The objective of this research was to test the following hypothesis regarding the nature of the soil moisture-rainfall feedback in semiarid regions. There is a dramatic land surface response to precipitation events and the associated rise in soil moisture, but this response is limited in duration. In addition, plant type does not influence the nature of this response. We tested this hypothesis by measuring the surface water and energy balances at three locations across the shrub-ecotone in the Sevilleta Wildlife Refuge. Measurements spanned the entire 2000 monsoon season.

Our analysis yielded four major results. First, changes in the evaporative fraction (EF) resulting from wet versus dry soil moisture conditions are dramatic. When the soil is dry, the evaporative fraction is typically 0.1, demonstrating that only 10% of the energy transferred from the surface to the atmosphere is via latent heating. In contrast, the evaporative fraction is ~0.5 when the soil is wet. This demonstrates that the surface energy balance (SEB) response to rainfall, at least in terms of latent heating, is substantial and could yield a feedback to the atmosphere. Second, net radiation and available energy both increase when the soil is wet, in both the grass and shrub environments. A volumetric water content change of ~5% yields an increase of 50 W m-2. If this result is accurate, then the soil moisture-net radiation feedback proposed by Eltahir (1998) would contribute to soil moisture-rainfall feedbacks in semiarid regions such as the Sevilleta. Third, changes in EF and net radiation following rainfall events are short lived – persistence is on the order of days. Therefore, a soil-moisture rainfall feedback will only exist if the atmospheric conditions conducive for convective precipitation occur within several days after a rainfall event. Fourth, the evaporative fraction and net radiation response to rainfall and the persistence of anomalous conditions are nearly identical across the grass-shrub ecotone. Therefore, we conclude that plant type does not influence the nature of soil moisture-rainfall feedbacks in semiarid regions.