Air temperature series [updated]!

In June of 2007 the Quelccaya AWS was supplemented by instrumentation compatible with NOAA’s Climate Reference Network (CRN). As nicely documented on the CRN website, several highly-accurate air temperature sensors housed within a continuously-ventilated radiation shield provide what is likely the best air temperature measurement possible in the extreme environment of the ice cap. (The link above now has comprehensive access to all CRN publications.)

Further information on Quelccaya temperature measurements is available here, along with access to the first 3 years of measurements.

Processing of the full 2007-13 temperature record from these sensors has just been completed, and access will be provided early next month. This record was assembled from 5-min averages based on 3-4 PRT sensors, quality controlled following CRN protocols (see Palecki & Groisman, 2011).

[UPDATE 3/25:  A complete record of quality-controlled hourly air temperatures for July 2007-June 2009 are available here, as well as the accompanying metadata. The full 2007-13 period has been processed, yet these are not yet available due to insufficient fan speed during early-morning hours through a portion of this period. Maximum daily temperatures were largely unaffected and this series will be available soon. Thanks for your patience.]

A sneak preview of one aspect of the new record is provided above, illustrating how melt energy is far from uniform through the year, or even through the accumulation season. The bar chart shows the number of hours each day in which the hourly mean temperature was above freezing, for each of the 6 years; also shown (as symbols) are the total number of hours above freezing - over 6 years - for each date. Although air temperature rises above freezing during only 0.42% of all hours, the associated melt is having a profound impact on the ice core record!

Snowy Vilcanota

This is the most extensive snowcover I've seen on a Landsat image of the Quelccaya area. Looking through a layer of thin clouds, Sibinacocha is barely visible in the upper left; the yellow circle tightly circumscribes the ice cap, which isn't visible here. This image was acquired on 26 October.

A context for the event is provided by snowfall telemetry from the summit AWS, via GOES telemetry. The figure below shows daily snow surface height at the station, with each year in a different color. The lowest horizontal dotted line is the zero reference, or the lowest height reached each year (defined here as June-May), and the upper horizontal line is 1 meter of accumulation. Accumulation for 2013-14 is shown in white, depicting a fairly-average surface height minimum date in early October. The snowfall event captured in the Landsat image occurred during the straight, steep increase in the white line, and accumulation by month's end was 2-3 times the mean (or median) since 2004. [Note that the straightness of the line for 8 days cannot be taken to indicate linear accumulation. Rather, heavy snowfall resulted in weak transmissions and intermittent loss of telemetry.]

Below is another Cordillera Vilcanota scene, this one acquired 11 November 2013 by Landsat 8, after the snowfall event (Landsat Scene Identifier:  LC80030702013315LGN00). Snowcover off the glaciers is somewhat diminished, because higher air temperature and increased clouds make this time of year warmer - and snow melts. Snowcover shown here is primarily at elevations above ~5,000 m. Compare this with the seasonal cycle for 1998 depicted in the right-hand margin of this page, or available here.

Detail of the ice cap and surrounding terrain is shown in the final image, cropped from the same scene.

Failures, breaks, and solutions

During almost every venture into the field, something inevitably breaks. Try as we may to bring the most appropriate and robust gear with us, and to use it carefully, equipment fails during fieldwork in extreme environments for a variety of reasons: (a) it isn't quite up to what we ask of it, (b) it is simply worn out, and/or (c) it fails due to unforeseen stresses - such as getting dropped.

One of the challenges of gear problems during fieldwork is... solving them! Such tasks span the full spectrum from exasperating to fun, yet their outcomes influence the extent to which fieldwork objectives are met. A few approaches for mitigating gear and equipment failures include: (a) redundancy, in which multiple methods are planned - or equipment brought along - to accomplish the same task, (b) repair, usually requiring supplies and tools, and (c) substitution of a different piece of gear or a different approach.

