Welcome to CGD

expand   collapse










 

CGD 2009 Profiles in Science: Dr. David Williamson

Summary of achievements

I established the equivalent resolutions for the finite volume and spectral transform versions of CAM, and considered the question of convergence of simulations with increasing resolution, including what scales, if any, converge. These studies were based on aqua-planet simulations and consider free motions of the atmosphere only, not a forced component. Along with Mike Blackburn and Brian Hoskins (University of Reading, UK), I continue to organize the coordinated Aqua-Planet Experiment (APE) under the auspices of WGNE. I also continue to refine and apply the CCPP-ARM Parameterization Testbed (CAPT) method to examine parameterizations. This approach runs the CAM in a forecast mode using an ensemble of historical analyses from very recent years for periods during which high quality field program measurements exist, such as ARM IOPs. This allows direct comparison of the parameterized variables (e.g. clouds, precipitation) with observations from the field programs.

Publications

Williamson, D.L., J.G. Olson and C. Jablonowski. 2009: Two Dynamical Core Formulation Flaws Exposed by a Baroclinic Instability Test Case. Monthly Weather Review, 137, 790-796, doi:10.1175/2008MWR2587.1.



Figure 1: High resolution figure

Abstract: Two flaws in the semi-Lagrangian algorithm originally implemented as an optional dynamical core in the NCAR Community Atmosphere Model (CAM3.1) are exposed by steady-state and baroclinic instability test cases. Remedies are demonstrated and have been incorporated in the dynamical core. One consequence of the first flaw is an erroneous damping of the speed of a zonally uniform zonal wind undergoing advection by a zonally uniform zonal flow field. It results from projecting the transported vector wind expressed in unit vectors at the arrival point to the surface of the sphere and is eliminated by rotating the vector to be parallel to the surface. The second flaw is the formulation of an a posteriori energy fixer that, although small, systematically affects the temperature field and leads to an incorrect evolution of the growing baroclinic wave. That fixer restores the total energy at each time step by changing the provisional forecast temperature proportionally to the magnitude of the temperature change at that time step. Two other fixers are introduced that do not exhibit the flaw. One changes the provisional temperature everywhere by an additive constant, and the other changes it proportionally by a multiplicative constant.

Figure caption: The 12 norm of the surface pressure differences (hPa) between various versions of the semi-Lagrangian model and the Eulerian spectral model, all at T170L26 resolution. All curves except FIXER 1 lie on top of each other.

Willett, M.R., P. Bechtold, D.L. Williamson, I.C. Petch, S.F. Milton and S.J. Woolnough. 2008: Modelling suppressed and active convection: Comparisons between three global atmospheric models. Quarterly Journal of the Royal Meteorological Society, 134, 1881-1896, doi:0.1002/qj.317.



Figure 2: High resolution figure

Abstract: The Met Office Unified Model, the European Centre for Medium-Range Weather Forecasts (ECMWF) Integrated Forecasting System and the National Center for Atmospheric Research (NCAR) Community Atmosphere Model 3 were used to compare simulations of suppressed and active convection by global atmospheric models (GAMs) in the Tropical West Pacific. The case used it three-week period selected from the Tropical Ocean and Global Atmosphere-Coupled Ocean-Atmosphere Response Experiment (TOGA-COARE). The global models were initialized from ERA-40 on each day of the case and the 12-36 h forecasts were concatenated together to create a continuous sequence of forecasts. One active and two Suppressed subperiods were identified and used to examine the behaviour of the models within different convective regimes. All the GAMs were able to clearly distinguish between the Suppressed and active regimes in terms of precipitation rate, precipitable water and radiative fluxes, and they all produced strong deep convection during the most active period. However, there were significant differences in the behaviour of the GAMs convection parametrizations during the most suppressed period. There was good agreement in the long-wave radiative fluxes where the GAMs were in good agreement in terms of convective activity, but there was less agreement in the short-wave fluxes, with the biggest differences being in the most active period. Throughout, there were large differences in the relative proportion of convective and large-scale precipitation produced by the GAMs. By making comparisons with reference data in the form of observations, reanalyses and cloud-resolving models, several issues that were specific to individual GAMs are identified and discussed.

