Discharge Missing for Some Years Despite Precipitation in wflow.jl Model

[AldenShadd1000]

I am encountering an issue in my wflow.jl model where discharge is not being generated for certain years, even though precipitation data is present during those periods. However, for other years with similar precipitation input, the model successfully calculates and outputs discharge.

Could anyone suggest potential reasons or solutions for this issue?

Any guidance or suggestions would be greatly appreciated.

Thanks.

[verseve]

Thanks for posting this question @AldenShadd1000 .

You are mentioning that discharge is not generated for certain years. Do you mean that discharge has zero values? Or at some point NaN values (I would also expect that the run at some point fails)? And do drying rivers occurs in the catchment?

If the model runs fine and discharge is zero, and flow intermittence occurs, it is perhaps caused by higher potential evapotranspiration values?

[AldenShadd1000]

Thanks for your response @verseve .

Discharge doesn't have zero values nor any nan values, rather it has lower values from 1985 to 2005 when compared to discharge in 2006 to 2023, while precipitation pattern is similar in both periods.

I have attached the following excel file of precipitation and discharge comparison and temperature for whole model time period (1981-2023) for better understanding.

Thanks and looking forward!

Discharge_Precip_Temp_Comaparison.xlsx

[verseve]

Indeed a strange pattern. What wflow version are you using? And is water demand included in your setup? What is the potential evapotranspiration (PET) pattern for the whole period?

Difficult to say what might be the issue here. Besides checking the PET pattern, as a first step I would recommend to output gridded variables, for example variables precipitation, temperature, PET, river flow, overland flow and satwaterdepth for a part of both periods to check for possible causes.

[AldenShadd1000]

Wflow version that I am using is 1.10.4 (2024-06-04). Potential evapotranspiration for the whole period lies between 0.0 to 6.0 mm. Can you elaborate more on water demand?

I have thoroughly analyzed all the variables precipitation, temperature, PET for both periods and the their pattern is consistent and there are no anomalies in the data that could explain the issue. I have attached the forcing data file for your review to seek further clarification. Link for file: https://1drv.ms/u/s!AtArswnSDMnghHgLt_IWPJowgVrA?e=UahRHJ

Subsequently, I ran the model for the extended period from 1963 to 2023 instead of the initial range of 1981 to 2023. Interestingly, the discharge calculations started only from 1988 to 2023, while the period from 1963 to 1987 showed previous phenomenon of flat discharge values. Can you please explain why this happened? Is there a more efficient approach to run the model without requiring a 25-year warm-up period, as it involves significant labor costs.

[verseve]

The version you are mentioning is the Julia version. Do you also know the wflow version?

Did you setup the model with hydromt_wflow?

In the docs the water demand option is explained from here. Yes, as a first step checking the external forcing data is a good idea. However, I would also recommend to check the forcing variables and the other variables internally, as mentioned previously, to check if forcing variables are picked up correctly by wflow and if states/output variables can explain the pattern. Below an example to get gridded model output for some variables (TOML file):

[output]
path = "output.nc"

[output.vertical]
satwaterdepth = "satwaterdepth"
zi= "zi"
ustorelayerdepth = "ustorelayerdepth"
precipitation = "precipitation"
temperature = "temperature"
potential_evaporation = "potential_evaporation"

[output.lateral.river]
q_av = "q_av_river"

[output.lateral.land]
q_av = "q_av_land"

A 25 year warm-up period is not required, normally a 1-5 year period is sufficient.

And are reservoirs or lakes included in the catchment you're modelling? These can have a large impact on discharge patterns (e.g. these locations could have zero outflow when volume is small).

[AldenShadd1000]

Sorry for late response!

My Wflow version installed in julia is 0.8.1.

Yes, I did setup my model thorough Hydromt Wflow.

I have checked the forcing and other variables, files are correct. I"ll also print those above gridded variables to double verify it.

No, there are no lakes or reservoirs in our model.