Monsoon Research at Reading

SHIVA

CGAM's research programme on the Asian Summer Monsoon has recently gathered momentum with the 3-year funding of the EC Framework IV proposal SHIVA: Studies of the Hydrology, Influence and Variability of the Asian Summer Monsoon. SHIVA is being coordinated from CGAM and involves the major modelling groups at CNRM, ECMWF, LMD, MPI, UGAMP and UKMO. 3 posts have been funded at Reading, two in CGAM and one with Paul Valdes to study palaeomonsoons. SHIVA is a coordinated program of climate model analysis and experimentation with the following aims:

• To improve the simulation of the mean evolution of the monsoon, including its intraseasonal characteristics.
• To assess the ability of the participating models to simulate the observed interannual variability in the date of onset and subsequent strength of the monsoon from extended integrations with observed sea surface temperature (SST) anomalies. The large scale versus regional characteristics of the variability will be investigated.
• To investigate the mechanisms involved in the interannual variability of the monsoon, including its regional and intraseasonal characteristics. This will be achieved by coordinated sensitivity experiments to study the role of changes in the boundary forcing, specifically anomalies in tropical SST and springtime Eurasian snow cover.
• To assess the ability of the participating models to simulate the intraseasonal characteristics of the monsoon, particularly active/break phases and monsoon depressions.
• To investigate the mechanisms involved in the intraseasonal variability of the monsoon through coordinated sensitivity experiments to study the roles of land surface processes, atmosphere-ocean interactions over the Indian Ocean and Arabian Sea, and internal dynamics.
• To investigate the relationship between intraseasonal and interannual variability through analysis of ensemble integrations and through idealized experiments.
• To investigate the seasonal and interannual variability of the Asian Summer Monsoon in coupled ocean-atmosphere models, particularly the coupling between the monsoon and ENSO.

EuroCLIVAR

As well as SHIVA, the EC has also funded a concerted action, EuroCLIVAR, which will provide support in the form of workshops for CLIVAR-related research within the EC community. I shall be on the EuroCLIVAR committee as the coordinator of SHIVA, so if you need more information. please contact me. For those that donŐt know, CLIVAR is a new 15-year WCRP programme to study climate variability from seasonal to centennial timescales, including the response of the climate system to anthroprogenic forcing. CLIVAR supersedes the very successful TOGA programme.

Monsoon depressions

We have a new visitor at CGAM, Dr. G. D. Roy from Shahjalal University, Bangladesh, who is here for 10 months on a Commonwealth Fellowship. Dr. Roy has worked on storm surges in the Bay of Bengal. While he is here we hope to do some modelling studies of monsoon depressions using the limited area version of the UM.

Interannual variability

Fay Nortley is continuing to investigate the interannual variability of the Asian Summer Monsoon in an ensemble of 40-year integrations with the UM using observed SSTs. She plans to extend the work to look at the intraseasonal variability also, and how that might be linked to the interannual variability. This will complement the research almost completed by Laura Ferranti at ECMWF for her Ph.D. thesis. Her supervisors are myself, Brian Hoskins and Tim Palmer. Laura has been analysing an ensemble of AMIP integrations with the ECMWF model. Her results suggest that intraseasonal variability is not reproducible although the large scale interannual variability of the monsoon might be. She has also been investigating the role of land surface processes and the influence of El Nino on the intraseasonal behaviour of the model when it is run in perpetual July mode. These experiments have been interpreted in terms of the predictability ideas of Lorenz in which the boundary forcing may influence the probability of occurrence of one regime rather than another without affecting the structure of the regimes themselves.

Monsoon experiments with the UM

Mark Rodwell has been extending his research on various aspects of the Asian Summer Monsoon to include experiments with the UM. The results from idealised modelling studies of the monsoon, such as those concerning cross-equatorial PV (Rodwell and Hoskins 1995) and the role monsoons play in desertification (Rodwell and Hoskins 1996), are now being tested in the UM. Many of the ideas have a theoretical basis to them and the UM modelling should complete the chain of reasoning from theory to simple model to complex GCM to observations. If the results do validate previous theories, they may lead to better predictive ability in future. In collaboration with Kevin Hodges (ESSC, formerly NUTIS), Mark has also begun a project to investigate the life cycle of monsoon depressions. Previous papers have pointed to the importance of the westward migration of depressions along the monsoon trough from the South China Sea. Indeed, such depressions can strongly intensify over the Bay of Bengal leading to severe flooding and loss of life in Bangladesh. However, prior to arriving in the Bay of Bengal, these depressions become very weak as they cross South East Asia and in the past it has been difficult to track them. Using better tracking techniques developed by Kevin, and the newly produced ECMWF re-analysis data, it is hoped that a comprehensive study of these depressions can be made.