During fieldwork on Quelccaya in July we were presented with an unusual number of equipment challenges, minor and major, solvable and not. Upon return we pondered, investigated, and resolved them in a variety of ways. Below are our 2013 gear challenges and their solutions or outcomes, in no particular order. In cases where the original manufacturer was contacted, note the variety of support (customer service) we received!

• After more than a decade of faithful service, a Black Diamond Gemini headlamp must have been damaged during travel, for it was unresponsive when needed the first night at base camp. Neither changing batteries nor disassembling it revealed the problem, which may have been a broken wire within the insulation. A spare headlamp was borrowed from one of our guides, and ironically the Gemini was found on within a duffel late in the trip. The headlamp continues to function fine, but has been replaced and is now relegated to use closer to home.
• An apparently faulty switch on one of the dual-frequency GPS receivers resulted in loss of some data from one of the base stations. This well-used unit has only LED lights to indicate functionality. Suspecting a potential power problem when the unit was deployed, we bypassed all connections between the GPS and power supply. This initially appeared to solve the problem, which we later learned did not. Although data were lost spanning several days, power was continuous through the most critical period when multiple base stations operated.
  
• Continuing with power problems, a "powergorilla" portable power supply station (Powertraveller, Ltd.) came along to supplement aging laptop batteries. Although the unit had worked well on several previous trips within the past 2 years, it appeared dead upon plugging in the laptop. Strangely, it appeared to take a charge just fine via a "solargorilla," yet when disconnected it had no power. Powertraveller was contacted about the problem and agreed to replace the unit, despite expiration of the warranty. A new powergorilla will power a laptop and recharge mobile phone batteries on Kilimanjaro, later this month!
• At the AWS, three problems developed in the year of autonomous operation. One was the complete failure of the transducer within one of the station's Campbell Scientific SR50 Sonic Ranging Sensors, a relatively uncommon event. These transducers are replaced annually, both to prevent such failures and to yield the cleanest possible measurements; without replacement however, the transducers sometimes last years. When one failed on Quelccaya in mid-May this year, 55 days before our field visit, almost all data were lost. This is why 2 sensors are used on all UMass stations, and replacement sensors with new transducers and desiccant are brought in each year, following careful lab testing. 
  
• A second sensor problem at the AWS involved one of the Rotronic MP101A temperature and humidity probes. This was not a new issue and the equipment is not to blame, but is likely due to a combination of the technology used, intermittent ventilation of the radiation shield, and a relatively humid environment during the Quelccaya wet season. The probe which develops problems has been housed within a radiation shield which is intermittently ventilated by fan (2 of every 10 minutes), and under certain conditions ice crystals are hypothesized to grow during ventilation and then melt during the subsequent unventilated 8 minutes. Despite microscopic analysis of several probes by both Rotronic and Vaisala - which clearly reveals damage to the thin-film polymer - the problem with the humidity sensor has never been satisfactorily explained. So, during this field season the probe and shield were both eliminated!
  
• The third AWS problem in 2013 was due to the fan motor on the shield referenced above, which allowed measurement of the fan revolutions. This ceased functioning at the end of March. The decision to remove the shield was not difficult, due to known radiation-loading errors with this shield design.
  
• A new batch of "harsh-environment" cable ties was purchased for this fieldwork, in an effort to economize on well-proven Tefzel ties. In every attempted installation, the tooth which allows only one-way motion immediately broke; they may have been a defective batch. McMaster-Carr promptly refunded the cost of these.
• To measure snowpit stratigraphy and sample depth, we have been using a Lufkin brand tape measure with a powerful spring to retract the tape into a housing. When blowing snow accumulated in the pit and buried the tape cartridge, a bit-too-much pulling caused the tape to detach, creating 8 meters of chaos. Replacing the tape was not possible in the field or lab, and an e-mail to Lufkin was unanswered.
  