Figure caption: Mean temperature budgets in K day-1 during B-Suppressed (top row), B-Active (middle row) and C-Suppressed (bottom row) from the MetUM (left column), IFS (middle column) and CAM3 (right column) in K day-1. The figures show the tendencies from the short-wave radiation (blue), the long-wave radiation (green), the dynamics (purple), the convection (red), the stratiform precipitation, stratiform cloud, boundary layer and vertical diffusion (orange), and the total tendency (black).

Williamson, D.L., F. Gasse, F. Chalié, A. Vincens and M.A.J. Williams. 2008: Climatic patterns in equatorial and southern Africa from 30,000 to 10,000 years ago reconstructed from terrestrial and near-shore proxy data. Quaternary Science Reviews, 27, 2316-2340, doi:10.1016/j.quascirev.2008.08.027 .



Figure 3: High resolution figure

Abstract: As part of a wider study of last glacial and deglacial climates in the Southern Hemisphere continents, we here review terrestrial and near-shore marine records from equatorial and southern Africa between 30,000 and 10,000 years ago (30–10 ka). This time interval covers the lead-up to the Last Glacial Maximum (LGM; 21 ± 2 ka), the LGM proper, and the ensuing deglacial. Records selected for review needed to meet three requirements: continuity or near continuity over the period; a well-established chronology; and at least one but preferably several unambiguous proxy(ies).

We aim to show how regional climates of the sub-continent have responded to orbital forcing as opposed to other global glacial-interglacial boundary conditions, and how they are related to high latitude climates, sea and land surface conditions, positions of the Intertropical Convergence Zone (ITCZ) and of the westerly belt. Evidence of past climates derived from many independent proxies is given from west to southwest Africa (moisture from the Atlantic Ocean), then from equatorial East Africa to the southern subtropical summer rainfall domain (moisture mainly from the Indian Ocean). The LGM was cooler than today, and generally drier in the tropics. North of 8–9°S, glacial to Holocene increase in monsoonal precipitation, primarily related to orbitally-induced summer insolation in the northern hemisphere, occurred by steps of increasing amplitude (not, vert, similar17–16, 14.5, 11.5 ka). Major wet–dry spells coincide with abrupt warm–cold events in high northern latitudes and related ITCZ migrations. In the southern tropics, the main post-glacial increase in tropical rainfall generally appears more gradual and in phase with Antarctic warming. Data suggest a restricted northward migration of the ITCZ and concentration of tropical rainfall well south of the Equator during the LGM and the Younger Dryas. Drier glacial conditions prevailed in southeastern Africa, while parts of southwestern Africa point to enhanced humidity during the LGM, suggesting that the winter westerly belt was either stronger than today or displaced further north possibly as a result of more extensive Antarctic sea-ice. Inferred African climatic fluctuations show the competing influences of tropical and high latitude climates of both hemispheres, and suggest changes in both meridional and zonal circulation modes. This review also reveals major geographical and methodological gaps, and a number of unresolved issues providing pointers for future research.

Figure caption: Difference in mean annual Sea Surface Temperature (ΔSST) or Surface Air Temperature (ΔSAT), and in mean annual precipitation (ΔMAP) or precipitation minus evaporation (Δ[P - E]) between the LGM (23–19 ka) and present-day. (A) and (B) are as inferred from the palaeoclimate proxies discussed in this review. Sites are numbered in B as in Table 1 and Fig. 3 (A) ΔSST or ΔSAT (red: lower than today). Digits are estimated anomalies compared to modern (mean, and extreme in parentheses). (B) ΔMAP or Δ[P - E]. Red and dark blue dots indicate generally drier or wetter conditions, respectively. Pale blue dots indicate increased available moisture (not necessarily increased P). Arrows connected to dots indicate that the authors have used a marine sediment core to infer moisture conditions over the continent. Digits in rectangles represent the estimated ΔMAP values in percentage relative to modern (mean, and extreme in parentheses). The red dotted line is the limit of the present-day Congo Basin. The dark blue line is the approximate boundary of wetter SW Africa according to Chase and Meadows (2007). (C) Simulation of Δ[P - E] from the atmospheric-GCM NCAR CCM3 with a grid size of not, vert, similar75 km (Kim et al., 2007). Units are in mm per day.

Williamson, D.L. 2008: Equivalent finite volume and Eulerian spectral transform horizontal resolutions established from aqua-planet simulations Tellus Series A-Dynamic Meteorology and Oceanograpy, 60, 839-847, doi:10.1111/j.1600-0870.2008.00340.x.



