Irrigation nudging in COSMO-REA6

Irrigation, predominantly carried out in southern Europe, has a considerable impact on the state of the atmosphere, but is not yet included in numerical models used for producing reanalyses. A common approach to correct the soil moisture (and, thus, also lower boundary condition in the atmospheric model) is to alter the soil moisture field to fit 2m temperature observations in a data assimilation scheme. In this way, water is added to or removed from the soil in the form of assimilation increments. This raises the question whether this assimilation process also recognizes soil moisture changes from irrigation and adjusts the model state accordingly.

To answer this question, we analyze the effects in the Po valley in northern Italy – a region where irrigated fields are the predominant land use. Here we compare three years in the late irrigation period July and August with each other, one particular rainy year, one drought year with irrigation and one drought year when irrigation was prohibited by law.
For this purpose we conduct a spatial analysis of the soil moisture increments in the neighbourhood of assimilated stations with respect to irrigation and precipitation bias. The resulting findings are then interpreted in terms of their spatial relation to other stations and their values of irrigation and precipitation bias.

The results indicate that the assimilation process is able to compensate for the deviations caused by irrigation. However, the effictiveness of this process also depends on the spatio-temporal density of observations. Since the vast majority of observing sites are located in non-irrigated areas, considerable model deviations in soil moisture and corresponding atmospheric parameters are still expected to be present in irrigated areas.