Executive summary:
The present deliverable D3.2, developed within the Task 3.2 building upon D3.1 and D2.1, aims to define a timeline of the steps towards assembling, calibrating and validating the actionable modeling suite for each living lab. To this end, D3.2 provides a conceptual description of the key principles guiding the modeling suite, namely uncertainty quantification (Section 2) and the protocol-based modular approach (Section 3), as well as the key methods available in the literature. Next, Section 4 describes in detail how these two principles are to be addressed in each of the seven living labs, including a description of the specific modules, models, protocols and uncertainty quantification methods, as well as an implementation timeline via a Gantt Chart.
Regarding the modules adopted, all labs will model endogenously (i.e., with an explicit model that is part of the modeling framework) the hydrologic and economic systems. All labs will model the climate system, albeit only one endogenously (the Reno, using ad-hoc statistical and dynamical downscaling) and the remaining exogenously by incorporating existing outputs from modeling and downscaling experiments such as CMIP6 and Cordex. 6 out of 7 labs will model the agricultural system endogenously (Orontes, Nitra, Júcar, Reno, Titicaca-Desaguadero-Poopo, Mahanadi), and one exogenously (Tympaki). One lab (Reno) will model the meteorological system, endogenously.
Selected models for the hydrological module include SWAT+ (Orontes, Mahanadi), HYPE (Nitra, Reno), Ribasim (Reno, as water balance model linked to HYPE), WEAP (Tympaki, Titicaca-Desaguadero-Poopo), and TETIS (Júcar); for the economic module, PMP and PMAUP (Orontes, Nitra, Reno, Titicaca-Desaguadero-Poopo, Tympaki, Mahandi) and linear models (Júcar, Mahanadi); for the agronomic module, GYMEE (Orontes), DAISY, WOFOST (Nitra), Aquacitrus (Júcar), CRITERIA (Reno), AquaCrop (Titicaca-Desaguadero-Poopo), SWAT+ tool (Mahanadi) and CMIP6 and AgMIP databases (Tympaki); for the climatic module, CMIP6 (Orontes, Júcar, Nitra, Titicaca-Desaguadero-Poopo, Mahanadi and Tympaki) and ad-hoc statistical and dynamical downscaling experiments (Reno); and for the meteorological module, the ARPAE-CO model (see D2.1 for a detailed description of these models).
Regarding the protocols and coupling approach adopted, the Orontes, Reno, Titicaca-Desaguadero-Poopo and Tympaki labs have opted for a combination of bidirectional and sequential coupling protocols, where bidirectional protocols are adopted for the coupling between the hydrological and economic module, and sequential protocols adopted for the couplings climate→hydrological, agricultural→economic, agricultural→hydrological, and meteorological→agricultural (i.e., climate and agronomic modules provide inputs to the economic and hydrological modules, which are then run iteratively). The Júcar lab has adopted a fully sequential approach, while for the Nitra the conceptual design of the coupling has been delayed until the availability of data for the calibration and setup of the proposed models is confirmed.
Regarding uncertainty quantification, all labs quantify (at least to some extent) input, parameter, and structural uncertainties (except for the Nitra for which the definition of uncertainty quantification methods has been delayed until the availability of data for the calibration and setup of the proposed models is confirmed). Input uncertainty is quantified using scenario-based approaches for all labs (excluding Nitra) and most systems, typically relying on outcomes from selected RCP and SSP scenarios; while global sensitivity analysis methods are proposed for the economic module in the Orontes, Reno, Titicaca-Desaguadero-Poopo, Mahanadi and Tympaki; and for the hydrologic module in the Orontes and Mahanadi leveraging on SWAT+ tools. Parameter uncertainty is quantified using local sensitivity analysis for the Júcar (economic and/or agricultural modules), ans using global sensitivity analysis in the Orontes (economic), Júcar (economic and/or agriculture), Reno (hydrologic, economic), Titicaca-Desaguadero-Poopo (hydrologic, economic), Mahanadi (hydrologic, economic), and Tympaki (hydrologic, economic). Finally, structural uncertainty is quantified using multi-model ensemble experiments in the Orontes (economic and climate modules, as well as water use data inputs), Nitra (climate), Júcar (slimate), Reno (climate, economic), Titicaca-Desaguadero-Poopo (climate, economic), Mahanadi (climate, economic), and Tympaki (climate, economic).