Figure 4: High resolution figure

Abstract: The equivalent resolutions for two different global dynamic cores are established when they are coupled to the sub-grid scale parametrization suite of the Community Atmosphere Model (CAM3). One core adopts the common Eulerian spectral transform formalism, the other adopts a finite volume approach. The equivalent resolutions are established over a range of resolutions employed today for climate models. The comparison is done in the context of the Aqua Planet Experiment (APE). Thus, it is based on the characteristics of free, unforced motions, due in large part to the dynamic component driven by the parametrized processes and explicit dissipation. The forced component arising from surface orography and land-ocean-sea-ice contrasts is not considered. The resolution equivalences are demonstrated for a number of model fields. These include selected time averaged, global and zonal averaged fields, the meridional structure of eddy kinetic energy and eddy temperature variance, the mean meridional eddy transports, the characteristics of tropical wave propagation and probability density functions of precipitation. These fields indicate that the 2 degrees finite volume model is equivalent to T42 spectral transform model, 1 degrees is equivalent to T85 and 0.5 degrees is equivalent to T170. This proportional relationship does not hold at lower resolutions.

Figure caption: Time average, zonal average of (a) surface pressure, (b) zonal surface stress and (c) cloud fraction for T42, T85, T170 and T340 spectral and 2°, 1° and 0.5° finite volume models.

Williamson, D.L. 2008: Convergence of aqua-planet simulations with increasing resolution in the Community Atmospheric Model, Version 3. Tellus Series A-Dynamic Meteorology and Oceanograpy, 60, 848-862, doi:10.1111/j.1600-0870.2008.00339.x.



Figure 5: High resolution figure

Abstract: The convergence of simulations from the Community Atmosphere Model with increasing resolution is determined in an aqua-planet context. Convergence as a function of scale is considered. Horizontal resolution (T42-T340) and time step (40-5 min) are varied separately. The simulations are sensitive to both. Global averages do not necessarily converge with increasing resolution. The zonal average equatorial precipitation shows a strong sensitivity to time step. Parametrizations should be applied in a range of time steps where such sensitivity is not seen. The larger scales of the zonal average equatorial precipitation converge with increasing resolution. There is a mass shift from polar to equatorial regions with increasing resolution with no indication of convergence. The zonal average cloud fraction decreases with increasing resolution with no indication of convergence. Equatorial wave propagation characteristics converge with increasing resolution, however a relatively high truncation of T170 is required to capture wavenumbers less than 16. Extremes are studied in the form of the probability density functions of precipitation. The largest half of the scales of the model converge for resolutions above T85.

Figure caption: Time averaged, zonal averaged surface pressure for T42, T85, T170 and T340 with Δt= 5 min, total and for specified symmetric spherical harmonics (Pmn) , m= 0 : n= 2 ; n= 4 ; n= 6 ; n= 8 and 10; and n= 12 through the truncation limits.

Williamson, D.L. 2008: Aquaplanets, climate sensitivity, and low clouds. J. of Clim., 21, 4974-4991, doi:10.1175/2008JCLI1995.1.



Figure 6: High resolution figure

Abstract: The convergence of simulations from the Community Atmosphere Model with increasing resolution is determined in an aqua-planet context. Convergence as a function of scale is considered. Horizontal resolution (T42-T340) and time step (40-5 min) are varied separately. The simulations are sensitive to both. Global averages do not necessarily converge with increasing resolution. The zonal average equatorial precipitation shows a strong sensitivity to time step. Parametrizations should be applied in a range of time steps where such sensitivity is not seen. The larger scales of the zonal average equatorial precipitation converge with increasing resolution. There is a mass shift from polar to equatorial regions with increasing resolution with no indication of convergence. The zonal average cloud fraction decreases with increasing resolution with no indication of convergence. Equatorial wave propagation characteristics converge with increasing resolution, however a relatively high truncation of T170 is required to capture wavenumbers less than 16. Extremes are studied in the form of the probability density functions of precipitation. The largest half of the scales of the model converge for resolutions above T85.

Figure caption: Wavenumber-frequency diagrams of log of power of equatorial precipitation for (left-hand column) T42 with Δt= 40 , 20, 10 and 5 min; and (right-hand column) T85 with Δt= 20 , 10 and 5 min.