• Lastly, at some point during travel to or from Cusco the airlines broke the frame of a Mountain Hardwear rolling duffel (Juggernaut) - one which had endured numerous prior trips. This has always been a crucial piece of luggage, as it offers more protection for items than a soft duffel and rolls exceptionally well. Mountain Hardwear inspected the break - which was considered unusual - and quickly provided a new Juggernaut.

For forthcoming fieldwork on Kilimanjaro we will be ready for breakages and failures, as always, but we are hoping for considerably fewer than at Quelccaya. After all, there are always other types of gear issues, such as forgetting and loosing things, even though we try to avoid these as well!

Fieldwork!

We are just back from an ambitious program of fieldwork on the ice cap, including 8 days at the summit. The AWS has been raised and serviced, accumulation has been measured, and we succeeded in the GPS survey of motion/ablation stake positions established in 1983 and 1984. Here are a few images from the trip:

Below illustrates the AWS "as found", after digging out the enclosures and raising them. To the right of center is our 2013 snowpit, and the orange flag marks the position of 1983 ice-core drilling done by Lonnie Thompson (Ohio State).

A cold, clear evening at the summit of Quelccaya Ice Cap (2.5 minute exposure).

Our final morning at the AWS with our Perúvian crew from Vicencio Expeditions (left to right:  Koky Casteñeda, Carsten Braun, Theodoro, Eusebio, Felix Vicencio & Doug Hardy; Dave Chadwell not shown).

Base station QSP-2 following 24 hours of high-frequency position measurements.

Position measurement at a (former) stake location near Qori Kalis outlet glacier. 

Final pre-fieldwork accumulation update

During the month of May, snow depth at Quelccaya summit has increased in about half of all years since 2004. By the beginning of June the dry season has typically begun, and until early August, snowfall is only associated with occasional, relatively-brief winter precipitation events.

For the accumulation season 2012-13, snowdepth on 1 June amounted to 1.70 meters, increased slightly by a mid-May snowfall event. This accumulation season is thus second only to 2009-10 (1.64 m) in terms of low snowfall, and considerably below the median 1 June depth of 2.06 m. In contrast, La Niña year 2007-08 saw the greatest snowdepth on 1 June, at 2.40 m.

This update will be the last on accumulation until after our fieldwork, when among other activities we will measure density and determine the water equivalent accumulation for 2012-13. This will provide a more accurate measure of precipitation, and put the accumulation season into a proper longer-term context. Hopefully, we will encounter easy digging!

Preparing for fieldwork!

A component of our forthcoming 2013 fieldwork will involve high-accuracy GPS measurements, uploading data to Scripps for processing, downloading calculated coordinates, and then re-occupying dozens of points comprising a strain net on the glacier which was established and surveyed in 1983. In preparation, Dave Chadwell & Carsten Braun are in San Diego this week testing GPS receivers, a BGAN terminal, laptops, charging systems, and Iridium phones. Multiple redundancies will hopefully insure that things go smoothly on the ice cap!

1998 snowcover from Landsat 5

Through calendar year 1998, Quelccaya was exceptionally well documented by Landsat imagery. From January into early November the ice cap is visible in almost every scene, as the satellite passed overhead every 16 days; 4 are entirely cloudless. Inspired by the Earth Observatory images of Quelccaya posted recently, collaborator Carsten Braun compiled and registered 16 scenes to demonstrate the annual cycle of accumulation and ablation at Quelccaya. These are shown below as an animated GIF cycling continuously through the 16 images, with the 1998 date shown in the upper-left hand corner.

