Modelling Glaciers in the ALADIN-HIRLAM Numerical Weather ... · Scan QR code to go to...
1
Modelling Glaciers in the ALADIN-HIRLAM Numerical Weather Prediction System Emily Gleeson* 1 , Ruth Mottram 2 , Kristian Pagh Nielsen 2 , Xiaohua Yang 2 1 Met Éireann, Dublin, Ireland 2 Danish Meteorological Institute, Lyngbyvej 100, Copenhagen, Denmark Model Evaluation: snow cover and albedo Abstract HARMONIE (a canonical configuration of ALADIN-HIRLAM) is a convection permitting non- hydrostatic model that includes the multi-purpose SURFEX surface model. It is developed for high resolution (1-3 km) weather forecasting and used for a number of different regions in Europe and the Mediterranean. The version HCLIM is being developed for regional climate modelling. Here we run it for Greenland. By improving the characterisation of glacier surfaces within SURFEX we show that weather forecast errors over both the Greenland ice sheet and smaller glaciers can be significantly reduced. The improvements also facilitate increasingly accurate ice melt and runoff computations, which are important both for ice surface mass balance estimations and for commercial applications such as hydropower forecasting. These improvements will also benefit forecasting for HARMONIE domains that cover the Svalbard archipelago, the Alps and the Scandinavian mountain glaciers. Future uses of HCLIM for these regions, where accurately characterising glacial terrain will be crucial for climate and glaciological applications, are also expected to benefit from this improvement. Figure 1. Harmonie NWP (numerical weather prediction) domains for Greenland and Iceland: a) is GLB NWP domain for Greenland (also used in development here) b) IGA domain over which model updates will be applied when complete c) PROMICE automatic weather stations around Greenland, QAS_L station is used for model testing. Scan QR code to go to polarportal.or g HARMONIE Set-Up HARMONIE has many different set-up possibilities and can be run with different physical, dynamical and surface schemes. Here we use the DMI GLB operational NWP set-up shown in Figure 2. Future Work This preliminary work is encouraging; we expect to have a working glacier scheme in HARMONIE this year but the following are still missing: • Surface albedo of bare ice, Surface Mass Balance computation; Retention and Refreezing in the snowpack (CROCUS in SURFEX is one solution). Figure 3. Operational (GLB) HARMONIE represents weather over the ice sheet well as shown here - clockwise from top left: 2m temperature, wind speed, cloud cover (note observations are uncertain below 0.2) and relative humidity when compared with QAS_L data, upper 4 plots March, lower 4 plots July 2014. Figure 4 compares skin temperature at QAS_L to output from HARMONIE for the default and modified cases. Figure 6. (left) Comparison of observed and modelled skin temperature compared with measured snow height at QAS_L (right) the albedo in different shortwave bands over same period. Figure 7. (upper left) Total albedo with cloud cover masked out for one day in July, (upper right) over glaciers (lower) MODIS (MOD10A) average over 1 week. HARMONIE total albedo is too low over bare ice SURFEX: Land surface Aladin Dynamics ECMWF IFS cycle 41r1 D95 snow scheme IFS Radiation QAS_L a b c Figure 8. Latent (LHF) and Sensible (SHF) heat fluxes for two days in July 2015, the 7th was a high melt day, the 27th was a relatively low melt day compared with the Greenland melt extent courtesy of Peter Langen, DMI calculated from the HIRLAM-newsnow model available on polarportal.dk. LHF is much higher than the SHF on both days (note different colour scale) and the pattern correlates well with melt extent from HIRLAM. Fausto et al. (2016) have recently questioned model simulations of the different types of heat fluxes during extreme melt events. Further work on this is needed for which both HCLIM and HARMONIE will be useful. Non-radiative fluxes LHF SHF Melt extent MESO-NH: Physics When snow melts away from the glacier surface and bare ice is exposed, modelling results deteriorate, suggesting an inadequate surface parameterisation. We modified HARMONIE to simulate melting over ice by fixing the radiative surface temperature. Modified HARMONIE simulates the melt out of surface snow reasonably well but there is a cold bias in the skin temperature as shown in Figure 4 below. Figure 5. Change in Snow Water Equivalent (kg m -2 ) from (left) 1st April to (right) 31st August 2015. LHF SHF Melt extent 7 th July 2015 27 th July 2015
Transcript of Modelling Glaciers in the ALADIN-HIRLAM Numerical Weather ... · Scan QR code to go to...
