Regional wet-season failure

A new post on the NASA Earth Observatory website reveals the regional extent of precipitation deficit partway through the 2019 long rains. The NASA soil moisture anomaly map for April (above, from MODIS) depicts a large anomaly extending into northern Tanzania. [The Kenya-Tanzania border jogs around the mountain just below the 'KENYA' label on the image above.]

The seasonal snowcover situation on Kilimanjaro is discussed below; the extent did not increase during May.

Included in the EO article are some helpful references detailing the human impact of this developing East Africa drought.

Dry-season forecast: above-average ablation

The long rains (Masika) of 2019 are concluding with virtually no snow accumulation on Kilimanjaro glaciers, in stark contrast to last year's long-rain season - demonstrating the extreme interannual variability of precipitation at the summit.

The Sentinel-2 image above from 2 days ago (14 May) reveals a largely snow-free crater. Small areas of last year's snow persist (e.g., east of the Northern Icefield, adjacent to the Furtwängler Glacier). Elsewhere, only a dusting of snow can be seen on Kibo's south side - which not coincidentally spans the elevation range and azimuth of remnant glaciers there! (Very preliminary analysis suggests that the responsible snowfall event was somewhat more extensive, yet we know that at this SSW sector of the mountain, convection enhances snowfall and clouds reduce ablation.)

During the long rains last year - extending from 27 February until this date (16 May) - net accumulation of snow on the Northern Icefield was over one meter (as discussed here). Contrast this with 2019 long rain accumulation, shown in the figure below (blue line); prior to the minor event last week the AWS recorded a net lowering (ablation) of over 30 cm. Additional long rain snowfall may still occur this year, however, the long rains rarely extend into June at the summit.

Absent a major event bringing sufficient snow to reduce solar radiation penetration (e.g., 30-50 cm), the forthcoming extended dry season will probably begin with a snow-free crater. As a result, ablation of both horizontal and vertical glacier surfaces is likely to be dramatic in the months ahead.

(The timelapse image below provides a perspective on summit snowcover since early August of 2018. Within the crater, note the persistence of long-rain accumulation through the dry season, and the ephemeral nature of spatially-extensive-but-thin accumulation during the period February to April 2019.)

South-side glacier changes

The images above depict how Kibo's south-side glaciers have changed in area and thickness over the past 13 years. Thanks to Catrin Stadelmann and her advisor Thomas Mölg for sharing them (Friedrich-Alexander-University (FAU) in Erlangen-Nürnberg, Germany).

Note that both snow and ice are visible in these images, particularly the lower one (Jan. 2019). The largest glacier shown is the Kersten, which Dr. Mölg has published extensively on. Shortly after 2006, a gap began widening between the upper ice and the steeper slope portion of the Kersten. Similarly, a gap also formed between the Kersten and the Decken Glacier to the east (right-hand side of the Kersten). The Decken has gradually narrowed, and very little ice remains today. On the Kersten's left (western) side, remains of the Heim Glacier are visible in 2006; the lower-most extent of this glacier was still present by 2019, surrounded by considerable transient snowcover in the image above.

2018 long rains review

Nature does not follow calendars... but yesterday marks the typical beginning of the northern Tanzania "long rains season". Coincidentally, the long rains last year (2018) began precisely on March 1st (see earlier posts, beginning with this one). Images acquired on the day prior, both last year and this year, are shown above (Sentinel-2 L1C).

Despite partial cloud cover, both images depict limited snowcover other than on glaciers (e.g., north-facing side of the Uhuru Peak summit ridge, the crater's south rim). Most important to this discussion is that the summit crater (approximately circled) is largely free of snow. The 2019 image shows some recent snow on the northern flanks, which was present to a lesser extent on mid-February images (not shown).

Even limited snowcover at the end of February (images above) is in stark contrast to the same time in 2000. Note in our prior post (link) that a snow shovel is visible; don't be fooled by this, for there was neither snow nor firn anywhere on the glacier or within the crater. Indeed, this was an exceptionally dry period which continued through the long rains of 2000 (see figure below; red line); March through May snowfall that year totaled only 26 cm, the least of any long-rain season in our period of record.

As the long rains concluded in mid-May last year, snowcover on the mountain was extensive - as depicted in a Sentinel image from the 29th (below). Indeed, the daily snowfall total at the AWS for the 2018 long rains was double the 19-year average, resulting in more than 1 m of net accumulation. The graph below shows how anomalous this accumulation (thick blue line; 2001-2017 as thin blue lines, 2000 in red).