SST sensitivity paper

A paper describing the results of the SST sensitivity experiments performed by Soman when he was here has been completed and submitted for publication in the Quarterly Journal:

'Sensitivity of Asian Summer Monsoon to Aspects of Sea Surface Temperature Anomalies in the Tropical Pacific Ocean' by M.K. Soman and Julia Slingo.

The response of the onset and strength of the Asian Summer Monsoon to regional aspects of the Sea Surface Temperature (SST) anomalies in the tropical Pacific Ocean, associated with El Nino, has been investigated through a series of sensitivity experiments with the Universities' Global Atmospheric Modelling Programme (UGAMP) General Circulation Model (UGCM). This paper builds on the work of Ju and Slingo (1995) in which they hypothesised that the relationship between the Asian Summer Monsoon and ENSO involved the latitudinal position and strength of the Tropical Convective Maximum (TCM) over Indonesia and the West Pacific in the preceding spring. The inference from their results was that the modulation of the TCM might be associated with either changes in the Walker circulation through the influence of the East Pacific SST anomalies, or with changes in the local Hadley circulation associated with the in situ SST anomalies in the West Pacific. The experiments described in this paper are designed to isolate the effects of the principal SST anomalies in the East and central Pacific, associated with El Nino/La Nina, and those of opposite sign which develop in the West Pacific as a complementary pattern during the mature phase of El Nino/La Nina.

The results of the experimentation have shown that, during El Nino, the modulation of the Walker Circulation, with implied additional subsidence over the eastern hemisphere, is the dominant mechanism whereby the Asian Summer Monsoon is weakened. However, the late onset during El Nino years may also be associated with the complementary cold SST anomalies in the West Pacific which delay the northwards transition of the TCM. During La Nina, the modulation of the Walker Circulation is not the controlling factor which determines the stronger monsoons. The UGCM results suggest that the complementary warm anomalies in the West Pacific enhance the TCM, and it is this in situ response by the TCM that leads to an early onset and stronger monsoon. The importance of West Pacific warm anomalies in the development of a strong monsoon has been investigated further through a case study of the 1994 season. 1994 was an El Nino year in which the monsoon was unexpectedly active, but which was also characterised by warmer than normal SSTs in the West Pacific.

The sensitivity experiments have also elucidated the role of El Nino in influencing the precursory signature of stronger subtropical westerlies over India and S.E. Asia during the winter and spring preceding weak monsoons. The results have suggested that the equatorwards shift of the subtropical jet is a remote response to the warm SST anomalies in the central and East Pacific associated with El Nino.

Julia Slingo (Reading, CGAM)

Observational and Modelling Studies of Transport in the Middle Atmosphere

The aim of these studies is to characterize the interseasonal and interhemispheric variability of the flow regimes in the middle atmosphere.

The discoveries of severe Antarctic ozone depletion, and of lesser ozone losses at northern middle and high latitudes highlight the need to understand the morphology, evolution and transport processes of the middle atmosphere, as well as the chemical processes that are receiving much attention. The recent availability of three-dimensional, contemporaneous data sets, such as data from the Upper Atmosphere Research Satellite (UARS) and meteorological analyses of the middle atmosphere produced by the UK Meteorological Office (UKMO), plus the availability of a suite of numerical models at UGAMP (whether mechanistic or GCMs), provide an unprecedented opportunity for studying the transport processes of the middle atmosphere.

Distinct flow regimes have been identified in the stratosphere by using water vapour tracer data from the Microwave Limb Sounder (MLS) instrument on the UARS (see Lahoz et al. 1994 for a discussion of the northern hemisphere stratosphere, and Lahoz et al. 1996 for a discussion of the southern hemisphere stratosphere). These results have been corroborated by a 'domain-filling' Lagrangian technique, which involves computing the paths of thousands of air particles, numerous enough to give a high-resolution depiction of transport processes. Two flow regimes have been identified. One is in the polar vortex, where unmixed descent throughout winter produces vertically aligned tracer isopleths, which is manifested by a plug of moist air extending throughout the depth of the middle and upper stratosphere. The other is at mid latitudes, where quasi-stationary and travelling anticyclones stretch, fold and coil up material lines, leading to weaker horizontal gradients in tracer concentrations, and which is manifested by the presence of quasi-horizontal layers in the water vapour distribution.

