Multi-period and multi-variable calibration of SWAT+ using gridded input datasets and a novel R package

For more than two decades, multi-objective optimisation (MOO) has imposed a new paradigm in the calibration of hydrological models, and over the years different algorithms and calibration approaches have been developed that aim to obtain consistent parameters for some specific hydrological model. However, the development of flexible multi-objective calibration tools has been scarce, making it difficult to spread these approaches to a wide range of researchers. 

The objective of this work is to show the application of a new multi-objective and multi-platform R package (hydroMOPSO) for the calibration of SWAT+, a widely used semi-distributed hydrological model. In particular, in this work hydroMOPSO is used beyond the traditional adoption of multiple objective functions. Instead, a multi-period (dry and wet years), and multi-variable (point streamflows and gridded soil moisture and evapotranspiration) are used as objectives, to illustrate the flexibility of hydroMOPSO to be linked with different model input and outputs, both with different file formats and temporal frequencies. Similar approaches could be applied with other hydrological models available in R (e.g., TUWmodel, airGR, topmodel) or any other model that can be run from the system console (e.g., Raven, MODFLOW, WEAP).