Research papersModel swapping: A comparative performance signature for the prediction of flow duration curves in ungauged basins

- A framework for the assessment of flow duration curves using distance-based method.
- Regression modeling is fringed with nearest neighbor analysis to do the estimation.
- The model with better global performance is selected to make initial estimation.
- The estimates are improved by swapping the selected model with another model.
- Two examples are reported to demonstrate the effectiveness of model swapping.

The issue of lack of donor basins for prediction of flow duration curves (FDCs) in ungauged basins (PUB) is an important area of research that is not resolved in the literature. We present a distance based approach to predict FDCs at ungauged basins by quantifying the dissimilarity between FDCs and characteristics data of basins. This enables us to bracket hydrologically similar basins and thus allowing us to estimate FDCs at ungauged basins. Generally, a single regression model is selected to make hydrological estimates at an ungauged basin. Based on established laws and theories of hydrology, we work to devise a method to improve the output of selected model for an ungauged basin by swapping it with another model in case the latter gives better coverage and statistical estimates of the nearest neighbors of an ungauged basin. We report two examples to demonstrate the effectiveness of model swapping. Out of 124 basins used in analysis, 34 basins in example 1 and 41 basins in example 2 fulfill the set criteria of model swapping and subsequently their estimates are improved significantly.