Identification and Analysis of Critical Water Futures in the Indus River Basin

Pakistan is a water-based economy and suffers from severe water scarcity in its primary river system, the Indus River Basin (IRB). The assessment of interactions among rising agricultural demand, socio-economic development and climate change is crucial to assess water scarcity in the IRB. Given the multiplicity of risks and the physical and social mechanisms that interact with them, estimating the future usage of the IRB requires models that represent plausible futures defined by a broad range of factors.

The Global Change Analysis Model (GCAM), an Integrated Assessment Model (IAM), is used to assess the complex connection and interactions between energy, water, land, climate, and the economy. GCAM divides the globe into  235 water basins, including the IRB, and 384 land use regions which are modelled based on combinations of 32 energy regions and overlapping water basins. Dolan et al. (2021) used GCAM to generate a large ensemble of 3,000 plausible future scenarios, varying parameters related to future socioeconomic conditions, climate impacts, and water supply. Each scenario represents a possible future from now until the end of the century, with detailed socio-economic, water supply and demand and land-use results at the basin level. Yet, while these experiments generate large databases, there is a need for specialised methods that extract useful information from that data.

Using the example of the IRB, we develop a methodology to leverage this type of database and (1) discover critical scenarios, i.e., scenarios with an outsized impact on water scarcity and economic costs, and (2) learn more about their characteristics, including what makes them critical. Here, we seek to identify outlier patterns by proposing a methodology that combines a machine learning technique, clustering, with dimensionality reduction. With clustering, we aim to identify hidden structures among scenarios and describe the clusters by a set of factors. Dimensionality reduction then assists us in determining which factors have the greatest impact on the critical scenarios that clustering identified.

Preliminary results suggest that our methodology is able to identify outlier scenarios for the IRB’s irrigated crops mix (dominated by cotton, wheat, rice, and sugarcane), understand the factors that make them outliers, and evaluate whether they could be critical. The analysis is also able to identify when an ensemble of scenarios becomes an outlier, and indicates that according to GCAM, the crop mix is susceptible to bifurcating in several contrasting directions after 2040. Thus, this methodology helps us to characterise the socio-economic uncertainties associated with the IRB’s water resources and their interaction with climate, land, food, and energy sectors under critical scenarios. It is being developed to have broad applicability in extracting valuable insights from a large ensemble of IAM simulations.

 

Dolan, F., Lamontagne, J., Link, R., Hejazi, M., Reed, P. & Edmonds, J. 2021. Evaluating the economic impact of water scarcity in a changing world. Nat Commun, 12, 1915.