Empirical validation of the InVEST water yield ecosystem service model at a national scale
2016
A variety of tools have emerged with the goal of mapping the current delivery of ecosystem services and quantifying the impact of environmental changes. An important and often overlooked question is how accurate the outputs of these models are in relation to empirical observations. In this paper we validate a hydrological ecosystem service model (InVEST Water Yield Model) using widely available data. We modelled annual water yield in 22 UK catchments with widely varying land cover, population and geology, and compared model outputs with gauged river flow data from the UK National River Flow Archive. Values for input parameters were selected from existing literature to reflect conditions in the UK and were subjected to sensitivity analyses. We also compared model performance between precipitation and potential evapotranspiration data sourced from global- and UK-scale datasets. We then tested the transferability of the results within the UK by additional validation in a further 20 catchments. Whilst the model performed only moderately with global-scale data (linear regression of modelled total water yield against empirical data; slope = 0.763, intercept = 54.45, R2 = 0.963) with wide variation in performance between catchments, the model performed much better when using UK-scale input data, with closer fit to the observed data (slope = 1.07, intercept = 3.07, R2 = 0.990). With UK data the majority of catchments showed less than 10% difference between measured and modelled water yield but there was a minor but consistent overestimate per hectare (86 m3/ha/year). Additional validation on a further 20 UK catchments was similarly robust, indicating that these results are transferable within the UK. These results suggest that relatively simple models can give accurate measures of ecosystem services. However, the choice of input data is critical and there is a need for further validation in other parts of the world.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
52
References
111
Citations
NaN
KQI