focussing on the Antarctic Mesoscale Prediction System (AMPS)


The International Antarctic Weather Forecasting Handbook:

IPY 2007-08 Supplement


Jordan Powers

Mesoscale and Microscale Meteorology Division

National Center for Atmospheric Research

Boulder, Colorado, USA


Submitted March 2008

*Contribution relevant to Section 4.2 NWP Model Fields.

Editors’ note: at this time, the contribution has not been adapted to the original Handbook style, especially wrt numbering of figures etc.

Only a few institutions in the United States undertake real-time numerical weather prediction (NWP) covering the Antarctic and sub-Antarctic.  Global models naturally address these regions, and the U.S. operational centers running these are the National Weather Service’s National Centers for Environmental Prediction (NCEP) and the U.S. Navy’s Fleet Numerical Meteorology and Oceanography Center (FNMOC).  NCEP runs the Global Forecasting System (GFS), and FNMOC runs the NOGAPS (Naval Operational Global Atmospheric Prediction System) Model. 

Apart from the global modeling, two groups have applied mesoscale models to the region. The one targeting the Antarctic and sub-Antarctic exclusively is that of AMPS— the Antarctic Mesoscale Prediction System (Powers et al. 2003).  Detailed below, this is a U.S. National Science Foundation (NSF)-funded program to develop and run a high-resolution modeling system for Antarctica and key areas within it.  AMPS employs the Weather Research and Forecasting (WRF) model (described below).  The other mesoscale NWP over Antarctica is done by the U.S. Air Force Weather Agency (AFWA).  While AFWA runs WRF with a domain over the continent, distribution of the model forecast products is restricted.

The Antarctic Mesoscale Prediction System (AMPS) is an experimental mesoscale NWP capability providing support to the weather forecasting for the United States Antarctic Program (USAP).  The history of AMPS dates from May 2000 and the Antarctic Weather Forecasting Workshop held at the Byrd Polar Research Center (BPRC) of The Ohio State University.  The workshop recognized that output from NWP models was critical to forecasters at McMurdo Station, who provided the forecasts for USAP flights to McMurdo from Christchurch, New Zealand and flights to/from McMurdo over the continent.   The guidance of the models then available was felt to be lacking due to: (i) horizontal grid resolutions inadequate to capture features crucial for short-term (6–24 hr) forecasting and flight operations, (ii) inadequate representation of characteristics of the Antarctic PBL and troposphere, and (iii) poor representation of Antarctic topography and surface features (Bromwich and Cassano 2000).  To improve NWP capabilities for the USAP the workshop recommended an Antarctic mesoscale modeling initiative and the implementation of high-resolution forecast domains (Bromwich and Cassano 2000).  Other key recommendations addressed the need for a product suite tailored to the evolving requirements of the forecasters, implementation of parameterizations tuned for the Antarctic, and verification.  From these needs, the AMPS Project was conceived. 

The NSF-supported AMPS furnishes twice-daily numerical guidance for Antarctica and the sub-Antarctic using the version of the Weather Research and Forecasting (WRF) model called the ARW (Advanced Research WRF) (Skamarock et al. 2005).  The principals in the AMPS effort are the National Center for Atmospheric Research, The Ohio State University (the Polar Meteorology Group of the Byrd Polar Research Center), and the University of Colorado.  The USAP forecasters provide regular feedback, and the system is upgraded on an ongoing basis.  Beyond AMPS’s use for routine forecasting, special applications of it have figured prominently.  Such applications have included support for scientific field campaigns, medical and marine rescues, and international facilities.  In addition, to researchers in Antarctic meteorology and climatology, AMPS provides a capability for process and event studies, a platform to test parameterizations in a polar region, and a database of high-resolution numerical output.  The latter is the AMPS archive, and it contains the model forecasts since 2001.  AMPS provides guidance used in support of the Antarctic forecasting needs of a host of nations, including Italy, Germany, Australia, the UK, South Africa, and Chile.  

AMPS has six computational/forecast grids, shown in Fig. 1 (NWP - USA).  The configuration features an outermost grid of 60-km horizontal spacing, nesting down to a finest grid of 2.2-km spacing.  The 60-km grid covers the middle and high latitudes of the Southern Hemisphere, the 20-km covers the Antarctic continent, and three 6.7-km grids cover the western Ross Sea, the South Pole, and the Antarctic Peninsula (Figs. 1(a),(b)).  The 2.2-km grid (Figs. 1(b),(c)) covers Ross Island and vicinity.  The ARW in AMPS is currently run with 31 vertical levels with a model top at 50 hPa.

