Sensitivity to soil moisture initialization in the simulation of Indian pre-monsoon season, using a regional climate numerical model

Soil moisture content is crucial for the representation and predictability of hydroclimatic extremes of different spatial/temporal scales such as heavy rainfall, droughts and heatwaves. In order to include these effects and the relevant feedback with the atmosphere in a regional climate model, the soil moisture initialization has to be adequate.

This study explores the soil moisture precipitation (SM-P) feedback, the soil moisture temperature (SM-T) feedback and the heat fluxes over the entire domain and 3 smaller regions of interest. A hydrostatic version of the Regional Climate Model  4.7 (RegCM4.7) with Arakawa B grid is used to run the simulations. The simulations  are performed for the months February to May during the years 2008, 2009 and 2010 with a spatial resolution of 12 km and temporal resolution of 3 hours. The initial and boundary conditions(ICBC)  are derived from the ERA5 data.  We examine results from simulations initiated using three different soil moisture datasets, namely, the control, dry and wet datasets. The soil moisture data from the ERA5-Land reanalysis is used for the control simulation. A dry/wet simulation is run using dry/wet datasets derived from the ERA5-Land data. This is done by halving/doubling the soil moisture values from ERA5-Land data, giving rise to new soil moisture values with lower/higher soil moisture as compared to the control dataset (ERA5-Land). CMORPH (Climate Prediction Center (CPC) Morphing Technique (MORPH)) and CRU (Climate Research Unit) datasets are used as reference to evaluate the precipitation and temperature values resulting from the control simulation.

The results display the mean changes in the dry/wet simulation results with respect to the control simulation. Plots showing the vertical profile changes in relative humidity and air temperature, and changes in lower tropospheric wind and specific humidity, indicates the build-up of the observed precipitation events and temperature patterns induced by the initial soil moisture perturbation. Interestingly the simulation results show negative SM-P feedback.  In other words, the average precipitation seemed to increase/decrease for the dry/wet cases with respect to the control simulation. This is contrary to the general expectation that dry/wet soil moisture decreases/increases precipitation. The possible reasons for the negative SM-P feedback and its distribution over the region include the proximity to the ocean, topography, and the pre-monsoon dryline. The SM-T and the heat fluxes on the other hand display expected behaviour with few exceptions in some regions in the dry simulation case.