Integrating MODIS NDVI in the WRF model to improve simulations in the Iberian Peninsula: a case study in a Mediterranean grassland of central Spain.

Vegetation has a strong influence on the interaction between the land surface and the overlying atmosphere, modifying the relative contribution of sensible and latent heat to the surface energy balance (SEB). Vegetation dynamics must be considered to improve atmospheric surface models, especially in the understanding of the land-atmosphere coupling.

Nowadays, remote sensing is an excellent tool to monitor vegetation functioning at different spatial and temporal scales. Spectral indices obtained from remote sensing data, such as the Normalized Difference Vegetation Index (NDVI) are strongly related to photosynthesis and vegetation functioning. Thus, the role of vegetation in the energy partitioning processes could be assessed using these spectral indices.

The Weather Research and Forecast Model (WRF) uses by default monthly values of the Green Vegetation Fraction (GVF) dataset derived from the NDVI of the NOAA Advanced Very High Resolution Radiometer (AVHRR) at 0.144° resolution for the period 1985-1990 to generate its simulations. The overall aim of this work is to integrate monthly MODIS NDVI values at 1 km resolution (MOD13A3 v061 product) in the WRF model to improve simulations. Validation of the simulations will be carried out using an Eddy Covariance flux tower with data from 2017 to the present, which belongs to the GuMNet network (https://www.ucm.es/gumnet), placed in a grassland ecosystem of Central Spain close to a forest influenced by the Guadarrama mountains (El Escorial, Madrid).

Results show that including NDVI in the WRF model improves the simulations in the grassland ecosystem, especially in the estimation of turbulent heat fluxes. Preliminary results in this grassland environment showed that high-resolution NDVI was strongly correlated with latent (a) and sensible (b) heat, especially for three months (May-June-July). These three months coincided when the grassland showed its greatest p changes, presenting its maxima of biomass and then drying quickly during the following months due to the lack of soil water availability.

In conclusion, this work shows that it is essential to include the vegetation dynamics to improve the atmospheric and land surface models, being a first step to use spectral indices as a proxy to improve the models.