In a year with one of the strongest recent El Niño events, the 1998 annual cycle of snowfall likely differs somewhat from a typical year. The total annual amount of snow, for example, was almost certainly below normal - although we do not have measurements from the summit that year. Nonetheless, these images nicely demonstrate the seasonality of precipitation at Quelccaya:

• In mid-January, when the wet season was well underway, the 1998 transient snowline is generally only 100-200 m higher than the ice cap margin.
• Two months later when the wet season is typically concluding, most of the glacier appears covered by snow - yet not quite to the margin and not on terrain adjacent to the ice.
• Two April images depict the greatest spatial extent of wet-season snowcover on the glacier, with thin cover on other high-elevation areas.
• By mid-May snowcover had decreased considerably, with snow present only on the glacier - and a transient snowline higher than in the mid-January image. (The 10 May image is shown above; registration issues precluded including it in the sequence below.) Snowline moves slightly higher by mid-June.
• Surprise! 27 June is well into the dry season, yet snowline reaches the lowest elevation for the year (at least to early November 1998). This scene also has the greatest spatial extent of regional snowcover for all of 1998.
  
• By the end of July almost all snow was gone from the landscape again. The transient snowline on the glacier rose continuously through July, August, and September. Very little accumulation for 1997-98 remained by the 15 September image.
• The image of 1 October shows what appears to be thin snowcover entirely blanketing the glacier. Little snow is visible elsewhere, except to the north in this image; the larger, un-cropped scene on this date shows a very different pattern of snow distribution on the landscape than after the winter (dry season) event of 27 June.
  
• By early November the next accumulation season appears to have begun, with deeper snow covering the ice cap and extended slightly beyond the margins.

Snowcover: 2010 and 2013

The previous post compared two Quelccaya satellite images discussed on NASA's Earth Observatory website. Among the many interesting features these show is the transient snowline at the time each image was acquired. Landsat images for any particular location (e.g., Quelccaya) can only be acquired when the operative satellite passes overhead, which has typically been limited to an interval of ~16 days. Useful image frequency is further limited by clouds obscuring the scene. In a recent manuscript submitted to The Cryosphere Discussions, authors Maiana Hanshaw and Bodo Bookhagen tabulate many of the best images that include Quelccaya and the Cordillera Vilcanota, beginning with those from Landsat 2 in 1975.

Our measurements on the ice cap provide a context for the 2010 EO image. The graph below shows how surface height at the summit generally decreases through the dry season. Each year of our measurements is shown in black, and for dry season dates common to all years the mean daily height is blue. Note that height decrease is not linear, which reveals important information about the processes involved!


Also shown (in red) is height through the 2010 dry season, with a pink circle indicating the EO image date. Most of the surface lowering (e.g., ablation) took place after the latest available image that year - and the 2010 dry season was the longest of our 8-year record. Before the dry season began, accumulation for the El Niño wet season 2009-10 was the lowest of our record (until this year), with maximum snowdepth reaching 1.79 m on 12 April. So, with lower than normal snowfall and a prolonged dry season - in which albedo steadily decreases - the snowline likely reached a considerably higher elevation than shown in the 2010 image, acquired two months before the dry season ended.

This year, maximum snowdepth was comparable to 2010 (1.78 m), yet reached a month earlier (18 March). A few snowfall events during April and early May have added mass, yet as of 15 May the surface is dropping below the mid-March height. This year and 2010 are the 2 lowest years of accumulation since our measurements began in 2004 (measured as snow depth, without consideration for density). For the first day of May both years, accumulation was >25 cm below the median depth (not shown in figure).

On Quelccaya this July we will measure density profiles and determine the more important measure of accumulation, which is water equivalent. If our team stays strong and our shovels don't all break, we will attempt to reach what remains of 2009-10 accumulation - assuming the 2010 snowline didn't rise above the summit!

Quelccaya from space

NASA's Earth Observatory "Image of the Day" (IOTD) has just published a pair of Quelccaya images from Landsat 5's Thematic Mapper, taken 12 years apart. Perfect registration and color correction provide a fantastic opportunity to view and explore environmental change:  click here, and then below the images click the shaded button labeled "VIEW IMAGE COMPARISON".  Now note the vertical slider bar in the image center which can be moved from side to side with the mouse to compare the scenes in 1988 and 2010.

