Dear Yu Liu ,

In general, if  observed data of streamflow is not available, then for validation of the model, some other method has to be adopted. When you are trying to validate by any other method, you must have very strong justification. If you have got any reference where, in the absence of observed streamflow data, application of pressure data to estimate the flow has been performed, its definitely welcome.

Now, regarding your particular case, I have mentioned in an answer to your last question that, calibration don't have much meaning when the data available is less than even 1% (9 days in 6*365 days).

Moreover, in the datalogger flow calibration, the R2 value is very low i.e. 0.3238. This is the result when matched for only 9 days data. If you use such an erroneous method to estimate observed flows, the result will be misleading. You can't really justify it.

I hope I am able to answer your queries.

All the best !

Hi, Sabyasachi Swain. Thank you indeed for your help!

If I begin to measure the flow data today, until I get two month of flow data, then I use these data create a new formula which contain 60 points. After this, I will apply this new formula and 6 years pressure data to estimate the flow data, do you think it works?

Yes, if it will show a good agreement (R2), then that is definitely welcome. The model or equation if very well fits for two months, we can postulate that the conditions of land use or soil has not changed much in 6 years. So it can expected that it will work for 6 years as well. You can justify like this.

I hope my answer will help. Good luck!

Thanks a lot!