For the first time, the phenomenon of vortex merger has been identified in the stratosphere of the southern hemisphere (Lahoz et al. 1996). This phenomenon is well known in a variety of geophysical flows, and had been recorded (in a collaboration between UGAMP and UKMO scientists) in the stratosphere of the northern hemisphere. This work has furthered understanding of transport in the middle atmosphere, and provides a framework to test numerical simulations.

Future observational studies will involve analysing the MLS water vapour data in the tropical stratosphere, with particular emphasis on the interseasonal and interannual variability, and transport between the tropics and the extra-tropics.

CGAM has a vigorous programme of modelling studies of transport of the middle atmosphere, and UGAMP groups at Edinburgh, Oxford and RAL are carrying similar studies. These studies involve using the UGAMP Stratosphere-Mesophere Model (USMM), the Extended UGAMP GCM (EUGCM) and the UKMO Unified Model (UM). The USMM is a mechanistic model, derived from the EUGCM, which uses as input the geopotential height at 100 hPa. This input can be constant (as in a perpetual January run), timevarying (as in a seasonal run using daily UKMO 100 hPa geopotential heights) or analytic. A variety of USMM runs focusing on northern winter and southern winter have been carried out. These runs will be diagnosed and compared with results from other numerical models and UKMO meteorological analyses. Initial results from these comparisons are very promising.

This research on middle atmosphere transport has links with several international programmes: (1) the NASA Upper Atmosphere Research Satellite Guest Investigator programme, via the 'Coordinated UK Studies of the Atmosphere using UARS data' proposal; (2) the ESA ERS Global Ozone Monitoring Experiment (GOME), via the 'Diagnostic and Numerical Modelling Studies to Interpret Data from GOME' proposal; (3) the EC Framework IV programme, via the 'EuroGRIPS' (European contribution to GRIPS - GCM-Reality Intercomparison Project for SPARC) project.

References

Lahoz, W.A., A. O'Neill et al., 1994: Three-dimensional Evolution of Water Vapor Distributions in the Northern Hemisphere Stratosphere as Observed by the MLS, J. Atmos. Sci., 51, 2914-2930.

Lahoz, W.A., A. O'Neill, A. Heaps et al., 1996: Vortex Dynamics and the Evolution of Water Vapour in the Stratosphere of the Southern Hemisphere, Quart. J. Roy. Met. Soc., 122, in press.

Alan O'Neill, William Lahoz, John Thuburn, Dave Knowles and Andy Heaps (Reading, CGAM)

Experimentation with IFS

Introduction

We have been experimenting with recent versions of ECMWF's forecast model (part of their Integrated Forecast System, IFS) for several months, following initial integrations of Cycle 12R1 at various horizontal and vertical resolutions, which were described at the UGAMP Summer Meeting in 1995. This article summarises results of these experiments and plans for future work. Thanks are more than due to Roger Brugge, Jeff Cole, Lois Steenman-Clark and all at ECMWF for their contributions to this work.

Cumulus Momentum mixing in Cycle 12R1

IFS Cycle 12R1 represents the last version to use derivatives of the physical parametrization schemes also developed for the UGAMP GCM (UGCM Version 2). In particular Cy12R1 includes envelope orography and orographic gravity wave drag based on the Palmer et al. (1986) scheme, and the diagnostic cloud scheme of Slingo (1987). Our experiments with Cy12R1 have used the Eulerian advection scheme and have been run using climatological SSTŐs in most cases for a single year, which severely limits statistical significance, particularly in mid-latitudes.

A further annual-cycle integration has been run at T42L31 resolution, with the mixing of mo- mentum by cumulus convection removed. As expected, the modelŐs tropical divergent circulation is stronger and there are also some improvements in the distribution of precipitation, cloud and surface fluxes, notably in the Australian Summer Monsoon and West Pacific, where the widely-split ITCZ appears to be corrected. There is more tropical variability overall. However comparison with analyses suggests that systematic errors in fields such as stream-function in the tropics are generally worse, simply because their amplitude is increased.

New Physical Parametrizations in Cycle 13R4

IFS Cycle 13R4 contains two major changes in physical parametrization, namely the use of mean orography with a revised orographic drag formulation (Lott & Miller, 1995), and a prognostic scheme for clouds (Tiedtke, 1993). The new schemes aim to capture the complexity of processes in a range of situations, since they are based on much more physically consistent representations of the relevant processes than the previous schemes.

