Sunday, December 8, 2013

Project Zongo

I'm taking an online course in climate change from the University of Chicago via Coursera (The Science of Climate Change).  David Archer, the professor, is excellent.  The final project (shown below in expanded form if you want to skip to it) was a fairly open ended assignment that amounted to something like get some data and do something interesting with them.  Although there were data available through the course materials, I wanted to do a project specifically on the Zongo Glacier ("Zongo") on Huayna Potosi so I had to search for some data elsewhere because the course materials did not have sufficient temperature records for El Alto which is the nearest weather station to Zongo.

The Zongo Glacer from the summit ridge of Chacaltaya:

Fortunately, I am friends, through climbing, with Jean Emmanuel Sicart who has published many papers on tropical glaciers, including quite a few on Zongo.  He was kind enough to point me in the right direction for acquiring some mass balance data on Zongo from 1991-2010 and also to spend some time explaining the more esoteric (to me anyway) points about the radiative physics of tropical glaciers which involve the interaction between various atmospheric components that feed into the glacier's energy balance such as wind, temperature, relative humidity, clouds, precipitation, and solar radiation.  We'll get back to all that below.

I have spent a lot of time reading blogs and papers about climate over the past few years but I have no formal academic background in math or science.  The last math or science class I took was in high school in 1983.  Although I felt pretty comfortable having a qualitative conversation about most topics in climate science, I had zero experience actually working through the math or using models in any sort of quantitative way.  Through this course, I have learned that it's one thing to watch a lecture or read an article and another, much more difficult, thing to get into the nuts and bolts of the process.

On the Zongo glacier taking in some short wave radiation:

My goals for the class were to:  (1) learn more about climate science; (2) learn some statistics because it seems like you can't understand anything about climate data without a good grasp of statistical concepts; and (3) learn the basics of manipulating data and performing statistical tests using the programing language R.  So this project is an attempt to achieve those three goals, at least at a very basic level.

The project was limited to 500 words and two figures.  I found it nearly impossible to say anything coherent about the subject within these limitations and eventually settled on the simplest thing I could think of which was correlating a nearby temperature record to the annual mass balance of the glacier.

Other Investigations

In addition to the simple project that I submitted, I attempted to correlate average December precipitation on the coastal zone of Brazil at the latitude from where it appears (from Carvalho 2004, Fig. 1) that the prevailing winds circulate over Zongo during the rainy season.  The precipitation data came from the Global Precipitation Climatology Center (GPCC) Landsurface Monitoring Product 2.5 degrees which uses in situ rain gauges to measure precipitation.  The data are available here.  I used 15 2.5 degree grid cells containing between 10 and 17 stations between 5N and 5S latitude and 55W to 42W longitude.

My idea was that, from the literature (e.g Sicart 2005, Sicart 2011), it appears that the timing of the onset of the rainy season, which usually occurs sometime in December, is a major factor in the annual glacier melt.  I would have liked to see if the timing of the annual southernly movement of the Intertropical Convergence Zone (ITCZ, pronounced "itch") was correlated with the annual mass balance but I couldn't find any ready to use data on the ITCZ and the methods described in the literature for locating the ITCZ by identifying the latitudinal median of the precipitation within the ITCZ's annual range was beyond me.  Just to get the ocean precipitation data in a usable form would have required learning a whole additional set of software tools and my plate was already too full.

Anyway, using just the average December precipitation from a region on the coast of Brazil, I found very, very low correlation to Zongo's annual mass balance which either means there is no correlation between December precipitation in that region and the annual mass balance of the glacier, the data from the GPCC product are unreliable, or I just did my analysis wrong.  Perhaps a more sophisticated look at when the precipitation begins in earnest, a larger or different region of study, or using different data might yield a different result.  It's something I might return to now that I am getting better at handling data.

Interestingly, I also attempted to correlate the December average precipitation for the one degree grid cell containing Zongo with its annual mass balance but also found very low correlation.  There was one gauge station in this grid cell but I do not know where it is located.  I was a little surprised by this result but there it is.

Finally, I attempted to correlate the average Southern Oscillation Index (SOI) for each hydrological year (September through August) to the annual (by hydrological year) mass balance of Zongo from 1991-2010.  SOI is a commonly used measure of the El Nino Southern Oscillation (ENSO).  Although the literature indicates there is a correlation between Zongo's mass balance and ENSO (e.g. Rabatel 2013), particularly during large ENSO events, I found very, very low correlation in the data I used.  I did not lag my SOI data which I believe was done in the studies that found correlation so that might be a big flaw in my methodology.  Also, the timing of the changes to the SOI might be more relevant than the yearly average.  In addition, I might have had different results by using MEI, an arguably more comprehensive ENSO index, instead of SOI.  Or again, I might simply have messed up my data analysis.

Although the results showed no correlation, I found the process interesting and, with each analysis, I improved my ability to work with data in R.

