Hi Martin

i working on time series models. i think you can use R software for this goal and i propose to assessment AR, MA, ARMA, ARIMA and SARIMA models and finally select the best model (in simulation) and then use the best model for forecasting.

thanks Mohammad. Can I use time-series models to model spatial changes? Instead I was thinking more in some kind of kriging analysis

Yes you can do it. but you must use the spatial changes instead of time series changes. a column must be X1,X2,X3,... Xn  that 1,2,3 indicated the spatial changes and another column is DO or BOD. when you forecast DO or BOD (for example 48 month ahead), you will can create the zoning map of  DO or BOD in basin.

I'm sorry for the bad writing.

thank you very much Mohammad

Hopefully this very brief answer can help you Martin .

The air to water mass-transfer rate of oxygen (Na) can possibly be considered as rate-controlling. It depends on the local oxygen deficit (Cs - C), where Cs is the dissolved oxygen concentration (DO) for complete saturation, while C stands for the actual (measured) DO: Na = KLa·(Cs-C). Here, KLa is a mass transfer coefficient that binds together the mass transfer coefficient for the liquid side with the interfacial area, since it is not practical to treat interfacial area as a separate variable. From the time-series array of DO at distinct spatial locations, estimates of KLa can be derived. This estimated KLa array could possibly provide an insightful complementary perspective to the analysis of the original DO data. Such KLa estimates can be derived off-line from previously acquired data, or on-line and on real-time while data is being acquired. Additional clues to last approach can be taken from my MSc Thesis: C.M.G.A. Queirós, "Controlo do Oxigénio Dissolvido em Fermentadores para Minimização de Energia Consumida", MSc Thesis, Instituto Superior Técnico, Lisbon, 1997 (in Portuguese).

Thesis Controlo do Oxigénio Dissolvido em Fermentadores para Minimi...