Improving stream flow estimation in a montane rainforest stream in Costa Rica : The impact of adding a high flow estimation

Sammanfattning

A key piece of information necessary for water management is accurate streamflow information. However, the available data and resources for gathering information vary around the world. This study forms a part of a research effort to improve discharge estimations for a montane rainforest stream located in the San Lorencito headwater catchment in Costa Rica. The aim of this study is to improve the current rating curve where the high flow part is currently ungauged and look at the impact of adding a high flow estimation to the rating curve. When adding information that is not gauged data, it is important to make sure that the information is unbiased. It can also be questioned if information based on less accurate observations will aid a model. The high flow estimation in this study was calculated using the Manning’s equation based on a survey after a flood event in August 2015. Field measurements were performed at the location in February 2017 and were added to the already existing data record from the gauging station dating back to 2015. Based on the previous data together with the new measurements, rating curves and corresponding hydrographs were established. The impact of adding the Manning’s estimation included a statistical assessment and comparison of the different rating curves and hydrographs. Results showed that at higher flows, the rating curve without the Manning’s point estimated higher discharge than the rating curve determined including the Manning’s point. However, at low flows below 0.9 𝑚3/𝑠 (or around a stage of 0.4 𝑚), the rating curve without the Manning’s point is estimating less discharge than the rating curve including the Manning’s point. Error plots and the uncertainty intervals in the graphs illustrate that the uncertainty at high flows are reduces when including a high flow estimation. The results suggest that it is preferable to add information, even if it has a lot of uncertainty, to no information at all.