DOI: https://doi.org/10.1016/j.scs.2024.105676

Abstract: Conceptual urban water metabolism (CUWM) models provide a holistic view of the efficiency of urban water systems. These models can be linked with multi-agent reinforcement learning (MARL) models to mimic stakeholders' reactions to various strategies. However, the outcomes derived from CUWM-MARL models come with uncertainty. As a result, this paper introduces a decision support system (DSS) that is aware of these uncertainties and uses this information to select robust management strategies based on the output of CUWM-MARL models.

 First, future scenarios are generated by exploring all combinations of critical values of deeply uncertain variables and values of uncertain variables sampled from an appropriate multivariate copula distribution. Then, a computationally efficient surrogate model is developed to alleviate the computational load of the MARL sub-model. The surrogate model is run through all future scenarios to calculate the system's key performance indicators (KPIs) for each management strategy. Urban water managers can use these KPIs and social choice methods to find consensus management strategies. 

The proposed methodology has been implemented in the western part of the Tehran metropolitan area, Iran. This study evaluates 26 management strategies consisting of various projects, including reducing water distribution network leakage, implementing urban demand control projects, and utilizing semi-centralized or decentralized wastewater treatment plants. The strategy chosen by the uncertainty-aware DSS enhances the total utility and fairness indices by 125% and 4%, respectively. Moreover, it effectively improves groundwater quality, reduces energy consumption and greenhouse gas emissions, and enhances water supply reliability, ultimately contributing to farmers' job security.

Keywords: Decision Support System; Deep uncertainty; Reinforcement learning; Water-energy-food-GHG nexus; Surrogate model; Urban metabolism.

Link

Published paper title: Combining urban metabolism and reinforcement learning concepts for sustainable water resources management: A nexus approach

DOI: https://doi.org/10.1016/j.jenvman.2022.117046

Abstract: Modeling Water-Energy-Food (WEF) nexus is necessary for integrated water resources management (IWRM), especially in urban areas. This paper presents a new urban water metabolism-based methodology for WEF nexus modeling and management. A behavioral simulation model is used to incorporate the characteristics of stakeholders in an urban area. Modified versions of the Borda count, Copeland rule, and fallback bargaining procedures are implemented to choose the socially acceptable management scenarios. Finally, the selected scenarios’ effectiveness is evaluated using the fairness and total utility indices.

The applicability of the proposed methodology is evaluated by applying it to the Kan River basin, Tehran, Iran, which is suffering from some water and environmental issues. The considered management scenarios include adding new water sources, leakage control plans, using rubber dams for enhancing groundwater recharge, revising water allocation priorities, and developing semi-centralized or decentralized reuse strategies for reclaimed wastewater. Results illustrate that considering different fluxes (i.e., water quantity, pollutants, energy, greenhouse gases (GHG), and materials) is as important as incorporating the social characteristics of stakeholders. Simulating the socially acceptable scenario shows that the aquifer's average water level improves by 3 (m), and its average nitrate concentration reduces by 16 (mg/L) in comparison with the business as usual (BAU) scenario. In addition, by implementing different water reuse strategies, which are energy-intensive, total energy consumption is reduced by 5% due to less groundwater pumping. Also, the selected scenario decreases GHG emissions by 18% and increases the sequestrated carbon dioxide by 20%. In conclusion, the proposed decision support tool can provide policies for sustainable water resources management considering water quality and quantity issues, energy usage, and GHG emissions.

Keywords: Water-energy-food-GHG nexus; Social choice procedures; Wastewater reuse; WaterMet2; Tehran city; conceptual-behavioral simulation;

DOI: https://doi.org/10.1016/j.jenvman.2023.118767

Abstract:  Market-based approaches are increasingly considered reallocating instruments that put water consumption at its highest economic value among competing water users. Setting up a water market can have a lot of environmental, social, economic, and legal complexities. One of the main issues is the uncertain nature of the available water, which can cause the failure of markets, especially during drought conditions. Therefore, there is a need for market mechanisms to consider and reduce the adverse impacts of available water uncertainty on market outcomes. Accordingly, this paper proposes a new real-time seasonal smart water market framework for basin-wide surface water pricing and allocation. The framework uses the results of the reservoir water allocation optimization models and ANFIS-based monthly river discharge forecasts to better assist the water users with their bidding. The market manager uses updated available information at the beginning of each season to provide users with a more accurate understanding of available water to adjust their tradings for the rest of the year. The applicability and efficiency of the proposed framework are evaluated by applying it to the Gorganrood River basin in Iran. According to the results, the framework increased users' benefits from 721 to 1050 billion rials, which is more efficient than an annual market. Water markets can use this framework to improve their ability to cope with the uncertainty of available water, increase their users' benefits, and encourage them to improve their efficiency. Furthermore, the proposed framework allows the decision-makers in water sectors (e.g., industrial, agricultural, etc.) to discover time and location-specific water allocation and price for different water users.

