Reanalyses products in the polar regions

Reanalyses products in thepolar regions

Ian Renfrew (UEA)

Kent Moore (U. Toronto)Ben Harden (WHOI/Sea Education Assoc)

Richard Jones (UEA)

Outline

• Potential advantages of regional reanalyses• Reanalyses performance: 

• The Arctic & the Subpolar Seas• The Antarctic

• Common concerns• Conclusions

Potential advantages 

• Increased model resolution• Bespoke parameterizations of key processes

• Surface exchange (e.g. sea ice)• Atmospheric boundary layer parameterizations

• Cloud parameterizations• Bespoke data assimilation? 

• e.g. sea ice thickness, snow depth, etc

Model Resolution• Many mesoscale weather systems in the subpolar seas

Tip jets Barrier winds

Katabatic flows Polar lows

• Many mesoscale weather systems in the subpolar seas• However all these systems are mesoscale in nature

(length scales ~100-500km) and so may be under-resolved

Power Spectrum of 10m wind speed near Greenland from the Athena Dataset(Kinter et al 2013 BAMS)

k-5/3 stratified 2D turbulencek-2.2 midlatitude scatterometer

windsIn the mesoscale (100-500km)T1279 spectra ~k-2.2

Skamarock (MWR 2005)Effective horizontal resolution~5x=> ERA-I (T255) has effective resolution ~400km

Model Resolution

• 10-m wind speed in ERAI and ASRv1

• Orographic influences clear• Sheltering & confluence

Model Resolution

Frequency of high wind speed events: 

NE flows: 

NW flows: 

Arctic System Reanalyses• Given the profound changes that are occurring in the

Arctic, there is a pressing need for a high resolution representation of the structure and variability of the Arctic troposphere.

ASR uses PolarWRF forced by ERA-I at boundaries

2 versions exist:ASRv1 30kmASRv2 15km

Arctic System Reanalyses

• ASRv1 (30 km) overall performs similarly to ERA Interim• Slightly better in near‐surface temperature & humidity• Wind speed biases smaller

• ASRv1 (30 km) overall performs similarly to ERA Interim• Wind speed biases smaller• Slightly better in near‐surface temperature & humidity• SW radiation biased high in ASR• Although RMSE comparable to ERAI & correlations bit better• Inadequacies in model physics, e.g. convective and radiation schemes

Topography of Southern Greenland (km) as represented in the ERA-I and ASRv2DMI stations in the region are indicated

Arctic System Reanalyses: Orographic flows

Orographic winds

Time series of 10m wind speed at selected DMI stations (from Moore et al. QJ, in review)

• Winds stronger to the south during first half of observational period (Easterly Tip Jet event B268).

• Winds stronger to the north during 2nd half of observational period (Barrier Wind event B274, B276, B277, B278).

• Observed winds at Tasillaq generally lower then model winds with ERA-I generally having the largest bias. This is due to a topographic sheltering effect that is partially captured by the ASR (Moore et al. GRL 2015).

=> Increase in resolution has positive impact on the representation of the coastal flow in the region.

Orographic winds

Observed and model wind speed profiles during GFDex flight B268 (from Moore et al. QJ, in review)

Narsarsuaq(onshore)

GFDex sonde # 9(jet core)

The ASRv2 is able to better represent the vertical structure of the observed easterly tip jet.

Orographic winds

Open ocean winds

GFDex flights (from Renfrew et al. 2009, QJRMS)

Both reanalyses have a systematic low wind speed bias.No significant difference between the ERA-I and ASRv2.

For offshore flow, resolution is not as significant Parameterization & DA more important?

ERA-I

ASRv2

Central Iceland Sea buoy (from Harden et al. 2015, GRL)

Central Iceland Sea

• ERAI generally compared very well to buoy

• Worse correspondence for Southerly winds compared to northerlies 

• What do high heat flux events look like? 

• How frequently do they occur? 

• Conditionally sample, dependent upon wind direction at buoy 

• Location of composite low pressure centres

• Associated T & heat fluxes• High heat flux events from NW quadrant

2‐m Temperature Total heat flux

• What do high heat flux events look like? 

• How frequently do they occur? 

• Conditionally sample, dependent upon wind direction at buoy 

• Location of composite low pressure centres

• Associated T & heat fluxes• High heat flux events from NW quadrant

Central Iceland Sea

• Reanalyses able to characterise high surface heat flux events

• Cold‐air outbreaks

Low‐level jets climatology with ASRv1Tuononen et al., 2015, Atmos. Sci. Lett

Mean frequency of LLJ Mean LLJ wind speed (m/s)

Antarctica • No regional reanalyses• Global reanalyses products• Downscaled products (RACMO)• Antarctic Mesoscale Prediction System (AMPS) 

• 30 km outer• 10 km Antarctic• 3.3 km Ross Sea• 3.3 km Ant. Peninsula• 1.1 km McMurdo

• AMPS uses Polar WRF 3.7• SEB modifications over sea ice and permanent ice surfaces 

• Sea‐ice updates during run• Snow modifications• NOT a reanalysis 

Antarctica 

• AMPS, MetUM, RACMO comparison• Significant problems in SW fluxes in all models

• Related to clouds 

Antarctica 

• Global reanalyses products

From: Jones, Renfrew, Orr, Webber, et al. J. Geophys. Res, under review

• ERAI cold bias at these coastal AWS sites

• MERRA T2m poor• Seasonal temperature bias

• Related to SBL

From: Jones, Renfrew, Orr, Webber, et al. J. Geophys. Res, under review

• Cold biases at these coastal AWS sites

• MERRA T2m poor• Seasonal temperature bias

• Related to SBL

From: Jones, Renfrew, Orr, Webber, et al. J. Geophys. Res, under review

• Cold biases at these coastal AWS sites

• MERRA T2m poor• Seasonal temperature bias

• Related to SBL

• Cold biases at coastal Amundsen Sea sites too • MERRA T2m poor again• Radiosondes: off shore   coastal 

Antarctic Winds• Underestimate high winds• Overestimate low winds• Unable to cope with steep & complex coastal orography• Unable to represent low level jets

• Radiosondes: no LLJ  LLJ  

Common concerns in polar regions

• Stable BL poorly represented cold biases around coastal Antartica

• Poor clouds and SEB• ASRv1 has biases

• Treatment of sea ice, snow and clouds needs consideration

• Sharp vertical gradients (in wind, temp or humidity) are common and often poorly represented

• Can act as lids for BL processes, transport, etc• Regional reanalyses may be better?  

Conclusions• Higher model resolution allows better representation of mesoscale features

• Katabatic flows, barrier winds, tip jets, …• Higher correlations, smaller RMSE, regression slopes closer to one 

• But doesn’t aid representation away from complex orography

• And doesn’t solve biases• Bespoke ‘polar’ parameterizations could be of benefit

• Sea‐ice concentration & surface exchange• Clouds, BL and radiation….   

• But evidence incomplete