System dynamics modeling of lake water management under climate change

Abstract

Lake Urmia, the twentieth largest lake in the world, is the most valuable aquatic ecosystem in Iran. The lake water level has decreased in recent years due to human activities and climate change. Several studies have highlighted the significant roles of climatic and anthropogenic factors on the shrinkage of the lake. Management policies for water resources harvesting must be adopted to adapt to climate change and avoid the consequent problems stemming from the drought affecting Lake Urmia, and rationing must be applied to the upstream water demands. This study analyzes strategies and evaluates their effectiveness in overcoming the Urmia Lake crisis. Specifically, system dynamics analysis was performed for simulating the water volume of Lake Urmia, and the Hadley Centre coupled model was applied to project surface temperature and precipitation for two future periods: 2021–2050 and 2051–2080. Six management scenarios were considered for decreasing the allocation of agricultural water demand corresponding to two options: (1) one-reservoir option (Bukan reservoir only), and (2) six-reservoir option. The net inflow of Urmia Lake was simulated for the two future periods with the IHACRES model and with artificial neural network models under the six management scenarios. The annual average volumes of Lake Urmia would be 30 × 109 and 12 × 109 m3 over the first and second future periods, respectively, without considering the management scenarios. The lake volumes would rise by about 50% and 75% for the first and second periods, respectively under the management scenarios that involve strict protective measures and elimination of the effect of all dams and their reservoirs. Implementing strict measures would increase the annual average lake volume to 21 × 109 m3 in the second period; yet, this volume would be less than the long-term average and strategic volume. The human water use would be completely eliminated under Scenario 6. Nevertheless, Lake Urmia would experience a considerable loss of storage because of drought.

A methodology for parameter estimation in system dynamics models using artificial intelligence

“…Each of these models have been previously found to be useful for performing nonlinear regression of empirical data. [4][5][6][7] Their difference in structure and learning will provide insights into the efficiency of the selected methodology.…”

“…For the high dimensional space, the kernel functions are used. Two kernel functions used to perform nonlinear mapping are polynomial, shown in Equation (5), and the radial basis-function (RBF), shown in Equation (6).…”

“…The authors simulated the system and constructed another ANN model that estimated the change in salinity based on the simulation variables. Bozorg-Haddad et al 5 used ANN to simulate lake and reservoir inflows and outflows in a water management system. Carranza et al 6 compared dynamical modeling and random forests (RFs) for interpolating and extrapolating root zone soil moisture measures.…”

Novel Air2water Model Variant for Lake Surface Temperature Modeling With Detailed Analysis of Calibration Methods

“…HE problem of modelling the lake surface water temperature based on the air temperature has been widely discussed in the literature [1]- [3]. As water temperature in lakes affects the presence of such important gases as oxygen [4], [5] and impacts on the fate of aquatic biota [6], [7], the problem of water temperature modelling is important for both biodiversity [8], [9] and lake water management [10], [11]. Thermal modelling is becoming even more critical during the climatic change [12], [13].…”

Dynamic monitoring and evolution analysis of wetland in Guangdong-Hong Kong-Macao Greater Bay area from 2005 to 2020

“…Investigating spatiotemporal wetland dynamics has generated interest in future pattern simulations. Quantity-based models, like system dynamics models [9], predict wetland area and change rate, but lack spatial analysis. In contrast, spatial models, such as cellular automaton [10], CLUE-S [11], and multi-agent models [12], offer valuable spatial evolution simulations and geographical process visualization.…”