During drought, low soil moisture, and vapor pressure deficit cause plants to tightly co-regulate water loss and CO2 assimilation. This impacts primary terrestrial ecosystem production, an important indicator for numerous ecosystem functions. As a measure of this productivity, gross primary production (GPP) is highly relevant for estimating agricultural production because it represents the carbon fixed through photosynthesis and is therefore the foundation for the accumulation of crop biomass. However, measuring this productivity in the field can be difficult and expensive. Soil moisture and vapor pressure deficit are the primary water stress drivers that influence vegetation health, but soil moisture has traditionally been difficult to measure, especially at large spatial scales.

This study uses light use efficiency (LUE) models that include vapor pressure deficit and soil moisture factors to estimate GPP and investigate how these water stress metrics influence vegetation productivity and model skill. Differences in these models are assessed and compared relative to the specific site biome and how environmental scalar factors behave. Ecosystem production data was used from the Ameriflux network of towers and satellite-based soil moisture measurements from SMAP. We focused on sixteen sites in the upper midwestern United States that encompassed cropland, evergreen needleleaf, and deciduous broadleaf biomes.

This study found that incorporating satellite-based soil moisture estimates into an LUE model can improve GPP estimates in some scenarios including mid to late summer period primarily at the cropland sites. Late spring also saw periods where incorporating soil moisture may be beneficial for model accuracy. These findings show that vapor pressure deficit alone may not capture the full impacts of moisture stress on vegetation productivity. Especially in agricultural ecosystems, there are times of the year when soil moisture effects are necessary to capture changes in productivity.