With even an average short rains last year (typically Nov-Dec), could crater snowcover have persisted until these next, 2019 long rains? Quite possibly! Instead, the 2018 short rains included just 2-3 minor accumulation events, plus early snowfall during our late October fieldwork (link), for a total accumulation of less than 20 cm. By Christmas, crater snowcover was patchy. Then, despite a mid-January event, ablation predominated; by yesterday (see above) the crater was largely snow-free.

In contrast, the Northern Icefield surface at the AWS gained mass over the past 12 months, increasing in height by nearly 50 cm. Higher reflectivity and less re-radiated longwave energy from below (i.e., ice vs. dark volcanic soil) are among the factors.

In summary, the extent of 2018 accumulation and it's persistence demonstrates the sensitive balance of processes governing Kilimanjaro's summit glaciers. If seasonal snowcover does ever persist in the crater through an entire year, retention will be easier the next year and subsequently become even easier. This idea is explored in Kaser et al. (2010), and 2018 observations strengthen the argument!

19 years on the Northern Icefield

This week marks 19 years since AWS measurements began on Kilimanjaro's Northern Icefield (NIF). With enthusiastic help from our Tanzanian crew, Mathias Vuille and I installed a tower into the ice and connected the electronics. Remarkably, the same datalogger continues measurement and control functions, and the same solar panels continue to provide power. Most sensors have been swapped out for recalibration or replacement, yet the original barometric pressure sensor continues reliable measurements every hour.

Ice ablation since 2000 has substantially reduce the areal extent of all glaciers on the mountain. However, this portion of the NIF has "only" thinned by ~5 meters, because the low surface gradient retards meltwater runoff - which then refreezes in place as superimposed ice. Other portions of the NIF, and other glaciers, have thinned more dramatically. For example, ice no longer remains at February 2000 drill sites on the Furtwängler and Decken Glaciers, which were 9.5 and ~20 m thick at the time (respectively).

Asymmetric snowfall [updated]

The pair of Sentinel-2 images above demonstrate an interesting asymmetry in snowfall, visible despite partial cloud cover. These images, acquired 5 days apart on 14 and 19 January, are closely registered and show the Northern Icefield AWS location (click image to enlarge).

The southern slopes and south side of the crater appear not to have gained any accumulation through the 5-day interval. Although slopes to the west are difficult to resolve through the clouds, new snow on the north and northeast flanks is readily apparent above ~4,800 m elevation. Snow also accumulated just south of the Northern Icefield, on the crater's west side.

AWS measurements are still being processed for this interval, which will reveal snowfall timing. In the meantime, the website "earth" allows weather conditions during this period to be visualized. The 17th appears a likely time for this snowfall pattern to have developed. Winds were light and humidity was high at 500 hPa, while a bit lower in the free atmosphere (700 hPa) winds were from just east of north, and humidity was high.

This post will be updated as AWS data and the next S-2 image become available.

[UPDATE 01/25, 2/4:  AWS data from the Northern Icefield (via telemetry) reveal the difficulty of documenting subtle climate features on a large mountain, using measurements at one location. In this case, only 3.5 cm of snow accumulation was recorded over the 5-day interval between images (above). Despite use of 2 sensors, 3 m apart, the timing of minor snowfall events cannot be precisely established from the 4-hourly satellite data, possibly due in part to wind redistribution of snowfall. Once hourly snow measurements and other data are recovered from on-site storage (e.g., solar radiation, wind speed), we may be able to better resolve snowfall timing.

A best guess from the AWS measurements on snowfall timing between these images would be the 18th. Supporting this is the lingering presence of accumulating snow at ~4,800 m, which is unlikely to persist more than a day or two - especially at this time of year. However, collaborator observations suggests that widespread snowfall on the northern slopes occurred a bit earlier. For 14-17 Jan., they report heavy rain at ~3,500 m on the Shira Plateau and to the northwest of Kibo.

Additional satellite images are available from the 9th and 24th of January. The image from yesterday (shown below) shows little change in snowcover at high elevations, consistent with AWS data, yet ablation of snow from the northern slopes.

The image from 9 January (not shown) reveals nearly-uniform cloud cover over the mountain. According to AWS data, snowfall was just getting underway at the time, and resulted in a whopping 24 cm of new snow at the summit by the 14th. This is a relatively large snowfall event for the Northern Icefield, of similar magnitude as the late-October event we experienced (see prior post, below). These two are the largest events since the 2018 long rains. Measurements at the AWS suggest most of this snowfall occurred on the 10th or 11th, 3-4 days prior to the image from the 14th (above) - plenty of time for ablation of most new snow from the slopes.]

Wild weather on the mountain

Fieldwork is an essential component of climate and glacier research, providing basic measurements as well as a foundation for theoretical and modeling studies. Yes, fieldwork can be tremendously fun, sometimes even yielding unexpected discoveries – yet it can also be difficult and dangerous. A successful fieldwork campaign requires alignment of numerous components and factors; some of these we can control, and others we must manage. For both categories, past experience and planning is helpful. Sometimes though, the outcome also requires good luck.

