HYPE can simulate water flow with several different time steps, between one hour and one day. The most common is to use the original daily time step. That is the longest time step available in HYPE and what the model was first developed for. Other available time step lengths are 1 hour, 2 hour, 3 hour, 4 hour, 6 hour, 8 hour, 12 hour, and 24 hour, i.e. daily time step.

To use sub-daily time steps in a simulation of HYPE require more than a change in simulation settings, the model set-up is also affected. A model set-up of HYPE with shorter time step than daily will differ from a model set-up of daily time step in three main aspects. These are:

1. time series files (i.e forcing data and other observation data files)
2. simulation settings (e.g. steplength in info)
3. the parameter values.

Each aspect will be described in detail below. In addition, at the end of this section, some miscellaneous input that may need to be considered are described.

HYPE uses forcing data and other observations as time series of the time step of the simulation. These time series files have, as for a daily model set-up, a complete time series covering all time steps for the period it has data. The first column often holding only the date for daily files is exchanged for a column with date-time. The format of the date-time is YYYY-MM-DD HH:MM, or alternatively YYYYMMDDHHMM.

For HYPE it is the date-time of the beginning of the time step that is given in the file. For example, for a time step of 12 hours the date-times of January 2, 1995 are 1999-01-02 00:00 and 1995-01-02 12:00. Similar to the daily files, the values of the files are the average over the time step, e.g. air temperature, or the sum over timestep, e.g. precipitation. Thus for an 1-hourly model set up, the content of the forcing data Tobs at 2023-11-21 10:00 is the average of the air temperature between 2023-11-21 10:00 and 2023-11-21 11:00 (in degree Celsius), while the corresponding Pobs value is the precipitation between the same times (in mm/hour).

An exception is observed lake water state (HYPE variable id wstr) used in updating of HYPE, which are the water level at the end of the timestep. This means that for this example, the value of 2023-11-21 10:00 is actually the observation of water level at 2023-11-21 11:00.

The simulation settings are all given in info.txt. The default time step is daily, so for a daily simulation the code for steplength is not needed. For other time step lengths the code steplength is set to the wanted number of hours. Some examples:

steplength 1h
steplength 6h
steplength 1d

The start and end of the simulation are given after their respective codes words in the form of date-time (format described above). In addition, all other dates in the settings are needed in date-time format. Following are some of the most commonly used date-time codes of HYPE (the model has steplength 1h in the example). All settings that use date-time are found on the info.txt page.

bdate        2011-01-01 00:00
cdate        2012-01-01 00:00
edate        2021-12-31 23:00
outstatedate 2022-01-01 00:00

The setting regestimate for regional estimated parameters by regression can be used for sub-daily time steps. Beware that if it is used on parameters with time step dependence (see next section), the regression coefficients need to be suitable for the time step length.

HYPE has many options (modeloption) for using different formulations for model processes or other functions. They are set in info.txt and listed there. Many modeloptions related to water simulations work without problem with sub-daily time steps. See below for full list on status of modeloptions (for water and for substances).

Updating discharge and lake water level with the updating methods work for sub-daily time step simulations. Updating of discharge and lake water level with the auto-regressive method has time step dependent input in the update.txt file. The AR-factors used in these updating methods, qar and warupd, and defined in update.txt, need to be adjusted.

The simulation of substances (substance) with HYPE is not fully developed for sub-daily time steps.

The setting readdaily work and is recommended for sub-daily time steps. It means that files are read each time step to get the current forcing and observations.

The ordinary output formats work for sub-daily time steps. The available output aggregation times are the time step of the simulation, daily, monthly, yearly, and for the whole simulation period.

Criteria calculations should work on sub-daily time step simulations, but only the most common ones have been tested.

Model parameters that depend on the timestep need to be set accordingly. HYPE holds most of its parameters in the file par.txt (see list), but some of them can also be set in reg_par.txt. Both these files (if used) must be adapted.

