You can rely on SPSS to generate missing values which is simply the mean fot the entire period but you may use many other ways like fitting a time series model by ARIMA modelling / Time Series Analysis which is simply executed in SPSS.

All the big data handling software like SPSS and even Origin has fits to compensate for intermediate missing data points

Naveed Ahmed there is a nice program called RRL (Rainfall Runoff Library) if you have rainfall and maybe evapotranspiration data. And another good thing about the software is that it provides dynamic adjustment while you change your parameters until you get a good matching with your available data. Have a look at this link https://toolkit.ewater.org.au/Tools/RRL.

There are different techniques for filling the missing hydro/meteorological data such as linear regression, inverse distance interpolation, normal ratio..... I would recommend to read this paper Article Gap filling and homogenization of climatological datasets in...

Hi, you can use MICE package in R.

Amelia too is good. Rainfall Runoff library is also a good tool but you have to be aware that it's a combination of hydrological models like sacremento,awbm and tank model.

Mekonnen Dagnenet Nice sharing but this paper is related to Temperature and Precipitaions and very common techniques have been used in this paper. The discharges data is not such easy to generate by these techniques. this is my opinion. I have to wait for some more techinical comments.

Regards

Naveed.

Naveed Ahmed I have to advise you to look for techniques of streamflow regionalization. Statistical tools, like linear-nonlinear regression and models, may result in well adjusted values, but usually they may conduct you to poor result at extreme values. This article is a good starting point: https://ascelibrary.org/doi/10.1061/%28ASCE%29HE.1943-5584.0000690

The best way to do this is to build a predictive statistical model. But first I must ask are there any other data available in the region that could be used as predictive variables in a model? This would be discharge from another river, precipitation, ground water levels, soil moisture, etc. If not, then the problem becomes more challenging.

Naveed--if there are other rivers in the region whose flow has been gaged for some overlapping time period with your river of interest; a quick and easy way to estimate the missing data is to check for correlations (or even nonlinear relationships) between flow in your river of interest and flow in these other rivers; develop relationships separately for each monthly or day-of-year time period: for seasonal time periods where correlations are strong (e.g. R^2>~0.85) then using the inter-river flow relationship(s) can yield reliable and accurate flow estimates for the specific time periods where the data for your river of interest is missing. It is also straightforward to compute confidence intervals (using standard statistical methods) for the accuracy of such flow estimates.

There are more complicated ways to optimize the information used in such analyses; let me know if you're interested in this general approach and I should be able to find my old references on this and cite them for you--

Jim T.

Hi every body

As all colleagues stated there are several methods for filling missing data in a time series. based on my experiences KNN method is one of the best one for this issue. You can use MATLAB and load knnimpute function to find the missing data...in MATLAB help you find more information about KNN method

the best

Azizian

Hi Naveed,

Assuming you have temperature and precipitation data, you could setup and calibrate a rainfall runoff model to predict missing flows. If instrumental data are not available, most places in the world have gridded meteorlogical data which often works better for runoff simulation than direct measurements from a met. station. You could make the met data as simple or as complex as you like, we've had surprisingly good results for most of the Indian subcontinent using a single precipitation and temperature time series to model river flows. There are a range of straight forward and easy to use rainfall runoff models. My group has had success with Jan Seibert's HBV-Light model and my PERSiST model.

Sincerely,

Martyn

If you have data at other station in the basin, you can try to generate missing data based on correlation relationships or try to estimate using rainfall-runoff model or using rational formula with precipitation data or other approach.

one way is to plot all the available flows data with time, then use the trend equation to find missing data if the behavior was linear versus time.

another one for general behavior, building up an interpolation statistical model to predict miss data.