Initial and boundary conditions for the ARW in AMPS are derived from GFS model output.  The first-guess fields derived from the GFS are reanalyzed with observations using WRF-Var (Barker et al. 2004), a 3-dimensional variational data assimilation system.  At the present time, the observational datasets assimilated include conventional surface and upper-air observations (surface synoptic, METAR, AWS, radiosonde, ship, buoy, pilot and AMDAR observations/reports), satellite winds (geostationary and MODIS (Moderate-resolution Imaging Spectroradiometer), and COSMIC (Constellation Observing System for Meteorology, Ionosphere, and Climate) GPS radio occultations.  Sea ice analyses from the National Snow and Ice Data Center (NSIDC) in the U.S. initialize the sea ice coverage. 

Two forecasts per day are produced, initialized at 0000 and 1200 UTC.  Forecast lengths are 120 hr for the 60- and 20-km grids and 36 hr for the 6.7- and 2.2-km grids.  The AMPS forecasts are disseminated primarily via the web, at http://www.mmm.ucar.edu/rt/amps/wrf_pages .  AMPS products are also sent via e-mail to users with limited bandwidth.  AMPS output and products are also put on the Antarctic Internet Data Distribution (IDD) system, a capability to share Antarctic meteorological data though a collaborative network of participating institutions (Lazzara et al. 2006).

A scaled-down backup forecast for AMPS is produced by the Polar Meteorology Group of the Byrd Polar Research Center (The Ohio State University).  For this, BPRC runs a polar-modified version of the MM5 model (Bromwich et al. 2001; Cassano et al. 2001) twice a day on a single 60-km grid covering the continent.  The forecast products are displayed at http://polarmet.mps.ohio-state.edu/PolarMet/Antarctic nwp.html .


(a)                                                               (b)


Figure 1 (NWP – USA ): AMPS forecast grids.  (a) 60-km Southern Hemisphere (outer frame) and 20-km Antarctica (inner frame) domains.  (b) 20-km Antarctica, 6.7-km western Ross Sea, 6.7-km South Pole, 6.7-km Antarctic Peninsula, and 2.2-km Ross Island grids.  (c) 6.7-km western Ross Sea and 2.2-km Ross Island grids.

AMPS uses the Advanced Research WRF (ARW) model (Skamarock et al. 2005).  AMPS’s configuration of physics packages includes cumulus and explicit microphysical schemes and boundary layer and land surface model packages.  The AMPS website lists the schemes used.  The ARW in AMPS contains a number of polar modifications to capture conditions unique to ice sheets and the high latitudes, and this version is referred to as “Polar WRF”.  The majority of the modifications were developed at BPRC, with additional ones coming from NCAR and NCEP.  The modifications include: fractional sea ice coverage in grid cells; use of the latent heat of sublimation for calculations of latent heat fluxes over ice surfaces; assumption of ice saturation when calculating surface saturation mixing ratios over ice; modifications to the density, heat capacity, and heat conductivity of the snowpack; modifications to the thermal roughness length over ice surfaces; and an adjustment of the shortwave radiation scattering parameter. 

Three modes of verification have been used to analyze the NWP of AMPS.  The first involves quantitative reviews of forecasts for target periods.  Error statistics are computed and analyzed for periods of weeks and longer.  Bromwich et al. (2005), for example, present results for statistical analyses of a 2-year period (2001– 2003) of AMPS forecasts.  They found that the AMPS polar setup produced surprisingly good forecasts considering the complex coastal topography, the limited amount of surface-based observations assimilated, and the reduced quality of global analyses over the continent.  Positive benefits from AMPS’s enhanced grid resolution in the Ross Island area were found.  Shorter-term verifications of AMPS are also performed, such as that of Powers and Manning (2007).  This study examined two three-week periods in the warm and cold seasons to compare WRF/Polar WRF and Polar WRF/Polar MM5.

