For irrigated agriculture, land units have been previously revised to be consistent with major river basins and thus, irrigated water use draws water directly from the basin in which the crop is grown. No additional mapping of irrigated water demand to basins is necessary to capture the agricultural use of water by basins.
All other uses of water, electric power, industries, municipal, primary energy, and livestock are modeled at the country or GCAM regional scale. Representations of these end-use sectors and their water demands are not spatially disaggregated sufficiently to be directly consistent with water supply at more detailed basin-scale. This mismatch is more pronounced for larger GCAM regions with multiple basins contained within its boundary than smaller regions with few basins. To address this mismatch, a mapping structure of regional water demands by sector to basin supply is constructed based on gridded historical datasets of population and power plant locations in the case of electric power. The mapping structure represents the basin shares of total water withdrawal (or consumption) from each regionally aggregated end-use water demand categories. The mapping resolves water supplies and demands at the basin scale and allows the balancing of water supplies and demands to future changes in the basin supply and regional end-use demand for water.
Additionally, the mappings structure allows multiple GCAM regions to access water resources from overlapping and shared basins. Basins have independent geographic boundaries and are separate to the GCAM regional boundaries. Basins may be contained within a single or multiple GCAM regions. The mapping structure for non-irrigated water demands quantifies what portion of regional demands came from which basins. The entirety of the basin is available for use by any GCAM region or regions that are mapped to it. For instance, the US and Canada share a common basin, such as the Great Lakes, and both countries have access to the total available supply of water in that basin. In practice, basins supplies are typically shared by overlapping regions. Irrigated agriculture water demands from multiple GCAM regions may also draw water from common basins. Upstream versus downstream issues within a basin, whether shared by mutliple regions or not, are not modeled. There are no biases or limitations placed on the regional use of shared basin water resources.
Currently, mappings of water demands to basins are assumed to remain fixed over time. This implies no inter-basin transfer and also does not capture long-term shifts in regional demands for water (excluding irrigated water demand), either due to population migration and/or spatial shifts in industrial and energy production within the GCAM region. These distributional issues are likely to be of greater concern for large regions that contain multiple basins as mentioned above. Future work in which sub-regional representations of water demands are modeled (e.g. GCAM-USA, GCAM-China, GCAM-India) in conjunction with remapping of finer scale demands to basin supplies can provide more consistent views of long-term responses to local and regional water use. We remain cognizant of the difficulties in the consistent representation of spatial and temporal scales in modeling the supply and demand of water resources.
Water prices at the basin-scale are “cleared” with all other market prices or simultaneously solved. Supplies of water increase with higher water prices subject maximum limits, whereas demands for water fall. Substitution to alternative technologies or crops with lower water intensity, shifts in the trade of agriculture commodities, or direct reductions in the withdrawal or consumption of water are the demand responses to higher water prices. The rising cost of water may limit or alter the production of goods and services until water supplies and demands are rebalanced through the market mechanism. Market clearing water prices are determined for each water basin. The water balancing and market behavior currently utilizes demands based on withdrawal volumes. Market behavior based on consumptive volumes is possible; however, we find that the use of consumptive volumes, which is low relative to total available water volumes, is grossly unresponsive to water scarcity. Both the irrigated water subsidies and the choice of withdrawal or consumptive volume for water pricing can have a substantial impact on the behavior of water constraints in GCAM. Further exploration of the demand responses to withdrawal and consumption differences remain for future work.
When renewable water supplies are abundant relative to demand, water prices remain low (or negligible), and non-renewable groundwater and desalinated water do not contribute to supply. In this case, water supplies are not binding and do not constrain the production of goods and services. When water demands approach or exceed accessible renewable supplies, water prices rise more rapidly such that more expensive non-renewable groundwater and desalinated water contribute to supply. Generally, renewable water supplies are lower in cost and are utilized first and to a greater extent before depending on more expensive non-renewable groundwater or desalinated water. Over time, basins may produce from all three sources of water depending on the market price of water and the cost of alternative water resources in each basin.
In order to improve the solution behavior of all GCAM markets with interlinked water dependencies, and to limit the total number of new markets to solve, renewable and non-renewable water resources have been combined into one effective resource supply curve. This limits the number of new markets to solve to 235, the number of water basins included. Renewable and non-renewable water resources are separately tracked, however. In GCAM, two Sub-Resource objects are included within the Resource object for each water basin, one each to represent renewable and non-renewable water resources. The Sub-Resource supply curves are additive and together represent the total supply of water.
Adding renewable and non-renewable resource curves into one is conceptually a new feature to GCAM. Discontinuity in the aggregate supply curve due to changing renewable potential and/or depletion of nonrenewable water resources has been accounted for by ensuring that the two Sub-Resource curves are overlapping so as to ensure continuity of the aggregate supply curve. The ability to combine and overlay multiple supply curves into one provides great flexibility in the potential for alternative characterization of water resources and their contributions to the overall supply. For instance, both renewable and non-renewable resources can be further disaggregated, such as by separating renewable water into surface and ground water (recharge) portions or separating fossil groundwater by alternative aquifer characteristics. The shape of the multiple resource curves and the degree to which they overlap or not can determine water use behavior, such as the simultaneous or sequential production of alternative water resources.
One notable feature of the GCAM’s Renewable Sub-Resource object is the ability to vary the maximum available resource potential by time period. This feature enables the assessment of climate impacts on water availability. Projected future changes in precipitation from climate models (GCMs) are utilized to calculate river runoff and changes in the maximum available potential of renewable water supplies by time and basin.