To supplement the IOTD images and text which you just opened in another window, the following are some interpretations based on observations at the ice cap. The 2010 NASA image has been annotated with letters to show locations referenced below, and the AWS location is labeled:

A   To the east of the label, a marginal lake had just begun forming here in 1988, which gradually enlarged with ice recession. Note the thin ribbon of green extending north to the bofedal on the 1988 image; this meltwater runoff also provided water to the research camp for many years. Sometime between the 2006 and 2007 dry seasons, the lake abruptly drained subglacially into the lake below the "A" on the image, and then flooded down into the next valley to the south. Water from the glacier no longer flows past the camp, and neither a section of bofedal above the "A" nor the green ribbon are visible on the 2010 image. The bright tan-colored ribbon visible on the 2010 image below the "A" resulted from sudden drainage of the marginal lake discussed above. Water flowed under and beside the glacier into the lake (not even present in 1988) and then sloshed over a bedrock sill and cascaded as a flood to the valley below. Sediments displaced into the bofedal are visible in the image; note also that the dark-blue lake just above the North arrow has increased in size. Several lakes in this valley were cored by Meredith Kelly, Tom Lowell et al. in 2011, and the record they produce will put this event into a Holocene perspective.

B   Just north of the "B" label is Qori Kalis outlet glacier, regularly monitored by the Ohio State group (Henry Brecher & Lonnie Thompson; also see here for an dual-image comparison). With the terminus no longer in contact with the lake, the rate of ice loss due to recession has slowed - yet the glacier is rapidly thinning. On-the-ground panoramic views of the area from 2012 include Qori Kalis (here and in greater detail here) and the margin to the south (here) where on the 2010 image - just 2 years prior - this lake was barely visible.

C   This valley to the northeast of Quelccaya appears to have received increased meltwater runoff since 1988, as evidenced by the extent and saturation of green in the 2010 image (i.e., move slider back and forth to see changes). This illustrates the paradoxical impact of glacier recession on water resources: in areas of increasing melt, more meltwater reaches bofedals in the valleys below, promoting plant growth (typically water-limited here, at ~5,000 m elevation). There are a multitude of other downstream effects as well, on both natural and human systems. At some point as the glaciers continue shrinking, the enhanced runoff will begin to diminish. Note the tremendous ice loss on the ridge north of this valley!

Finally, the timeframe for this comparison must be put into context, as the dramatic changes shown by these two images span only a brief moment in the context of Earth history. Ice extent at Quelccaya has fluctuated in the past, of course, but likely never at the modern rate. For example, radiocarbon dating of plants emerging from beneath the ice near label "B" reveals a retreat rate that is nearly 2 orders of magnitude higher than the rate of advance (Thompson et al., 2013). Adapting to this rapid recession - at Quelccaya, elsewhere in the Andes, and globally - will be a major challenge for humans and all organisms dependent upon glacier meltwater runoff.

2012-13 accumulation update

The 2012-13 wet season on Quelccaya began in earnest ~1 November, when the glacier surface reached its lowest height for the year. A snowy interval at the end of September may have marked the seasonal change in larger-scale circulation, yet this snow all ablated by the end of October and was not preserved.

As of 1 April, net accumulation for the season amounted to ~1.7 meters of snow. This is the least accumulation measured on this date for our 10-year period of record. Only slightly more snowfall had accumulated by this date during the 2009-10 season, one that ended by mid-April and was followed by the greatest ablation we have observed (>0.7 m lowering).

In another month or so the wet-dry season transition will be underway, and we will have a more comprehensive perspective on 2012-13 accumulation. The actual mass addition (water equivalence) will not be known until we are on-site for fieldwork; plans are being made to dig a massive snowpit, for sampling and observations back to 2009-2010 accumulation!

Radiation data!

Measurements of all four radiation terms have just been processed for June 2010 to July 2012 - the period over which a Kipp & Zonen CNR4 operated at the station. This is an exciting milestone, for working with such data from an extreme-environment AWS is not trivial; riming, snowfall, and intense solar radiation all influence measurements and necessitate adjustments. More importantly, radiation balance fluctuations considerably influence mass balance at the site, and processes of ice core development.

