Reduced-Order Salinity Modeling of the Urmia Lake Using MIKE3 and Proper Orthogonal Decomposition Models

Abstract

Increasing salinity in Urmia Lake, located in the north-west of Iran, has turned into a critical issue, particularly because the lake is the habitat of a unique multi-cellular organism called Artemia Urmiana. During the past decades, several anthropogenic changes have taken place in the lake, which have resulted in increased salinity. This study introduces a reduced-order framework based on MIKE3 simulation model and proper orthogonal decomposition (POD) to simulate salinity patterns in Urmia Lake. Spatio-temporal variations of salinity in the lake firstly were simulated by MIKE3, and close matches were observed between salinity estimates from MIKE3 and those of the field data. Thereafter, 365 daily snapshots were taken from MIKE3 simulations, and subsequently 365 POD basis modes were computed. Due to high percentage of conserved energy of the lake system (salinity of lake) within the first ten POD basis modes, these modes were considered to develop a reduced-order salinity model (ROSM). Finally, results from MIKE3 were compared with the ROSM. It was shown that the first ten modes (among 365 modes) obtained by the POD conserved approximately more than 99.8% of the energy of the system. Moreover, using the first ten modes resulted in an error in magnitude of less than 0.01. Therefore, the ROSM could successfully capture the variations of salinity in the lake via its first ten modes.