Our first experiment with Cy13R4 was an annual cycle at T42L31, equivalent to the original resolution-sensitivity integrations of Cy12R1. There are several new diagnostic fields associated with the prognostic cloud scheme, and these reveal major changes in the cloud distribution, cloud-fraction and cloud-radiative forcing. Boundary layer cloud cover is larger at all latitudes and the major OLR bias associated with deep convective cloud in the tropics, seen in Cy12R1, is much improved. Storm-track cloud cover is both larger and more vertically coherent, and the previous isolated layer of cloud at the high-latitude tropopause has disappeared.

The net radiative cooling associated with both shallow and deep clouds occurs lower in the atmosphere (Fig.7). The boundary layer clouds are displaced from about 700hPa to 800hPa, leading to a smaller net long-wave flux at the surface. The radiative cooling from the tops of deep clouds is lower and stronger, with some improvements in temperature structure.

In mid-latitudes there appears to be a systematic improvement in the storm-track transient fluxes, though a single year is insufficient to be certain. The tropical divergent circulation is significantly weaker than in Cy12R1, but it is not clear whether this is associated with the parametrization changes alone. There is also a known bug in the convective downdraught mass-flux in Cy13R4, whose effect is as yet unknown. An intriguing question is whether removal of the cumulus momentum mixing in 13r4 will both achieve a realistic magnitude of the tropical circulation and also improve geographical patterns.

Semi-Lagrangian Advection in Cycle 13R4

We want to perform a clean comparison of Eulerian and Semi-Lagrangian advection in Cy13R4, particularly now that the S-L scheme uses 3-dimensional interpolation, parts of it quasi-monotone, for the trajectory calculation. The original version of the S-L scheme in IFS is described in Ritchie et al. (1995).

Some 30-day tests have been performed to check that a longer timestep in the S-L integrations, preferable because of the computational overhead of the S-L scheme, does not have systematic effects which could invalidate the Eulerian versus S-L comparison. A small increase in the strength of the hydrological cycle in the tropics is apparent, but differences are much smaller than Eulerian versus S-L differences using identical timestep lengths.

The main result of these S-L tests is a major impact on the zonally averaged temperature structure, particularly around the tropopause, consistent with other studies using vertically-interpolating S-L schemes (e.g. Williamson & Olsen, 1994). The high latitude tropopause is much improved, with the changes very similar but in the opposite sense to the tropopause errors seen in the Cy12R1 Eulerian integrations after a full yearŐs integration (Fig. 8). There is a corresponding equatorward movement of the Southern hemisphere (Summer) jet, again in the sense to correct previous errors. The test results are from a relatively short period within the model's spin-up, so it remains to be seen whether the improvements persist throughout the annual cycle.

Unfortunately the S-L test runs lose mass at a rate of roughly 1.5hPa/month, an order of magnitude faster than Eulerian runs, so it will be necessary to correct the global mass during model runs, as was done for the UGCM. Code has been obtained from the climate group at Meteo-France, which simply corrects the mass on model restart files, and either this or, preferably, a correction every timestep of the model will be implemented in Cy13R4 before a full annual cycle integration using S-L advection.

Future Plans

An annual cycle integration is underway, with cumulus momentum mixing removed in Cy13R4, to test the ideas already mentioned. Following this a Semi-Lagrangian integration will be performed.

It is impossible to draw conclusions on mid-latitude sensitivity to model version and options using the current single-year integrations, so it is proposed to run a small number further to obtain 5-year averages, dependent on availability of computer time. Using climatological SSTŐs, this should give sufficiently stable statistics, though interannual variability will be underestimated. However the results will complement ECMWF's seasonal integrations which include interannually varying SST's.

References

Lott, F. and Miller, M.J. 1995 A new sub-grid scale orographic drag parametrization: its formulation and testing. ECMWF Tech. Memo. No. 218, 31pp. (Submitted to Quart. J. R. Met. Soc.).

Palmer, T.N., Shutts, G.J. and Swinbank, R. 1986 Alleviation of a systematic westerly bias in general circulation and numerical weather prediction models though an orographic gravity wave drag parametrization. Quart. J. R. Met. Soc., 112, 1001-1039.

Ritchie, H, Temperton, C, Simmons, A.J., Hortal, M., Davies, T., Dent, D. and Hamrud, M. 1995 Implementation of the Semi-Lagrangian method in a high-resolution version of the ECMWF forecast model. Mon.Wea.Rev., 123, 489-514.

Slingo, J.M. 1987 The development and verification of a cloud prediction scheme for the ECMWF model. Quart.J.R.Met.Soc., 113, 899-927.

Tiedtke, M. 1993 Representation of clouds in large-scale models. Mon.Wea.Rev. 11, 3040-3061.

Williamson, D.L. and Olsen, J.G. 1994 Climate simulations with a semi-Lagrangian version of the NCAR CCM2. Mon.Wea.Rev., 122, 1594-1610.

