Optimal Management of
Groundwater-Surface Water
Resources:
A Brief Overview
James R. Craig
Assistant Professor
Dept. of Civil & Environmental Engineering
Outline
• Brief overview of groundwater and surface water
interactions
• Optimization problem
– Common constraints /objectives
• Discussion
– Current practice
– Important considerations
• A global search algorithm, DDS, presented by Prof.
Tolson
Premise
• Groundwater and surface water resources are
intimately connected
• Management of either resource requires knowledge
of the impact of management decisions on both
• Due to the complexity of these systems, predictive
models should be used to facilitate decision making
– Management problem posed as a “simulationoptimization” exercise
Water Management
• Water managers are tasked with determining how
best to obtain and allocate our water resources –
who gets water, how much they get, and where they
can get it from.
• In the case of groundwater allocation, the selection
of well locations and pumping schedules can
– impact the quantity and distribution of water present in
streams, wetlands, or aquifers and
– determine the quality of both the pumped water and
affected areas
Conceptual Models
Groundwater and surface water
exchanges occur in both
directions
-Behavior is generally transient
and can rarely be predicted in a
purely deterministic manner
Gaining Stream
Losing Streams
Stream Depletion
Recharge/Infiltration
Stream Depletion:
“Safe Yield”:
Pumping balanced by
recharge
Lowers water levels,
reduces base flow, effects
wetland ecosystems
Drying out of wellTypical
dominated by 10-100s of wells, extensive stream networks and
Water systems
rights infringement
complicated exchange patterns influenced (in part) by transient precipitation,
treated either deterministically or stochastically
“Typical” GW-SW optimization
problem
• Maximize groundwater withdrawal with minimal
impact to surface water resources
• By changing:
– Pumping rates, schedules, & locations
– Surface irrigation and storage measures
• Subject to multiple constraints:
– Groundwater quantity & quality
– Surface water quantity & quality
– Cost
Quantity Constraints
• Groundwater quantity / distribution
– Sustainable pumping rates (or close enough)
– No water rights infringements (penalty function)
• Surface water quantity / distribution
– Water levels must typically be maintained high enough to sustain fish
and bird habitats, recreation
– Flow rates have to be within desirable limits for
• Hydropower
• Dilution of agricultural & industrial waste
• Sediment transport
– All impacts propagate downstream – watershed-scale management is
common
Quality Constraints
• May wish to minimize or disallow amount of
surface water allowed to reach pumping wells
– Reduces/removes presence of surface water
contaminants
• Seawater intrusion
From www.lenntech.com
Current State-of-Practice
• “Manual optimization” still quite common
– Black-box management tools (rather than physics-based
models) used to test “what-if ” scenarios
• State-of-the science
– **Oversimplified Systems**
– The standard heuristic toolbox
• GA, PSO, SA, Integer programming, etc., etc.
– Limited by the ability to solve real-world problems in a
reasonable time frame
Important Issues
• The subsurface is unknown!
– How can the optimization process explicitly address the presence of
uncertainty?
• The systems are often large (watershed scale) and poorly characterized
– Computationally expensive – How to develop surrogate models?
• “Perfect” Global optimum is not the real goal
– How to replace the chase for perfection with the chase for “good enough”?
• Appropriate formulation of the objective function is an issue
– Philosophical question: How to quantify ecological damage?
– What are the impacts of changing the objective function?
• Long-term research objective:
– How to develop multi-objective tradeoff curves (e.g., cost vs. extraction vs.
environmental quality) under the presence of fundamental uncertainty at
watershed scales?