After 20 prior trips to my AWS on Kilimanjaro’s Northern Icefield, plans for October fieldwork came together within only a couple weeks. Telemetry of data revealed technical problems which could only be resolved by a visit to the station. However, budgetary constraints dictated that this trip would need to be done differently. While recognizing that a lighter and faster approach would reduce the factors we could control, and increase the required management of other factors, we decided that the potential rewards of a brief visit to the station outweighed the risks of this strategy. Supportive and generous collaborators* agreed to make the visit possible.

Our quickly-hatched plan was to acclimatize normally, and then in one day ascend the final 1000 meters, undertake 4-6 hours of work, cross the summit caldera, and descend the other side to rejoin our support team. Past experience on the mountain suggested to all involved that the concept was reasonable –given just a few hours of reasonable weather. For this trip, ‘reasonable’ weather meant conditions under which an ascent of the Western Breach was safe (i.e., cold and dry), followed by a few hours at the AWS without heavy snowfall, wind less than ~20 km/hr, and air temperature above -5 °C or so; any sun would be a bonus. Once finished at the station, we were confident about descending in almost any conditions.

Reasonable weather prevailed through our first two days on the mountain, followed by conditions more typical of the wet seasons (e.g., April-May). Warmth, rain, and wind appeared in the forecasts, and on day 3 became our reality on the mountain, conditions increasingly at odds with both our work needs and those required for the safety of all involved. Ascending in wind-driven rain to Arrow Glacier camp below the Western Breach, there was little ambiguity about what we were likely to encounter the following day – which proved accurate, as illustrated below and revealed by telemetry from the AWS. That afternoon a consensus emerged: continuing with our plan would unacceptably compromise safety, and that work at the AWS would almost certainly be impossible.

Disappointing? Absolutely. Station problems remain unresolved, and the 18-year nearly-continuous record may be compromised. In addition, not measuring and documenting the summit glaciers will prevent assessing the response of anomalous accumulation during the 2018 long rains. Furthermore, any compromise on fieldwork goals is disappointing in light of the carbon cost of traveling nearly 30,000 km. However, our decision to retreat was correct, for in contrast to several other groups on the mountain, we all returned safely.

Field scientists must fully exploit observational and quantitative opportunities during every moment in the field, and learn from every experience. This trip provided new insight into the development of weather systems on Kilimanjaro. Valuable photographs and observations of the slope glaciers were obtained, and new understandings were gained through interactions with others on the mountain. More difficult to accept was something we already knew, the false economy of an ambitious undertaking in too-little time. Future fieldwork must allow adequate time to accommodate difficult weather conditions, despite the higher financial cost of extended fieldwork time on Kilimanjaro summit glaciers (e.g., extra Park fees, staff salary premiums). Finally, the experience highlights the value of high-elevation climate and glacier data, which should never be taken for granted.

*Special thanks to Nicolas J. Cullen at University of Otago (New Zealand), and Thomas Mölg at Friedrich-Alexander-University (FAU) in Germany, for their encouragement and vital support!

Fig. 1. Timelapse of clouds over Kibo, 24 Oct. at 13:30 (~1 sec interval). Wind speeds began increasing on the 23rd, and remained high for 3-4 days. Airflow throughout our time on the mountain was from just south of east, as illustrated in Fig. 3 below.

Fig. 2. Kibo on 24 Oct. at 13:30 from near Karanga Camp. Although the mountain is quite snowy for mid-October, snowcover decreased during the days prior, due to rain and a high freezing level (note lack of fresh snow east of the Rebmann Glacier, on right-hand side of image).

Fig. 3. Airflow and relative humidity at 500 hPa over east Africa and Kilimanjaro (green circle), 24 Oct. at 14:00. Cyan color indicates RH above ~95%; 49 km/hr equates to ~30 MPH, not a particularly high windspeed for a mountain summit, but difficult to work in when humidity is high (see riming in Fig. 4). Source: earth.nullschool.net (c) 2018 Cameron Beccario.

Fig. 4. Summit scenes early on 25 Oct., when apparently only 2 Norwegians and their guide reached the top. Photos courtesy of Dismas Kishingo, via Emanuel Mutta of SENE.

Fig. 5. Landsat 8 scene from 28 Oct. at 10:43 local time. Fresh snow on Kibo and Mawenzi accumulated over the prior ~4 days, when AWS data show dropping air temperature and 20+ cm of accumulation.

Fig. 6. The proverbial calm after the storm. Kibo as seen from Moshi, 28 October at 08:00 (just prior to the Landsat image above).