the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Direct integration of reservoirs’ operations in a hydrological model for streamflow estimation: coupling a CLSTM model with MOHID-Land
Ana R. Oliveira
Tiago B. Ramos
Lígia Pinto
Ramiro Neves
Abstract. Knowledge about streamflow regimes and values is essential for different activities and situations, in which justified decisions must be made. However, streamflow behavior is commonly assumed as non-linear, being controlled by various mechanisms that act on different temporal and spatial scales, making its estimate challenging. An example is the construction and operation of infrastructures such as dams and reservoirs in rivers. The challenges faced by modelers to correctly describe the impact of dams on hydrological systems are considerable. In this study, an already implemented, calibrated, and validated solution of MOHID-Land model for natural regime flow in Ulla River basin was considered as baseline. The referred watershed comprehends three reservoirs. Outflow values were estimated considering a basic operation rule for two of them (run-of-the-river dams) and considering a data-driven model of Convolutional Long Short-Term Memory (CLSTM) type for the other (high-capacity dam). The outflow values obtained with the CLSTM model were imposed in the hydrological model, while the hydrological model fed the CLSTM model with the level and the inflow of the reservoir. This coupled system was daily evaluated in two hydrometric stations located downstream of the reservoirs, resulting in an improved performance compared with the baseline application. The analysis of the modelled values with and without reservoirs further demonstrated that considering dams’ operations in the hydrological model resulted in an increase of the streamflow during the dry season and a decrease during the wet season but with no differences in the average streamflow. The coupled system is thus a promising solution for improving streamflow estimates in modified rivers.
Ana R. Oliveira et al.
Status: open (until 19 Jul 2023)
-
CC1: 'Comment on egusphere-2023-915', Ningpeng Dong, 27 May 2023
reply
The authors present an interesting and significant work on the hydrologic modelling in reservoir-regulated basins. I have a minor point that I would like to discuss with the authors.
The authors stated in Section 2.3.1 that they did not use reservoir outflow as a forcing variable of the CLSTM model, because there can be an accumulation and propagation of errors when the simulated outflow at the current timestep recurrently serves as the model input in the next timestep. I would like to know if the authors have actually performed this experiment. Based on my personal experiences and some existing literature, this can be true if we simulate the reservoir outflow only with the machine learning algorithm and do not consider any physical constraints. However, this issue may be largely avoided if we impose some of the physical constriants of reservoir operation to the machine learning models, for example, the reservoir water balance, upper/lower limit of reservoir storage & outflow, etc. The authors may want to reconsider this expression.
See also: Ehsani, N., Fekete, B.M., Vorosmarty, C.J., Tessler, Z.D., 2016. A neural network based general reservoir operation scheme. Stoch. Env. Res. Risk A. https://doi.org/10.1007/s00477-015-1147-9.
Citation: https://doi.org/10.5194/egusphere-2023-915-CC1 -
AC1: 'Reply on CC1', Ana Oliveira, 02 Jun 2023
reply
Dear Ningpeng Dong,
First, we are happy that you found our study interesting, and we would like to thank you for your comment.
Concerning the CLSTM model developed in this study, it was not tested for the accumulation and propagation of errors when the simulated outflow at the current timestep recurrently serves as the model input in the next time step. However, that statement refers to the CLSTM model considered in the study, which does not include any constraint.
Nevertheless, I have read the paper and it is perceptible that authors conclude that inflow and outflow values are not the main variables impacting the outflow estimation. However, it was unclear to me that they tested to feed the model with released values already estimated by that same model. It seems that the authors used only observed values, which do not allow to ascertain how errors evolve when the model feeds itself.
Even so, and since this point made arise the discussion, we consider that it can be clarified in the revised version of the manuscript, stating that the authors of the referred study concluded that outflow values from previous days do not have a significant impact on the calculation of the outflow on the current day.Sincerely,
Ana R. Oliveira
Citation: https://doi.org/10.5194/egusphere-2023-915-AC1
-
AC1: 'Reply on CC1', Ana Oliveira, 02 Jun 2023
reply
Ana R. Oliveira et al.
Ana R. Oliveira et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
126 | 34 | 10 | 170 | 2 | 2 |
- HTML: 126
- PDF: 34
- XML: 10
- Total: 170
- BibTeX: 2
- EndNote: 2
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1