Evaluation and upgrade of the Telemark flood model
Abstract
Flood forecasting models are important tools that could be used to find the optimum way to operate hydropower reservoirs in order to reduce flood disasters of especially human settles close to these reservoirs. The flood forecasting model for Telemarkvassdraget (FMTV), a model specifically made for the lower part of the Telemark watercourse for operational flood forecasting was used during the two floods that hit the Telemark area in September 2015 but unfortunately significant discrepancies were found in simulated reservoir level by the FMTV and the observed reservoir levels from two out of a total of four unit in the FMTV. Each model unit uses inflow foresting and reservoir routing to determine the forecast flood levels. The forecasted inflow into each model unit in the FMTV is based on a scaling of 10 days forecasted runoff series from a rainfall-runoff model (HBV model) for three unregulated catchments calibrated on historical data.The aim of this master thesis is to evaluate the existing scaling for all model units in the FMTV model by comparing a manual computation of local inflow into each model unit and comparing to the FMTV computed local inflow. This would help identify the deficient model units that need improvement. Potential strategies were evaluated for improving the scaling and the best strategy was used to come up with new scaling factors. In addition to the evaluation and upgrade of the scaling, the reservoir routing at model units Norsjo and Hjellevatn had to be evaluated to determine the best option to improve the model set up if necessary.The hydrological analysis in this master thesis is based on the September 2015 flood. Again the evaluation and upgrade of the FMTV model focused mainly on the techniques used in scaling and reservoir routing. The results of the evaluation revealed that Tinnsjo and Heddalsvatn (the first two model units) had a fairly acceptable discrepancy between simulated and observed reservoir level. The last two models, Norsjo, and Hjellevatn had a very poor discrepancy as they had a significant mismatch between their simulated and observed reservoir levels. The results of the several evaluations of all inputs into Norsjo and Hjellevatn showed that Farelva ndf Skotfoss (a gauging station situated between Norsjo and Hjellevatn measuring flow out of Norsjo and flow into Hjellevatn) was faulty and therefore is the explanation for the significant discrepancies observed in last two models.The evaluation of scaling of local inflow into Tinnsjo and Heddalsvatn showed that the scaling factors at Tinnsjo and Norsjo were really correct and hence needed some adjustment to get a better match between observed and simulated reservoir level.Trial and error analysis was used as a method of upgrading the scaling factors better the models at Tinnsjo and Heddalsvatn. New scaling factors could not be found for Norsjo and Hjellevatn due to faulty Farelva ndf Skotfoss. Another master s thesis running parallel with this project focusing on spills in Tinnsjo revealed that the current FMTV model did not include spills from Brook Intakes in the Marvatn and Mosvatn Hydropower Systems. The inclusion of the spills in Tinnsjo influenced the scaling factors that were derived without the spills from brook intakes. In conclusion based on findings in this study, it was suggested for local inflow to Tinnsjo to be scaled with donor catchment Hørte with a scaling factor of 9 if a 50 years return period flood or more is expected and (4.5*Austbygdåi ) if expected flood levels is less than 50 years.It was recommended for a new gauging station to be built and positioned downstream of Hjellevatn to make it easier to estimate the total outflow from Hjellevatn due to the difficulty in determining the outflow from Hjellevatn from its complex gates. Outflow from Vrangfoss should be audited by appropriate authority to ensure reliable outflow data since the alternative of building another gauging station downstream of Vrangfoss may be challenging technically and financially costly. A reliable stage-storage relationship for Hjellevatn should be developed to reduce the uncertainty in the water balance equation of Hjellevatn.