Keywords: Water market; Smart market; Real-time Water reallocation; Water pricing; ANFIS model

Abstract:  The integrated management of water resources (particularly in urban areas) necessitates considering the water-food-energy nexus during simulation. According to the simulation of urban water metabolism and given the water-food-energy nexus, the current thesis has developed two decision-making support tools in two states, i.e. considering or not considering the effect of uncertainties in decision-making. In deterministic decision-making support tools, the urban water metabolism simulator is connected to a behavioral simulator to predict the stakeholders' response to different strategies. Different strategies were examined by using this hybrid model. Ultimately, improved social choice methods have been employed for selecting strategies agreed upon by superior stakeholders. 

The uncertain decision-making support tool helped decision-makers make better decisions for difficult-to-predict-and-simulate systems. Robust decision-making (RDM) is the framework employed to develop the uncertain decision support tool; besides considering the quantity and quality of water, also considered in the previous decision-making support tools, these two decision-making support tools investigate energy usage and greenhouse gas absorbed and emitted, too.

In order to measure the developed tools’ efficiency, the Kan River basin in Tehran (Iran) has been selected for evaluation, which suffers from many environmental problems. Twenty-six scenarios have been developed to improve conditions in this area, including a combination of projects like adding a new water source, reducing water distribution network leakage, managing urban potable water demands, treating wastewater through semi-centralized and decentralized treatment plants, and reallocating wastewater.

Both deterministic and uncertainty-based decision-support tools have resulted in the selection of the same strategy. By applying the final strategy, the investigated system will become more sustainable regarding water quantity and quality, energy use, and greenhouse gas emissions. Moreover, this scenario has provided fairer and more beneficial conditions for stakeholders.

These two decision-support tools’ effectiveness has been measured by two general indicators: fairness and stakeholders’ total utility. In the business-as-usual condition, the fairness and stakeholders’ total utility indicators were equal to 2.2 and 0.91, respectively. By selecting the proper strategy, the deterministic decision support tool reveals that fairness and the stakeholders’ total utility indicators reach the values of 4.83 and 0.945, respectively; and the uncertainty-based decision support tool indicates that fairness and the stakeholders’ total utility indicators increase to the values of 4.95 and 0.95, respectively.

Keywords: Water-Energy-Food-GHG Nexus,  conceptual-behavioral simulation, Social Choice Methods, Uncertainty Analysis, Robust Decision Making (RDM), Surrogate Modelling by GRNNs

Background:

Gorgan in one of the northern districts of Iran. And Golestan is the capital of it. The main water demands of the area are related to agricultural, fishery, and industrial users.  Following a prolonged and severe drought spanning multiple years, a series of flood occurrences of unparalleled geographical magnitude occurred in various regions of Iran from March 17th to April 1st, 2019 (Alborzi et al, 2022). Northern Iran was hit by frontal precipitation events (March 17–22, 2019) that caused the first major flood. GorganRood and GhareSoo Basin in Golestan Province received 50% of the local mean annual rainfall in three days, including 338 mm at TooskaChal Station, breaking the one-day and three-day records in over 70 years (Beitollahi et al., 2019).

Research Objectives, method, and results:

This project has four research questions, in the following, each of them is stated and the method to answer the question is discussed.

1) The primary objective of this project is to generate a flood extent map to estimate the areas affected by flooding in Gorgan, utilizing the Google Earth Engine (GEE) platform. The flood extent is determined through the application of a change detection methodology on Sentinel-1 SAR GRD (C-band Synthetic Aperture Radar Ground Range Detected) data developed and validated by UN-SPIDER.

2) After identifying the impacted regions, it is feasible to examine the demographic composition of these locations to ascertain the extent of the population affected. The GHS POP data collection will be utilized for this objective.

3) The utilization of NDVI (by using sentinel data) can be employed to compare long-term (and before and after the flood) surface characteristics alterations within the area, to detect some of the contributing factors to this flood.

4) The night light is utilized to ascertain the potential detection of alterations in urban dynamics before and after the occurrence of a flood event. Additionally, the duration required for the restoration of night light following its diminishment can serve as an indicator of the area’s recovery rate. The VIRS dataset was used for this purpose.

The repository contains code and presentation

This project analyzes 140 years of central park precipitation using methods like SARIMAX, Marked Point Processes, and LSTM networks. Gaussian processes include frequency analysis, predictive modeling, and uncertainty quantification to explore precipitation trends and improve forecasting accuracy.

Key Features:

Frequency Domain Analysis to assess periodicity.

SARIMAX Models for trend and seasonal predictions.

Marked Point Process (MPP) for impulse-like behavior analysis.

LSTM Neural Networks for temporal correlation modeling.

Gaussian Process for uncertainty quantification of residuals.