AWS on Ice

[Joe is a Senior Glacier Hydrologist at the International Centre for Integrated Mountain Development in Kathmandu, Nepal]

True fact: there have been not one but two workshops dedicated specifically to the installation of automatic weather stations (AWS) on glaciers.

Jiten_160508_MG_2806

The newly-installed AWS at Yala Glacier. We didn’t get these views when we did the work. (Photo credit: Jitendra Bajracharya)

One of the biggest unknowns in how glaciers will respond to climate change are the meteorological conditions and melt rates at the glacier surface, and how these conditions relate to data from standard observation networks and/or climate reanalysis products. But setting up precise sensors on a surface that can move, melt, and be buried by snow – sometimes all of these in the same day – is a big challenge. Unfortunately, for all challenges (including drinking milk upside down through a straw) you either learn by experience (AKA “mistakes”), or you learn from the experiences of others. For some reason I’ve tended to go with the former.

Our recent AWS installation at Yala Glacier is another attempt to obtain a year-long record of meteorological conditions at 5350 m in the Himalayas. At this altitude, temperatures are rarely above zero and the melt of snow or ice is basically controlled by the radiation balance at the surface (see below for a more technical discussion). So our station will record radiation received and emitted or reflected by the surface, air temperature and relative humidity, wind speed and direction, and surface height changes from melt and snowfall.

Experience tells us that ‘floating’ weather stations, such as tripods that simply sit on the top of the surface, don’t work so well on glaciers. The surface melts down unevenly, the station can be buried and damaged by heavy snowfall, and there is no way to get a record of surface lowering: the surface height sensor needs to be mounted at a fixed height in order to get information that makes any sense.

DSC01549

Our first attempt to measure conditions on Yala Glacier, found after typhoon HudHud in October 2014.

For the new station, we used a slick tower design that can be built up in the field (full credit to Alex Jarosch and Faron Anslow; tower recipe below or see P. 52-55 here). Essentially, we connect three 2.0 m aluminum pipes vertically to make a 6.0 m tall triangular structure. Horizontal supports brace the top 2.0 m of the tower, and the bottom 4.0 m of each leg is drilled in to the ice. If you’re going to try this at home, don’t forget to stick small plastic caps on the bottom of the pipes that go in the ice. Without these, the weight of the tower would be supported on a very small surface area and it would melt into the ice – probably due to heat conduction through the aluminum. If the tower sinks into the ice during the experiment, the surface height measurements are meaningless. (Thanks, experience!)

Once the base and the tower are installed and leveled, the waterproof enclosure (which contains the battery, solar charge controller, and the datalogger) and all sensors were mounted to the tower. In the time-lapse animation shown below, you can see the clouds rolling up and down over us as we mount the sensors. In response, we shed layers and then put them back on, because the thickness of the cloud layer really affects the ‘felt’ temperature at the surface (you should really read the technical explanation below). Air temperatures during the setup hovered around 0C.

 

DCIM100GOPRO

The tower from above: logger box and temperature/humidity sensor is on the left, wind sensor is top right, and it looks like three people are required to mount the net radiometer (which measures shortwave and longwave radiation – really, there is no longer an excuse to not read just a little bit more in depth below).

The full installation took only half a day, and we were back drinking tea in camp by mid-afternoon (though thankfully not upside down and with straws). But getting the equipment and the tower components up there literally took a small army. We have nothing but huge gratitude and respect for Dawa Sherpa and Ngawang Sherpa who helped haul everything up the glacier, and to all the trekking agency staff who carried everything up from the trailhead at 1600 m to the basecamp at 5000 m.

[Thanks to Maxime Litt and Desiree Treichler for their help in the field, but also for the pre-field testing and programming. This is a critical step in the recipe.]

 

Glacier Station Recipe

  • 9 x 2.0 m aluminum pipe (48.25 mm OD)
  • 9 x 0.50 m aluminum pipe (48.25 mm OD)
  • 3 internal pipe connectors
  • 3 external pipe connectors
  • 3 plastic cap ends (large corks also approved)
  • 18 x 90 degree joints (48.25 mm OD)
  • Ice auger (4-5 m)
  • Plumber’s level
  • Tools
  • AWS components and all mounting hardware (!)
  • Patience
  • Reasonably good weather
  • Preparation, preparation, preparation

Radiation Balance Details

The net radiation at the surface (Q*) can be calculated from incoming and outgoing shortwave and longwave radiation:

Q* = Sin – Sout + Lin – Lout

Shortwave radiation comes from the sun: its highest at solar noon, and zero at night. But the amount of radiation reaching the surface depends on clouds and the atmospheric conditions, and the amount of shortwave radiation absorbed at the surface depends on the reflectivity (or albedo) of the surface. Brighter surfaces reflect more radiation, and have a higher albedo, which means less energy available for melt.

