Regional Climate Modeling of Vegetation Feedbacks on the Asian-Australian Monsoon Systems

We are exploring the hypothesis of Notaro et al. (2011) that the global monsoon regions exhibit unique responses to vegetation feedbacks, with a greater sensible heat response for subtropical monsoon regions and greater latent heat response for tropical monsoon regions.  Specifically, that study concluded that a reduction in leaf area index (LAI) led to an earlier subtropical Chinese monsoon in contrast to a delayed, weaker tropical Australian monsoon.  The study applied the coarse (1.9° x 2.5°), fully coupled global climate model, the National Center for Atmospheric Research (NCAR) Community Climate System Model Version 3.5 (CCSM3.5), which unfortunately exhibited substantial biases in regional climate and LAI across the monsoon regions, with excessive forest cover across China due to the absence of land use in the model; these biases obfuscate the reliability of the findings on land-atmosphere interactions and resulting hypothesis.

            In response to these concerns, we developed a version of the International Centre for Theoretical Physics (ICTP) Regional Climate Model Version Four (RegCM4), coupled to the NCAR Community Land Model (CLM, as also in CCSM3.5) with land use, that applied the interannually-varying reconstructed plant functional type (PFT)-specific LAI during 1960-2013 from Stockli et al. (2008, 2011) as boundary conditions.  We ran RegCM4 at 40-km grid spacing for two study domains, namely China and Australia.  The regional climate model (RCM) matches the observed dominance of cropland, C3 grassland, and temperate needleleaf evergreen forest in southern China and of C4 grassland and temperate broadleaf deciduous shrubland in northern Australia.  In order to determine optimal model configurations for each domain, we produced 21 simulations of the Chinese domain and 26 simulations of the Australian domain for 2010-2012, excluding the first year for model spin up.  The runs differed in terms of planetary boundary layer scheme, explicit moisture scheme, convective scheme, cumulus closure scheme, and numerous coefficients and thresholds.  Using metrics of mean bias, spatial correlation, and root-mean-square-difference, we evaluated the 47 simulations in terms of over-land air temperature, over-land precipitation, over-ocean precipitation, and domain-wide cloud cover fraction for each month during 2011-2012, in comparison with the University of Delaware, Tropical Rainfall Measuring Mission (TRMM), and Pathfinder Atmospheres-Extended (PATMOS-x) products.  The optimal configurations were then used to produce extended control simulations for 1960-2013.  In general, air temperature was better simulated in the Australian domain, and cloud cover was better simulated in the China domain.  The Chinese domain is characterized by a cold bias in late spring-early summer, wet bias in winter-spring, and dry bias in mid-late summer, and the Australian domain is characterized by a summertime cold bias, wintertime warm bias, and late summertime dry bias.

            We next developed an ensemble of RegCM4 experiments, aimed at contrasting vegetation feedbacks between tropical and subtropical monsoon regions and examining potential nonlinearities and responsible mechanisms of land-atmosphere feedbacks.  For each month during April-September for the Chinese domain and November-April for the Australian domain, 30 experiments were produced, each one-month in duration, in which LAI was increased by 0.5 over either the Chinese monsoon region or the Australian monsoon region only and 30 experiment were produced in which LAI was decreased by 0.5.  It is demonstrated that 30 ensemble members in each direction of LAI anomaly is sufficient for achieving stable estimates of the atmospheric response to vegetation anomalies in the monsoon regions.  In general for both monsoon regions, greater LAI supported reductions in surface albedo, air temperature within the planetary boundary layer (PBL), low-level wind speed, PBL height, low-mid tropospheric ascending motion, and mid-level clouds and increases in diurnal temperature range (DTR), wind stress, evapotranspiration (ET - especially canopy transpiration), specific humidity within the PBL, low-level clouds, and the probability of precipitation.  In response to greater LAI, mean rainfall was enhanced in the pre-mid monsoon season for Australia but did not significantly change over the Chinese monsoon region.  Likewise, significant anomalous subsidence was generated during the entire pre- and full monsoon season across northern Australia yet restricted to the early monsoon month of June for southern China.  Modified LAI led to dramatic changes in the temporal distribution and intensity of daily rainfall events for the Australian monsoon only, with greater LAI favoring an increased probability of wet days, fewer dry days, more light-moderate rainfall days, and fewer heavy rainfall days.  Within the monsoon regions, the responses in surface albedo, DTR, wind stress, ET, and low-level wind speed and specific humidity exhibited clear spatial heterogeneity based on the distribution of biomes, with amplified impacts across China’s croplands and Australia’s shrublands.  It is suggested that the atmospheric response to modified LAI over the Chinese monsoon region is inconsistent between the present study and that of Notaro et al. (2011) due to biases in simulated vegetation in CCSM3.5, particularly an exaggerated tree cover and thus an amplified response in surface albedo.