Mike Blackburn (Reading, CGAM)

Developments with the Unified Model (UM) and access to Meteorological Data within the UGAMP community.

CGAM took delivery of the Unified Model code last October. We are in the process of running and testing the Global climate Version of the Unified Model on an HP workstation within CGAM. The Unified Model is a fully coupled system in so much as it contains a hierarchy of 3 atmospheric models embedded within each other. These are termed the Global Climate Model (GCM), Limited Area Model (LAM) and Mesoscale Model (MES) Details of the respective horizontal and vertical resolutions of the 3 models are given below. The atmospheric GCM can be run alone and also can be used to generate boundary conditions for a subsequent run of the LAM, over some period of the GCM integration. Similarly, the LAM can be used to generate boundary conditions for an integration of the MES. All 3 models can be run using initial and boundary conditions derived from the UK Met. Office Analyses or with climatological conditions.

Model Resolutions

GCM:

3.75 deg longitude by 2.5 deg latitude and 19 vertical levels

LAM:

0.44 deg longitude by 0.44 deg latitude and 19 vertical levels

MES:

0.15 deg longitude by 0.15 deg latitude and 31 vertical levels

Furthermore the atmospheric GCM can be coupled to an interactive 3D ocean model. The ocean model is at the same horizontal resolution as the atmospheric model. Potential therefore exists for coupled ocean-atmosphere work to be pursued. The coupled ocean-atmosphere model has not yet been tested within CGAM.

At present we are tied to generating the scripts which control the model integration on the Met. Office machines, using the Unified Model User Interface. A portable version of the User Interface is under development at the Met. Office and will ultimately allow external users of the UM to develop their own model integrations locally. We have a test version of this interface at CGAM and will receive a full release from the Met. Office in due course. This will increase the independence of UGAMP in terms of running the UM.

In the coming few weeks we will be testing the UM code on the J90 at RAL and will be subsequently doing some of the development work, such as testing the Limited Area Model, on this machine. Ultimately the UM code will be released for use by the wider UGAMP community. It is envisaged that UGAMP integrations will be performed centrally at RAL, or locally on workstations.

Allied to the UM project we are in the process of drawing up a data agreement with the Met. Office whereby data will be distributed to UGAMP scientists through CGAM. This agreement will cover the majority of non-commercial Met. Office data and will include the following:

1. Global Analyses and UM boundary condition files for the GCM, LAM and MES.
2. GCM output from Hadley Centre Integrations.
3. UM Ancillary files (e.g. Sea Surface Temperatures, Sea Ice coverage etc.)
4. Satellite Data
5. Some amount of RADAR data.
6. Radiosonde and upper air data
7. MRF Aircraft Data.

This is neither an exhaustive nor final list and we will reserve the option to add to it at intervals in the future.

Colin Jones (Reading, CGAM)

A Shape-Preserving Shallow-Water Model

In an earlier newsletter (number 12) I reported on a new shallow water model that has been developed. It uses a nearly uniform grid of pentagons and hexagons (Figure 9) and thus avoids the sort of problems seen with a regular longitude-latitude grid associated with small grid boxes near the poles. It uses potential vorticity (PV) as one of its prognostic variables. This means that PV should be handled more "cleanly" than in other formulations and should make PV-based diagnostics easier to interpret. PV, mass and any tracers are all advected using the same genuinely multidimensional shape-preserving advection scheme. "Shape-preserving" means it doesn't generate any spurious oscillations or unphysical negatives. One consequence of this is that the model runs without any supplementary scale-selective dissipation.

Since the previous report, the model has been subjected to a comprehensive set of test cases. These test cases were published with the intention that any new shallow water model could be compared with existing models that have already undergone the tests. A UGAMP Technical Report documenting the numerical schemes and the complete test case results is in preparation. The model performs reasonably well on all of the test cases. It shows none of the most conspicuous problems shown by some other models, such as problems near the poles, spurious oscillations or unphysical negatives, or lack of conservation. It is also reasonably efficient because it uses a semi-implicit time step; its CPU cost is similar to that of a spectral shallow water model.

In the short term I plan to apply the model to look at stratospheric wave breaking a la Juckes and McIntyre. In the longer term, if time permits, I would like to extend the model into three dimensions with an isentropic vertical coordinate.

John Thuburn (Reading, CGAM)

© 1996 Centre for Atmospheric Science/UGAMP. All scientific articles are unpublished. No text or graphics may be copied or used without permisson. Newsletter Editor: Glenn Carver, Cambridge University.