Quantifying the Surface Flux Variability Associated with Vegetation Variations Over the SGP site during the HI-SCALE Field Campaign

The Southern Great Plains (SGP) of North America is known as a region of strong land-atmosphere interactions in the warm season. Recent studies suggested that vegetation plays a significant role in the land-atmosphere coupling in the region, but plant biogeophysical properties vary across a wide range of spatiotemporal scales, making it challenging to quantify the influence of surface on the boundary layer and cloud processes. The 2016 Holistic Interactions of Shallow Clouds, Aerosols, and Land-Ecosystems (HI-SCALE) field campaign provides a suite of observations to address this challenge. This study uses measurements from the Department of Energy’s Atmospheric Measurement (ARM) SGP site to quantify the spatiotemporal variability in the surface fluxes, and contrast them between the two intensive observation periods (IOPs) conducted in the spring and summer. The observed statistics are then used to evaluate Weather Research and Forecasting (WRF) model forecasts at a convection-permitting resolution (1.3 km grid spacing). The three-dimensional ensemble-variational hybrid assimilation scheme is used to constrain the initial conditions in WRF. The model experiments, integrated with observations, help identify systematic linkages between turbulent statistics and heterogeneities in the surface and mixing layers.

During HI-SCALE, the mean surface sensible (SH) and latent heat (LH) fluxes change by 20 to 30 W m^-2 from IOP1 to IOP2. The grassland and cropland sites show opposite seasonal trends in each flux, such that Bowen Ratio (BR) decreases over the grassland but increases in the cropland. The (joint) probability distributions of the daytime surface fluxes generally follow the (bivariate) normal distribution, but they are positively skewed with high probability densities toward near-zero values. The preliminary site-level evaluation of a 12-h forecast, which assimilates only conventional atmospheric data for the initial conditions, shows that the model reliably forecasts temperature and moisture in the near-surface air and soil. However, in the deeper soil layers the model significantly underestimates soil moisture, presumably limiting plant's water use. This soil moisture bias is accompanied by underestimated LH and overestimated SH by 30 to 80 W m^-2, which can exceed the observed standard deviations among 17 sites used for evaluation. The model reasonably reproduces observed differences in BRs between the two surface types and characteristics of the flux probability distributions when all the grassland and cropland grid cells are sampled across the ~3° x 3° domain. But better constraints on deep soil moisture seem necessary to reproduce the observed magnitude of the surface fluxes, which is in line with previous studies suggesting the importance of vegetation for surface energy fluxes over the region.