So for a quick overview, the lower plot below shows monthly mean values for 2010-11 (left bars, blue line) and for 2011-12 (right bars, dark blue line). Net solar is shown in orange, net longwave is red, and the lines are net all-wave radiation. Note a clear seasonality to both net shortwave and net longwave, and the relative accordance of monthly values for both years. At 14°S latitude, solar irradiance at the top of the atmosphere is most intense in February and least so in June, yet halfway through the atmosphere on the glacier, the seasonal pattern differs for both incoming and net solar. The seasonal cycle of net longwave radiation broadly mirrors that of net solar, with the greatest energy loss during months of high solar gain.

The upper plot is a timeseries of snow ablation and accumulation for the two years, June through May. Plotting the datum for each year relative to the annual minimum surface height at the station highlights differences in the timing of height changes.

The snow plot provides an valuable context to account for the observations above. The dry season on Quelccaya is typically May through August, when incoming solar radiation is relatively low, yet increasing dust concentration lowers albedo - and thus net solar receipt. Clear skies result in dramatically less longwave energy from the atmosphere than is lost from the snow surface. The transitional periods between seasons are very important to the energy balance, as shown by the contrast between a year when the dry season lingered (e.g., 2010 blue line) and the following year (dark blue line) in which a snowfall event buried the dry season surface by mid-September. With continuing clear sky and relatively darker snow in 2010, net solar irradiance averaged roughly 50 W/m^2 higher through September, easily offsetting the greater net longwave loss (due less cloud cover)! Without a substantial snowfall event until the end of November, and incoming solar radiation seasonally increasing rapidly each day, 2010 ablation (blue line) during these months was considerable relative to June-August, and the transient snowline likely reached a high elevation on the glacier. Unfortunately, suitable Landsat scenes for that year have not been identified after mid-September (Hanshaw and Bookhagen, The Cryosphere Discussions, 2013).

During both years shown, snow accumulation was greatest during February. Cloud cover and fresh snow likely account for low values of both net solar and net longwave radiation. Lastly, the snow plot illustrates that the 2011-12 wet season (dark blue line) ended earlier than the prior year (i.e., March), as also indicated by higher net solar and longwave radiation during March.

Also essential in understanding the Quelccaya energy balance are turbulent fluxes of sensible and latent heat. Along with estimating the radiation balance for the entire period of record at the site, efforts to determine these energy fluxes continue. The objective is quantifying the magnitude and timing of mass flux due to melting and meltwater percolation, which profoundly influences all aspects of the ice core record.

View to the west

An automated camera on the AWS records both the glacier surface evolution and riming of the station. The clip below was assembled from images taken over a 5-month interval, at a time of day which best shows surface texture. Not all days were included here, as the surface isn't visible when rime obscures the camera housing window!

The date stamp on each image allows meteorological conditions to be assessed at the time. For example, drifting snow is evident on 24 October, when wind speed was ~9 m/s; or compare albedo on 25 or 27 October (~0.62) with that on 2 November (0.84). Although not immediately apparent in these images, the 2011 dry season on Quelccaya ended on 9 September, when the glacier surface was ~18 cm lower than that in the first image (10 July).

Accumulation update for 2012 - 2013

Accumulation on Quelccaya began in earnest during mid-November this year, which is relatively late. Two earlier snowfall events - in late September and early November - were followed by settling and ablation. Although these events contributed mass, they did not result in a surface height increase.

With a relatively prolonged 2012 dry season, and possibly melting of these two early snowfall events, the mid-November commencement of accumulation should be easily recognizable in the snowpack during 2013 fieldwork.

For the current accumulation season through yesterday, snow depth on the glacier is now at 1 meter. This is about the average for mid-January. With ENSO-neutral conditions in the tropical Pacific predicted to continue into the dry season at Quelccaya, another meter of accumulation is likely by early April.