Longwave radiation is a mainly function of temperature: incoming longwave radiation is emitted by the atmosphere, and the earth’s surface emits longwave radiation upwards. Temperatures near the surface will be warmer on cloudy nights because the clouds both (a) emit greater longwave radiation towards the surface than a clear sky and (b) trap some of the longwave radiation emitted by the ground. Incoming longwave radiation is also a function of water vapour in the atmosphere, which affects the temperature profile.

 

Contribution to EGU General Assembly 2016

This year’s EGU General Assembly has passed and we presented a number of topics in 4 different sessions.

In a session with numerous outstanding talks on Mountain Climates on Wednesday, Joseph Shea presented initial results from an analysis of glaciological and hydrological sensitivities in modeling in the Hindukush Himalaya region.

On Thursday, Walter Immerzeel opened the session on debris covered glaciers with a solicited talk, summarizing our recent efforts in quantifying mass changes on the debris covered glaciers. During the same row of talks Jakob Steiner looked at the spatial and temporal evolution of ice cliffs and lakes in the Langtang catchment in the recent decade.

The round of talks was followed by a poster session with 20 submissions specifically on the topic of debris covered glaciers, underlining the increased attention the issue has received recently. Philip Kraaijenbrink presented his work on the monitoring of glaciers using unmanned aerial vehicles. Pascal Buri presented some progress on the distributed modeling of ice cliff backwasting in the catchment. Evan Miles provided insight into the temporal change of supraglacial lakes and how to extract this information from Landsat imagery. He also presented recent work on deriving surface roughness oft he glacier with photogrammetric analysis. The contribution to this session was rounded off by Pascal Egli’s work on deriving debris thickness from remotely sensed temperature data. Jakob Steiner also presented some insights from analysis of the development of a large alluvial fan in the Langtang catchment in a session on sediment transport in pro-glacial environments.

The week ended with Tobias Bolch presenting Silvan Ragettli’s recent work on mass balance change in the Langtang catchment during the last decade.

 

Contribution to Alpine Glaciology Meeting

At the recent Alpine Glaciology Meeting (AGM) 2016 held in Munich we presented some of the ongoing work in the Langtang catchment.


Pascal Egli presented his work on the reconstruction of debris thickness on debris covered glaciers from thermal satellite images and meteorological data via an energy balance approach. We used three previously published methods based on the energy balance to compute debris thickness maps of Lirung Glacier, assess their performance and determined their sensitivity to several input parameters. We developed a new time-integrating energy balance model with which we intend to obtain more accurate estimates of debris thickness by accounting for the heat transport and storage in the debris layer. Debris thickness reconstruction is essential for the determination of melt rates of debris covered glaciers in the Himalayas.


Figure 1: Modelled debris thickness on Lirung Glacier using an energy-balance approach, with distributed surface temperature data from a Landsat 8 image from 23rd October 2015.


 

Pascal Buri presented a new 3D-modelling approach of ice cliff backwasting. Whereas some supraglacial cliffs remain stable in terms of shape over the melting season, some cliffs flatten considerably and disappear. We tested our physically-based model on selected cliffs on Lirung Glacier in order to simulate cliff evolution over one season. Both atmospheric melt and the interaction with adjacent lakes and debris slopes are implemented. This allows us to dynamically simulate the cliff geometry applying monthly geometric corrections. Simulated volume losses and melt rates for each cliff roughly agree with a TIN-approach used for validation but give also further insight into the inter-annual variability of melt processes and mechanisms behind cliff dynamics.

Figure 2: Simulated ice cliff outlines from May to October for cliffs 1-4. Observed outlines from UAV orthoimage in the background.


 
Jakob Steiner talked about traces of the Gorkha earthquake from April 2015 left on the cryosphere in the catchment. We could observe that the ice released mainly came from the top ridges pointing at topographic amplification playing a possibly significant role in this case. This could help to assess which hanging ice seracs pose a danger in future earthquakes. The massive deposits on the glacier tongues of two debris covered glaciers in the catchment, could provide a chance to observe how the heterogeneous debris cover develops over time.

Figure 3: The debris surface on Lirung glacier is nearly level on the upper parts after the earthquake. Observing how this surface changes into the hummocky terrain normally observed on such glaciers could provide insights into the general development of debris cover.

Special issue of Annals of Glaciology

In the recent special issue of the Annals of Glaciology on ‘Glaciology in High Mountain Asia’ our research team contributed four papers focused on the surface of debris-covered Lirung Glacier, located in the Langtang Valley in the Nepalese Himalaya.


