CALVIN

General Description

CALVIN is a statewide hydro-economic model for the California's intertied water supply system. The model includes 82 years (1921-2003) of monthly surface and groundwater hydrology; major storage, pumping, and conveyance infrastructure, selected hydropower facilities and agricultural and urban service areas. The model accounts for infrastructure capacities, operating and water scarcity costs, as well as environmental regulations for minimum stream flows. As an optimization model, CALVIN allocates water monthly such that total water scarcity and systemwide operation cost is minimized. In addition to monthly water deliveries and storage, CALVIN provides economic opportunity costs of infrastructure expansions and environmental flows. The website of the model is:  https://calvin.ucdavis.edu

Software and network database:  https://github.com/ucd-cws/calvin

Model Domain

Water supply for California 

Developer

Lund, Howitt, Medellin-Azuara, Jenkins at the University of California, Davis

Hardware computing requirements

NA

Code language

Originally, CALVIN employed a free linear solver (HEC-PRM, Hydrologic Engineering Center-Prescriptive Reservoir Model), a network flow optimization code developed by the US Army Corps of Engineers' Hydrologic Engineering Center in Davis. The model is now coded in Python and solved as a network flow model with open-access free solvers including GLPK, CBC, CPLEX and Gurobi. See Dogan et al. (2018) for further details.

Original application

The State of California Resources Agency funded an 18-month study starting in January 1998 to analyze finance options for California's future water supply. An initial model report came in 2001.

Public/proprietary and cost

Free

Physically or empirically based

Empirical 

Mathematical methods used

CALVIN is a hydro-economic optimization model (generalized network flow optimization, described by Jensen and Barnes, 1980).

Input data requirements

• Surface and groundwater hydrology
• Facilities and capacities (storage and conveyance)
• Urban water use
• Agricultural water use
• Environmental flow constraints
• Operating costs
• Economic value of water from ancillary models (e.g. SWAP)

Outputs

• Water allocations and delivery reliabilities
• Willingness to pay for water and reliability
• Economic benefits
• Conjunctive use operations
• Value of flexible operations
• Values of increased capacities
• Monthly water allocation, typically over 82-year.

Pre-processing and post-processing tools

Originally, HEC-PRM was required to run the model. Inputs must be provided through database connected to HEC-PRM. In the Python-based platform and database https://github.com/ucd-cws/calvin the network and other model inputs can modified. Output files are in comma separated value format which can be post-processed with conventional spreadsheets or other data processing platforms. 

Representation of uncertainty

CALVIN employs deterministic optimization yet it incorporates hydrologic variability in its 82-year monthly hydrology which includes various multiyear droughts. The calibration process is described in Jenkins (2001) Appendix. 

Prevalence

Modest use in CA water planning  and demand analysis, yet several dozen products including refereed journal papers, theses, dissertations, and research reports have employed the CALVIN model. Related literature available on https://watershed.ucdavis.edu/shed/lund/CALVIN

Ease of use for public entities

Requires training and knowledge of network flow optimization, and Python.

Ease of obtaining information and availability of technical support

Extensive documentation on UC Davis Calvin website, contact information for PIs is readily available. Training sessions available on demand through the UC Davis Center for Watershed Sciences.  

Source code availability

Status of model development

The CALVIN model continues development on the interface to input data and other elements. Results from the model are useful for management and planning in droughts and adaptation to climate change.

Challenges for integration

Complex relationship between elements of the model. Existing integration with SWAP could provide a go-between for other models. Specificity of model schematic and inputs may be obstacle to integration with other water supply/demand models. The model interface would use some more development to properly connect with other frameworks.