Keywords: long–memory properties, streamflow time series, Iran, classic R/, S analysis, aggregated variance method, detrended fluctuation analysis, DFA, MRS test, GPH test
Investigation of long–memory properties in streamflow time series in Gamasiab River, Iran
In this article, we have discussed about long–memory of processes of time series at Polchehr, Polkohne and Heydarabad hydrometric stations of Gamasiab River at Kermanshah by using daily, monthly and seasonal (1/4–year) discharge average data through heuristic, semi–parametric and parametric methods. Also, we have discussed about existence of long–memory and comparing its intensity at three timescales. If we were able to model xt series to (1 - L)dxt = εt in which εt ∼ N(0, σ2) is white noise and also 0 ≤ d ≤ 1, then series has long–memory property. If 0 ≤ d ≤ 0.5 so series variance is finite and series is generally stationary. If 0.5 ≤ d ≤ 1, series variance is infinite and series is non–stationary. The results of each of the three methods show that daily steamflow processes exhibit strong long–memory. By increasing timescale of processes, intensity of long–memory decreases. Streamflow processes in monthly timescale show less evidence of the existence of long–memory however in seasonal streamflow time series, there is no long–memory.