#L20 Implementing variable soil thickness in ALM

Poster Title

Variable soil thickness in ALM: Implementation without elevation classes and preparation for implementation with elevation classes

AuthorsMichael Brunke, Xubin Zeng
GroupLand
ExperimentWatercycle
Poster CategoryEarly Result
Submission Type
Poster LinkACME_Results_VARSOIL.pptx


Abstract

Currently, the ACME Land Model (ALM), like most other land surface models, has a constant depth (~3.8 m) that is hydrologically active. To compensate for this constant “soil” depth, a poorly-defined unconfined aquifer somewhere below this depth is included. As an alternative, variable soil thickness based upon our global ~30 arcsec (or ~1 km) estimate of bedrock depth (Pelletier et al. 2016) has been implemented into the ALM for inclusion into ACMEv2. This implementation allows for the removal of the unconfined aquifer when turned on. As seen in Brunke et al. (2016) for CLM4.5, soil moisture profiles are most impacted in locations where the soil column is made shallower than the original 10 layers. Hydrologic fluxes such as surface runoff and baseflow are also impacted. Surface runoff is less affected with small changes to the amplitude of the mean annual cycle at some locations. Baseflow is more affected from changes to annual cycle amplitude and temporal changes to the timing of the annual maximum.

We are furthering the implementation of variable soil thickness by improving the cold start spin-up and supporting the inclusion of elevation classes which will enhance the representation of variable soil thickness in the model. Spin-up with variable soil thickness takes at least 200 years for the soil temperature of deep layers to reach equilibrium. This happens because of the lag needed to move heat from the surface down to deeper layers due to an unrealistic initialization of a constant 274 K for non-urban and non-lake grid cells. We propose to initialize soil temperature based upon the climatological 2-m air temperature at each particular grid cell. We are also investigating the best ways to map atmospheric forcings from grid-average quantities to quantities for the individual elevation classes within grid cells. The quantities that need to be mapped include near-surface temperature, humidity, and wind as well as surface precipitation and surface downward shortwave and longwave radiation.