A second type of verification is that of the case study.  For example, Powers (2007) examined the performance of the ARW model in AMPS through the simulation and observational analysis of an extreme wind event at McMurdo.  It was found that AMPS running the ARW could capture the event well, with a significant benefit seen from the assimilation of MODIS satellite wind measurements.  Other case verifications are described in Monaghan et al. (2003) (AMPS model performance for the Shemenski rescue from the South Pole in April 2001) and Bromwich et al. (2003) (mesoscale cyclogenesis in the Western Ross Sea).

A third type of verification is reflected in the forecast scrutiny, ongoing comparison with observations, and daily review/discussions conducted by the SPAWAR forecasters.  They assess AMPS on an ongoing basis as verifying observations show how justified their confidence, or lack thereof, in the model progs was.  They develop an understanding of model biases for certain regions, synoptic patterns, and weather scenarios.  This information is important in the forecasters’ interpretation and application of the NWP guidance, and it is communicated to the AMPS developers for analysis and development of improvements.  These types of findings are also presented at yearly workshops, such as the Antarctic Meteorological Observation, Modeling, and Forecasting Workshop.

References cited by Powers

Barker, D.M., W. Huang, Y.R. Guo, Q.N. Xiao, 2004: A three-dimensional (3DVAR) data assimilation system for use with MM5: Implementation and initial results.  Mon. Wea. Rev., 132, 897–914. 

Bromwich, D.H., and J.J. Cassano, 2000: Recommendations to the National Science Foundation from the Antarctic Weather Forecasting Workshop.  BPRC Misc. Series M-420, 48 pp.  [Available from Byrd Polar Research Center, The Ohio State University, 1090 Carmack Rd., Columbus, OH, 43210-1002]

Bromwich, D.H., J.J. Cassano, T. Klein, G. Heinemann, K.M. Hines, K. Steffen, and J.E. Box, 2001: Mesoscale modeling of katabatic winds over Greenland with the Polar MM5.  Mon. Wea. Rev., 129, 2290–2309.

Bromwich, D.H., A.J. Monaghan, J.G. Powers, J.J. Cassano, H.-L. Wei, Y.-H. Kuo, and A. Pellegrini, 2003: Antarctic Mesoscale Prediction System (AMPS): A case study from the 200/2001 field season.  Mon. Wea. Rev., 131, 412–434.

Bromwich, D.H., A.J. Monaghan, K.W. Manning, and J.G. Powers, 2005: Real-time forecasting for the Antarctic: An evaluation of the Antarctic Mesoscale Prediction System (AMPS).  Mon. Wea. Rev., 133, 579–603.

Cassano, J.J., J.E. Box, D.H. Bromwich, L. Li, and K. Steffen, 2001: Evaluation of Polar MM5 simulations of Greenland's atmospheric circulation. J. Geophys. Res., 106, 33,867–33,890.

Lazzara, M.A., G. Langbauer, K.W. Manning, R. Redinger, M.W. Seefeldt, R. Vehorn, and T. Yoksas, 2006: The Antarctic internet data distribution (Antarctic-IDD) system.  22nd Int’l. Conf. on Interactive Information Processing Systems for Meteorology, Oceanography, and Hydrology.  Amer. Metor. Soc., Atlanta, GA.

Monaghan, A.J., D.H. Bromwich, H. Wei, A.M. Cayette, J.G. Powers, Y.-H. Kuo, and M.A. Lazzara, 2003: Performance of weather forecast models in the rescue of Dr. Ronald Shemenski from South Pole in April 2002.  Wea. Forecasting, 18, 142–160.

Powers, J.G., K.W. Manning, Y.-H. Kuo, and D.H. Bromwich, 2003: Mesoscale modeling over Antarctica: The Antarctic Mesoscale Prediction System (AMPS).  Bull. Amer. Meteor. Soc., 84, 1533–1546.

Powers, J.G., and K.W. Manning, 2007: Polar WRF testing in Antarctica.  8th WRF Users' Workshop, National Center for Atmos. Res., Boulder, CO.

Skamarock, W.C., J.B. Klemp, J. Dudhia, D.O. Gill, D.M. Barker, W. Wang, and J.G. Powers, 2005: A description of the Advanced Research WRF, Version 2.  NCAR Tech. Note., NCAR/TN-468+STR, 88 pp.  [Available from UCAR Communications, P.O. Box 3000, Boulder, CO 80307]