Hydroeconomic modeling to support integrated water resources management in China

The North China Plain is a 320,000 km2 alluvial plain with a population of more than 200 million people, that stretches across the Hebei, Henan, Beijing, Tianjin and Shandong provinces. The plain is an economic hotspot and is extensively used for irrigation agriculture and heavy industries. Population growth and rapid development of the Chinese economy have increased water scarcity and put the natural water resources and aquatic ecosystems on the North China Plain under pressure.

Dry rivers, rapidly decreasing groundwater tables and heavily polluted surface water bodies are consequences of the growing demand for water to irrigation, industrial and domestic uses. As a response, the Chinese authorities have launched the 2011 No. 1 Central Policy Document, which set targets related to water scarcity and water quality and marks the first step towards sustainable management of the Chinese water resources.

In this context, the PhD study focused on development of approaches to inform integrated water resources management to cope with multiple and coupled challenges faced in China. The proposed method is to formulate river water management as a joint hydroeconomic optimization problem that minimizes basin-wide costs of water supply and water curtailment.

Water users are characterized by water demand and economic value, turning the complex water management problem into a single objective cost minimization problem. The physical system and management scenarios are represented as constraints to the optimization problem, which was solved using a variant of stochastic dynamic programming (SDP) known as the water value method.

This method determines and stores shadow prices of water for all system states. Three different method implementations to the Ziya River, a complex Chinese river basin on the North China Plain, were used to assess water conflicts and interactions between water users and ecosystems. To overcome the curse of dimensionality associated with SDP, the multiple surface water reservoirs were aggregated to a single reservoir.

Natural runoff upstream the reservoirs was estimated with a simple rainfall-runoff model based on the Budyko framework and auto-calibrated with measured discharge. The monthly serial correlation was described by a Markov chain and the estimated runoff used as stochastic input. The first method used a simple linear and convex formalization of the management problem with a single surface water reservoir state variable.

A comparison of different management scenarios was used to evaluate how the South-to-North Water Transfer Project will impact optimal water resources management. Scenarios with unregulated groundwater pumping at realistic pumping costs verified that the water users will keep pumping until all water demands are fulfilled.

The second implementation introduced a groundwater aquifer state variable, and linked groundwater drawdown to pumping costs. Non-convexity and non-linearity caused by these head-dependent pumping costs were accommodated with a hybrid genetic algorithm and linear programming formulation. Besides regional drawdown, local drawdown cones estimated with the steady state Thiem solution were included.

This enabled analysis of dynamic groundwater and surface water interactions. The results showed that the groundwater aquifer buffered the system and allowed overdraft in dry years in return for increased recharge in wet years. Further, cost-effective recovery of an overdrafted groundwater aquifer was demonstrated.

The third implementation assessed interactions of water resources and water quality management. Biochemical oxygen demand (BOD) was used to represent water quality, and water uses were associated with BOD generation and BOD reduction treatment costs. Constraints on downstream river water quality were included as dissolved oxygen, computed with the Streeter-Phelps equation.

Nonlinear constraints were overcome with the hybrid genetic algorithm and linear programming formulation. Costs of compliance with the different Chinese water quality standards were found to be relatively small compared to the water scarcity costs found in the second study. In contrast, the optimal management was highly affected, and allocations were shifted between the users as the model utilized the surface water for dilution.

The developed methods were successfully used to demonstrate how hydroeconomic modeling can guide optimal water management for complex systems. The method is highly flexible and can be applied to systems which can be formalized as up to two reservoir state variables. Linearity and convexity reduce the computation time, but are not required to solve the problem.

Cost-effective management can be found across traditionally separate disciplines, and this method thereby represents the type of integrated assessments needed in the context of the China 2011 No. 1 Central Policy Document.