumes that some meteorological events that occurred in the past
are representative of events that may occur in the future. Each year of historical
meteorological data is assumed to be a possible representation of the future and is
used to simulate a streamflow trace. The simulated streamflow traces can be scanned
for maximum flow, minimum flow, the volume of flow, reservoir stage, and so on, for
any period in the future. ESP produces a probabilistic forecast for each streamflow
variable and the period of interest. Historical simulation is generated as part of a 12month interval. The error depends on the type of flow: for minimum flow, 10–15%;
for average flow 50%. For a long period (up to 6 months), another viable method to
apply is the Multi-model forecast technique (applied by the Romanian experts Adler
Mary-Jeanne and Valentina Ungureanu in 2006) [1].
Four models (Deterministic statistical model; Hydro-meteorological analogues;
Stochastic models; Deterministic model) have been used for low flow forecasting
in the extreme droughty period between the months of June and November, 1994.
The models from which the best data were obtained are the Stochastic model and
the Stanford model. The use of the Multi-model improves the forecast; during these
months, the forecasting errors represent 10%. Therefore, the multi-model technique
is highly effective. The use of the Multi-model technique increases the forecasting performances, and this is why it is recommended for operational activity (see
Table 15.8) [15].
Hidroelectrica S.A. declared insolvency and made a change in the Energetic Strategy of Romania for the period between the year 2011 and 2035. In this case, there
is the possibility of compromising the objectives assumed by Romania according
to Annex I of the Directive, 2009/28/EC, which stipulates that the objective set for
G. Romanescu and I. Minea
– mathematical modelling of surface and underground water runoff;
– warnings in case of extreme hydrological phenomena (floods, droughts, frost) for
crisis prevention and management.
Forecasts always involve probability. A physical model features the values that
appear rarely, and that favours extreme events, such as droughts or tidal waves. The
forecasting of hydrological elements is only possible under the conditions of cyclic
repeatability, over a long period. Unfortunately, the hydrological information is on
a recent date. The future cannot be an replica of the past. From both a theoretical
and a practical perspective, it is impossible to regulate completely, the mean multiannual flow used. The reservoirs capable of such annual or multi-annual regulation
are situated on the rivers of Bistrita (HPP Stejaru), Arges (HPP Corbeni), Lotru (HPP
Ciunget), Somes (HPP Mariselu), and Sebes (HPP Calceag). On the Danube and on
the Olt River, storage is not possible, though, this river does hold 33% of Romania’s
hydroelectric power potential.
The most common long-term—up to 12 months—forecast method used, is the
Extended Streamflow Prediction (ESP). The ESP system is that part of the NWSRFS (National Weather Service River Forecast System) that provides the possibility of making long-term forecasts for streams. The ESP uses conceptual hydrologic/hydraulic models to forecast future streamflow, using the current snow, soil
moisture, river, and reservoir conditions with historical meteorological data. The
ESP procedure assumes that some meteorological events that occurred in the past
are representative of events that may occur in the future. Each year of historical
meteorological data is assumed to be a possible representation of the future and is
used to simulate a streamflow trace. The simulated streamflow traces can be scanned
for maximum flow, minimum flow, the volume of flow, reservoir stage, and so on, for
any period in the future. ESP produces a probabilistic forecast for each streamflow
variable and the period of interest. Historical simulation is generated as part of a 12month interval. The error depends on the type of flow: for minimum flow, 10–15%;
for average flow 50%. For a long period (up to 6 months), another viable method to
apply is the Multi-model forecast technique (applied by the Romanian experts Adler
Mary-Jeanne and Valentina Ungureanu in 2006) [1].
Four models (Deterministic statistical model; Hydro-meteorological analogues;
Stochastic models; Deterministic model) have been used for low flow forecasting
in the extreme droughty period between the months of June and November, 1994.
The models from which the best data were obtained are the Stochastic model and
the Stanford model. The use of the Multi-model improves the forecast; during these
months, the forecasting errors represent 10%. Therefore, the multi-model technique
is highly effective. The use of the Multi-model technique increases the forecasting performances, and this is why it is recommended for operational activity (see
Table 15.8) [15].
Hidroelectrica S.A. declared insolvency and made a change in the Energetic Strategy of Romania for the period between the year 2011 and 2035. In this case, there
is the possibility of compromising the objectives assumed by Romania according
to Annex I of the Directive, 2009/28/EC, which stipulates that the objective set for