Modelling Glaciers in the ALADIN-HIRLAM Numerical Weather Prediction SystemEmily Gleeson*1
,Ruth Mottram2, Kristian Pagh Nielsen2, Xiaohua Yang2
1 Met Éireann, Dublin, Ireland2 Danish Meteorological Institute, Lyngbyvej 100, Copenhagen, Denmark
Model Evaluation: snow cover and albedo
AbstractHARMONIE (a canonical configuration of ALADIN-HIRLAM) is a convection permitting non-hydrostatic model that includes the multi-purpose SURFEX surface model. It is developed for highresolution (1-3 km) weather forecasting and used for a number of different regions in Europe andthe Mediterranean. The version HCLIM is being developed for regional climate modelling. Here werun it for Greenland. By improving the characterisation of glacier surfaces within SURFEX we showthat weather forecast errors over both the Greenland ice sheet and smaller glaciers can besignificantly reduced. The improvements also facilitate increasingly accurate ice melt and runoffcomputations, which are important both for ice surface mass balance estimations and forcommercial applications such as hydropower forecasting. These improvements will also benefitforecasting for HARMONIE domains that cover the Svalbard archipelago, the Alps and theScandinavian mountain glaciers. Future uses of HCLIM for these regions, where accuratelycharacterising glacial terrain will be crucial for climate and glaciological applications, are alsoexpected to benefit from this improvement.
Figure 1. Harmonie NWP (numerical weather prediction) domains forGreenland and Iceland: a) is GLB NWP domain for Greenland (also used indevelopment here) b) IGA domain over which model updates will beapplied when complete c) PROMICE automatic weather stations aroundGreenland, QAS_L station is used for model testing.
Scan QR codeto go to polarportal.org
HARMONIE Set-UpHARMONIE has many different set-up possibilitiesand can be run with different physical, dynamical andsurface schemes. Here we use the DMI GLBoperational NWP set-up shown in Figure 2.
Future WorkThis preliminary work is encouraging; we expect to have a working glacier scheme in HARMONIE this year but the following are still missing:• Surface albedo of bare ice, Surface Mass Balance
computation; Retention and Refreezing in the snowpack (CROCUS in SURFEX is one solution).
Figure 3. Operational (GLB) HARMONIE represents weather over the ice sheet well as shown here -clockwise from top left: 2m temperature, wind speed, cloud cover (note observations are uncertain below 0.2) and relative humidity when compared with QAS_L data, upper 4 plots March, lower 4 plots July 2014.
Figure 4 compares skin temperature at QAS_L to output from HARMONIE for the default and modified cases.
Figure 6. (left) Comparison of observed andmodelled skin temperature compared with measuredsnow height at QAS_L (right) the albedo in differentshortwave bands over same period.
Figure 7. (upper left) Total albedo with cloud cover masked out for one day in July, (upper right) over glaciers (lower) MODIS (MOD10A) average over 1 week. HARMONIE total albedo is too low over bare ice
SURFEX: Land surface
AladinDynamics
ECMWF IFS cycle 41r1D95 snow
scheme
IFS Radiation
QAS_L
ab c
Figure 8. Latent (LHF) and Sensible (SHF) heat fluxesfor two days in July 2015, the 7th was a high melt day,the 27th was a relatively low melt day compared withthe Greenland melt extent courtesy of Peter Langen,DMI calculated from the HIRLAM-newsnow modelavailable on polarportal.dk.
LHF is much higher than the SHF on both days (notedifferent colour scale) and the pattern correlates wellwith melt extent from HIRLAM. Fausto et al. (2016)have recently questioned model simulations of thedifferent types of heat fluxes during extreme meltevents. Further work on this is needed for which bothHCLIM and HARMONIE will be useful.
Non-radiative fluxes
LHF SHF
Melt extent
MESO-NH: Physics
When snow melts away from the glacier surface andbare ice is exposed, modelling results deteriorate,suggesting an inadequate surface parameterisation.
We modified HARMONIE to simulate melting over iceby fixing the radiative surface temperature. ModifiedHARMONIE simulates the melt out of surface snowreasonably well but there is a cold bias in the skintemperature as shown in Figure 4 below.
Figure 5. Change in Snow Water Equivalent (kg m-2) from (left) 1st April to (right) 31st August 2015.
LHF SHF
Melt extent
7th July 2015
27th July 2015