Time step dependent parameters (i.e those with a unit of time steps) need to be adjusted according to the timestep. If there is a daily model setup that is to be adapted, they can as a first estimate be rescaled (unit transformed) according to the difference in time step length. There is a function in the R-package HYPEtools (link to HYPEtools) that can do this for you (ScalePar).

Table with current (HYPE version 5.25.0) time step dependent parameters for water simulation.

NOTE: The parameters for snow density increase per day (snowdensdt) and snow compaction for lake and river ice model (licesndens and ricesndens) have the unit g cm^-3 d^-1 in the current HYPE code and do not need to be rescaled. They have erronous had the unit g cm^-3 ts^-1 stated on the wiki page of par.txt. These parameters may become time step dependent in future HYPE version.

Only one time step dependent parameter can be found in reg_par.txt. That is rrcs3. This parameter can be used with regional parameter estimation by regression.

This method to rescale time step dependent parameters by unit transformation, has been used and shown to be working rather well (Fuentes-Andino et al. 2022). Still, some may work better with another rescaling. Other parameters may need recalibration or adjustment for a simulation with sub-daily time steps. Interesting parameters/processes for further adjustment, may be found reading the model description, studying the model simulation, or reading the detailed description of how HYPE functions are working for shorter time steps (section below). The parameters recalibrated in previous work can also be used as a starting point if the directly rescaled model is not working well.

Table with parameters adjusted/calibrated in previous work at SMHI with simulation with sub-daily time steps.

Starting state files saved from one model cannot be used for another model with other time steps. The number of states will be different.

There are no other input related to flow simulation that need to be changed for a simulation with a sub-daily time step. It is good to remember though, that the flow input in PointsourceData.txt and GeoData.txt should not be changed. This because, the flow from point sources and local diffuse sources should be given in m3/day also for simulations with sub-daily time steps.

NOTE: The above adaptations is true for a HYPE simulation of water. For substance simulations there is other input that might be affected. HYPE is not yet fully adapted to handle a simulation of sub-daily time steps than days with substances. See next section for details on which parts of the HYPE model work and which may not.

This section holds information on if and how specific HYPE model parts works for sub-daily time steps. There are functions which workings have not been checked or tested.

This section is under construction and will be complemented.

In general, the water simulation parts of the HYPE model work, with some exceptions, while the substance simulations have several model parts that is known not to work well. The substance simulation processes have not been checked in any systematic way. A few processes has been adapted when other changes made it suitable. Still, simulations with substances are running, but the inner function of some processes are unknown. Models using sub-daily time steps have been used and reported in e.g. Fuentes-Andino et al. 2022, Olsson et al. 2017, and Ivarsson et al. 2017.

The status of model parts has been divided into four tables. The split is made between models of water and substances, and between modeloptions set in info.txt and other model choices.

Table with status of current modeloption (HYPE version 5.25.0) related to water simulation with sub-daily time steps.

Table with status of model parts related to water simulation with sub-daily time steps. These are part of the model that is goverened by input data or model parameters.

Table with status of current modeloption (HYPE version 5.25.0) related to substance simulation with sub-daily time steps.

Table with status of model parts related to substance simulation with sub-daily time steps. These are part of the model that is goverened by input data or model parameters.

Fuentes-Andino, D., Y. Hundecha, G. Lindström and J. Olsson, 2022. Exploring the potential for parameter transfer from daily to hourly time step in the HYPE model for Sweden. Hydrological Science Journal, 67(13):1988-2001. https://doi.org/10.1080/02626667.2022.2121165

Ivarsson, C.-L., Olsson, J., Pers, C. and Y. Hundecha, 2017. High-resolution ensemble flood forecasting: A case study in Höje å, Sweden, Vatten - Journal of Water Management and Research 73:85–92.

Olsson, J., Pers, B.C., Bengtsson, L., Pechlivanidis, I., Berg, P., and H. Körnich, 2017. Distance-dependent depth-duration analysis in high-resolution hydro-meteorological ensemble forecasting: a case study in Malmö City, Sweden. Environmental Modelling and Software. 93:381-397. https://doi.org/10.1016/j.envsoft.2017.03.025.