I would suggest to callibrate only with the high quality gauge station because otherwise you will introduce errors in to parameter values. Wrong data will lead to wrong model parameters and hence a wrong model identification. May be you can use the second discharge gauge station to compare timing of discharge changes and use the data for evaluating time lags between the stations using cross correlation analysis. Most times water level discharge relationships are the most important error of discharge measurements.

Michael

For such a large catchment I would definitely go for a multi-site calibration. It's hard to imagine that an area of 150,000 km2 is homogenous. You can grasp some of heterogeneity by using spatial calibration. So I would suggest using not 1 or 2 but as many as possible stations.

As regards question 2, it depends what is your calibration approach. For instance, if you're planning to use SWAT-CUP, in this software you can define the measurement error in your observed data, ie. you can set a higher error for your most downstream station. This would affect calculation of the uncertainty bounds.

Hope this helps.

We test with one discharge gauge because we don´t have more gauges for calibrating the model. But we are now fixing crops inputs because evapotranspiration doesn´t generate a good performace. When we finish this work i´ll send you results but in spanish for now.

What parameters do you would like to calibrate ? How you are confident in measured parameters ? All the uncertainties will be lumped in the parameter that you are calibrating. In many cases is more important a good assessment of the model parameters respect to an extensive calibration on measured data. Calibration could be carried out in the high quality catchment, and the other catchment, even if data are not of good quality, can be used for a informal (or formal) validation of the model. If you use a multiple site calibration, in general calibrated values are inside the range of physically acceptable range. IN some case, with a single site calibration the calibrate values are unrealistic. There is a large literature about model calibration on the methods and on the choice of the objective function.

Please do calibration at multiple stations

i am also modelling in ArcSwat and i m using three rain gauge stations, i m getting good result. but I would suggest to calibrate with high quality data.

It is better to calibrate the model with one station firstly (perphaps upper-mid reaches one), since the downstream station may be largely influenced by human activities. Of course, the model should be calibrated by multi-gauges due to high hetergenities of land surfaces for the final results.

I have the same questions and i am using MikeShe by DHI from my Thesis , but for my luck i have only one gauge station and the same area... , i have many meteorological stations at least.. but no edaphological data .I think that with one station you will be ok .. because most of this models works better if you keep it simple

i am new for this intellectual site, it is great job; if anyone of you know that there are also people experimenting with ARCSWAT in Nepal, i would appreciate you so much, thanks.

As a follow up to the answers given I suggest contributors should read this comment: (http://onlinelibrary.wiley.com/doi/10.1002/hyp.6491/abstract) They give some very good insight into multi-site calibration. Please read this reference and share your thoughts.

http://onlinelibrary.wiley.com/doi/10.1002/hyp.6491/abstract

i like the topic very much but unfortunately, i dont know the implementation part of the model (although i want to learn it).. good luck. by the way, how can i get the ARC SWAT software?

I would recommend you to do a multi-site calibration. SWAT-CUP is a good software to know the error of your measurements.