UJCC Science Highlights in year 2
Written by Pier Luigi Vidale   
Friday, 05 January 2007


SSTs in the Tropical PacificTropical Pacific simulation

  The tropical Pacific simulation in the HadGEM1 model is much too cold at the equator, leading to a poor simulation of the ocean thermocline and atmospheric Walker circulation. Both these may play an important role in vairability processes such as ENSO. Using the model matrix, it is found that enhancing the ocean resolution leads to a significant improvement in the SST and deeper ocean simulation, and also helps the atmospheric mean state. Analysis suggests that tropical instability waves (TIWs), which form near the equator due to shear instabilities in teh ocean, play a primary role in converging heat onto the equator and hence warming the ocean surface there. Such waves are poorly resolved at low ocean model resolution, particularly in the East Pacific (Figure shows the SST standard deviation from models and observations), even when model dissipation parameters are given extremely low values. The improvement of the SST mean state, due to the TIWs, has important consequences for the atmospheric circulation, giving an improved Walker circulation along the equator and better placements of convection int he Weste pacific warm pool. A paper on this work will shortly be submitted to Journal of Climate.

Simulation of tropical cyclones


Analysis of the atmosphere-only (AMIP) integrations at different model resolutions (and including results from Japanese GCMs) shows that the properties of tropical cyclones (TCs) vary strongly with resolution. The intensity distribution shifts towards a profile with relatively less weak events and more of the very strong events (see Figure), with a long tail. Statistics also suggest enhanced formation of TCs off Africa, which improves the Central and North Atlantic distribution. These results suggest that resolution is crucial for such processes. Further work will analyse winds and precipitation associated with the TCs, and in particular regional tracks and impacts and interaction with ENSO and MJO. New simulations are underway, using idealised SST and also high-resolution SSTs from the high-resolution work, to address the nature of tropical cyclones under a warmer climate. A joint UK-Japan publication on the impact of high resolution on tropical cyclone simulatio is planned, with a possible successor, using the scenario simulations.





Nino 3 power spectrum

El Niño-Southern Oscillation (ENSO)


The ENSO simulation in HadGEM1 is rather poor, with few strong warm or cold events and consequently poor global patterns of SST and precipitation in ENSO years. Since the annual variability in many regions of the globe is strongly linked to ENSO and other modes of variability, it is very important to simulate such modes properly so that, for example, climate change signals can be detected. Increasing the atmosphere or ocean model resolution from HadGEM1 have some impact on the ENSO simulation, but it is only when both components take the higher resolution that the power spectrum of Nino3 variability starts to look more realistic (Figure). We are actively researching what it is about having enhanced resolution that makes such an important difference - perhaps better initiation by westerly wind bursts, or the improved mean state, or atmospheric and oceanic mesoscale variability.



Tropical convection


 A fundamental mode of variability of the global climate system is the diurnal cycle, which is associated with a large and well-defined variation in solar forcing. The response to this forcing is therefore a key test of the correctness of many, interacting processes in a climate model.


The solar insolation is reflected and absorbed in the atmosphere, dependent on the structure and extent of the clouds. The tropical land surface warms up and soil moisture evaporates during the day, creating a warm and moist boundary layer. This builds up potential energy in the atmosphere, which is released in the afternoon by convection. As a result, convective precipitation peaks roughly between 2 and 4pm. However, the precipitation in the HadGEM1 model peaks already around noon, because the convection responds too fast and too vigorous to the buildup of potential energy. We have changed the initiation of convection to respond to the vertical velocity of air, which is an indicator of the ease with which convective towers can develop. The figure below is the local time of the peak in convective precipitation, and it shows that this change in the model brings the simulation much closer to the observed afternoon maximum over tropical land areas.




The Madden-Julian Oscillation (MJO)



The MJO is a major feature of the tropical climate, characterised by large-scale areas of organized convection, which typically form in the East Indian Ocean and propagate towards the Central Pacific over the course of a month or more. Via its profound effects on the tropical circulation, the MJO has strong connections with a number of other tropical phenomena, including ENSO, the Indian Monsoon and Tropical Cyclones, all of which are of interest to UJCC.During the year, a basic analysis was performed to assess the variation of MJO characteristics within the UJCC model matrix; particularly the sensitivity to resolution and basic-state changes. Later, this was extended to include data from an atmosphere-only version of MIROC, kindly provided by our Japanese collaborators.The coupled UJCC models were all found to produce world-class simulations of MJO, with relatively minor variations between them. In particular, the strength and the spatial and temporal scales are all well-captured. The plots in figure X illustrate the propagation of (a) observed , and (b) modelled convective anomalies on MJO timescales. The model captures well the observed propagation of the anomalies across the Indian Ocean (40E-100E), but fails to capture its propagation into the Western Pacific (120E-). This is thought to be related to basic state errors in the Equatorial Pacific.





NUGAM, best monsoonMonsoon and soil moisture dynamics

The Indian summer monsoon is one of the main issues for the Met Office Hadley Centre and NCAS-Climate, as the different resolution versions of HadGEM1a don't simulate it properly. It is however better in the 60kms atmosphere model (NUGAM) and in some years, the observed regime is represented quite well (Fig.). One monsoon especially, with good characteristics, has been simulated by NUGAM (Fig.timeseries), but the model generally tends to generate either no monsoon at all, or a break event, preventing heavy rain to occur. Some hypotheses have been suggested in order to understand what would initiate and maintain the monsoon in that particular model configuration. Among these are the resolution and especially the inclusion of mesoscale mountains, which affect the local circulation and represent more realistic Himalayan and Eurasian snow cover, important for the onset of the monsoon. Another difference between the low and high-resolution models is in the formulation. formulation_precip_india_timeseries_monthlyDue to the stability requirements of NUGAM, it has been necessary to modify the treatment of snowmelt infiltration in the land surface parametrisation. This had a large impact on the surface energy balance and consequently on the Asian continent-Indian ocean temperature gradient, which drives the monsoon. The land surface also seems to have an impact on the maintenance of the monsoon process, via evapotranspiration limitations; in this context, a better representation of soil physical properties is important. A new set of soil physical parameters, using Earth Observatory-generation data, has been constructed for the high-resolution model. Further hypotheses about the importance of high-resolution for land surface processes (e.g. precipitation frequency and distribution) are currently being formulated and will allow us to test and improve the low-resolution models.

Last Updated ( Thursday, 12 July 2007 )