Variation in environmental lapse rate

In a study of the distributed air and surface temperature of the debris surface (link) we propose to adapt the normally used environmental lapse rate when using off-glacier data for an energy balance of the glacier, as the debris surface heats up much more than the surrounding environment (Fig. 1). We also show that the lapse rate has a diurnal cycle and the relation between air and surface temperature changes between day and night and for the dry and wet season. This has implications for using surface temperature data acquired from satellite products used to determine local air temperature over a debris covered glacier.

An additional paper looking at the wider catchment (link) shows a strong seasonality of the environmental lapse rate. Lateral variability at transects across valley is high and dominated by aspect, with south-facing sites being warmer than north-facing sites and deviations from the fitted lapse rates of up to several degrees.

Steiner J. and Pellicciotti F. (2016), On the variability of air temperature over a debris-covered glacier, Nepalese Himalaya. Ann. Glaciol., 57(71), (doi: 10.3189/2016AoG71A066)
Heynen M, Miles E, Ragettli S, Buri P, Immerzeel W and Pellicciotti F (2016) Air temperature variability in a high elevation Himalayan catchment. Ann. Glaciol., 57(71) (doi: 10.3189/ 2016AoG71A076)

Figure 1

Figure 1: Mean air temperatures measured at the T-Loggers in the daytime (red) and at night-time (blue) in all three seasons (a–c) and from 2012 to 2014 (top to bottom). The lines show the temperatures determined with the ELR from the off-glacier AWS Kyanjing. The plots at the bottom show absolute deviation of each value from the ELR.


Distributed modelling of ice cliffs

Investigating ice cliffs further (link) we developed a first distributed model of cliff backwasting for two cliffs. The physically-based model includes an improved representation of shortwave and longwave radiation, and their interplay with the glacier topography. Diffuse radiation is the major shortwave component, as the direct component is strongly reduced through self-shading. Incoming longwave radiation is higher than the total incoming shortwave flux, due to radiation emitted by the surrounding terrain. We could show a considerably high variability in melt rates across the cliff surface and ice cliff melt to be more than 10 times higher than melt under debris per unit area (Fig. 2).

Buri P, Pellicciotti F, Steiner J, Miles E, Reid T and Immerzeel W (2016) A grid-based model of backwasting of supraglacial ice cliffs over debris-covered glaciers. Ann. Glaciol. (doi: 10.3189/ 2016AoG71A059)
Figure 2: Distribution of daily melt rate (cm w.e.) for cliff 1 (left, NW-aspect) and cliff 2 (right, NE-aspect) for pre-monsoon (PRM), monsoon (M) and post-monsoon (POM) periods.

Figure 2: Distribution of daily melt rate (cm w.e.) for cliff 1 (left, NW-aspect) and cliff 2 (right, NE-aspect) for pre-monsoon (PRM), monsoon (M) and post-monsoon (POM) periods.


Modelling of supraglacial lakes

In another study (link) we focus on supraglacial lakes. This research advances previous efforts to develop a model of mass and energy balance for supraglacial ponds by applying a free-convection approach to account for energy exchanges at the subaqueous bare-ice surfaces (Fig. 3). We develop the model using field data from a pond on Lirung Glacier, Nepal, that was monitored during the 2013 and 2014 monsoon periods. Supraglacial ponds efficiently convey atmospheric energy to the glacier’s interior and rapidly promote the downwasting process.

Miles E, Pellicciotti F, Willis I, Steiner J, Buri P and Arnold N (2016) Refined energy-balance modelling of a supraglacial pond, Langtang Khola, Nepal. Ann. Glaciol., 57(71), 29–40 (doi: 10.3189/2016AoG71A421)

A conceptual diagram of the energy balance developed for supraglacial lakes.

Figure 3: A conceptual diagram of the energy balance developed for supraglacial lakes.


Seasonal surface velocities

The last study (link) reveals seasonal differences in velocities of Lirung Glacier in detail using cross-correlation feature tracking of imagery from an unmanned aerial vehicle (UAV).The glacier has considerable spatial and seasonal differences in surface velocity, with maximum summer and winter velocities 6 and 2.5 m a–1, respectively, in the upper part of the tongue, while the lower part is nearly stagnant (Fig. 4). UAVs have great potential to quantify seasonal and annual variations in flow and can help to further our understanding of debris-covered glaciers.

Kraaijenbrink PDA, Meijer SW, Shea JM, Pellicciotti F, Jong SM, Immerzeel WW (2016). Seasonal surface velocities of a Himalayan glacier derived by automated correlation of unmanned aerial vehicle imagery, Ann. Glaciol., 57(71), 103-113 (doi: 10.3189/2016AoG71A072)
Surface velocity and flow direction

Figure 4: Surface velocity and flow direction obtained by noise-filtered frequency cross-correlation for the summer (left) and winter (middle) period.


annals-of-glaciology-logo
1 2 3 4 5 9