Project Zongo

Correlating temperature to glacier melt appears totally obvious at first glance.  Temperature goes up and the glacier melts.  What could be simpler?  Well, tropical glaciers exist only at high elevation where the air is cold, thin and dry so the air is usually pretty cold without a lot of seasonal variability, has little heat content, or ability to absorb/emit long wave radiation which makes the situation more complicated than you might think.

On Zongo at about 5,100m:

What follows is the paper I submitted for the class but with citations to the literature, links to the data used, and additional discussion to try to clarify the energy balance issues that have an impact on melt plus some additional figures to show the work that led to my conclusion.  I have also inserted some photographs where I thought the photograph would help illustrate the concept under discussion.  It's still a work in progress.

Project Zongo


This project examines the correlation between the annual mass balance of the Zongo Glacier ("Zongo") and nearby annual temperature anomalies over the period 1991-2010.  The hypothesis is that, despite the absence of a rising temperature trend, the average annual temperature anomalies will correlate to the annual mass balance changes.  The project concludes that the average annual temperature anomalies are well correlated with the annual changes to the glacier's mass balance because the annual average temperature integrates the underlying complex ablation processes with annual accumulation even though, according to the relevant literature, the ablation (melt/sublimation) processes are not well correlated with short term measurements of surface temperature.  It also concludes that Zongo was out of balance with the climate of the tropical Andes during this period and likely remains so.  

Zongo is a large tropical glacier located in the Bolivian Andes (16S, 68W, elevation 4900-6000 m.s.l) approximately 25 km northeast of El Alto/La Paz.  It lies on the southwestern slope of Huayna Potosi (6088m), possibly the most climbed 6000m peak in the world.  Zongo is an important source of water for hydropower plants that provide electricity to more than 1.5 million residents of El Alto/La Paz.  (Vuille 2013.)  These plants provide approximately 25% of Bolivia's electricity.  During the austral winter dry season, Zongo's runoff irrigates crops in the Altiplano, the most extensive high plateau on Earth outside of Tibet.

A group of climbers heading across the Zongo dam up to the one of the mountain huts on the regular route on Huayna Potosi: 

Another view of Zongo:

Taking the ladder to the base of the Zongo Dam:

Looking down on the Zongo reservoir:

Observations and Methodology

Like virtually all tropical glaciers, Zongo has experienced a negative mass balance over the last few decades.  (IPCC, AR5, Ch.4. here.)  Using data from GLACIOCLIM and linear regression in R, the cumulative mass balance loss from 1991 to 2010 is shown in Figure 1:
r -.96, R squared: .93, F statistic: 241.1, p value: 7.2E-12, no apparent pattern in residuals.  The GLACIOCLIM data are available here.  Subject to certain concerns, such as the small sample size, potential autocorrelation, and the appropriateness of using a linear model, Figure 1 strongly suggests a statistically significant negative trend.  This result is consistent with the literature indicating that the retreat of Zongo over the past three decades is unprecedented since the Little Ice Age with rapid and continuous decline since 1975.  (Soruco, 2009; Rabatel, 2013.)  Models project that Zongo may disappear entirely between 2030 and 2090.

The Chacaltaya Glacier, located just a few kilometers from Zongo but at a lower altitude (c. 5400m) has disappeared entirely.  Here's a photo of the old cable ski tow which now hangs many meters above the bare rock with La Paz in the background:

A weather station on the ridge running from Chacaltaya's summit in the direction of Huayna Potosi with seasonal snowfall that still allows some skiing:

I was unable to find adequate temperature records for Zongo itself for the period in question so I used data from the nearest weather station I could find.  The temperature data from the Berkeley Earth Project for El Alto (25k from Zongo) show slight cooling (not statistically significant, R-squared less than .01) over this period:
As explained at the Berkeley Earth Project's webpage, the data have been adjusted to eliminate known break points due to station siting and other issues so I have fairly high confidence in the data's integrity.  The data are available here.  Figure 3 shows the relationship between annual average temperature anomalies and mass balance, both in hydrological years:
r -.69, R-squared: .44, F-stat: 16.22, p value: .0008.  Figure 3 suggests that there is a strong negative correlation between annual temperature anomaly and mass balance.  Based on an identical R-squared of .44, this result is quite similar to the result in Jeschke 2009 which used average annual temperature for the 2.5 degree grid cell in which Zongo lies.  Nevertheless, it is lower than the r of .76 that Sicart 2008 reported for the period 1996-2006 but Sicart 2008 used temperature data from just outside the glacier so the higher correlation that they found is not surprising.  Because Zongo sits on the eastern edge of the Andes, it is more exposed to cloud cover rising from the Yungas than El Alto which is located further west.  It is possible that the differences in cloud cover would give significantly different annual temperature anomalies at Zongo and El Alto.

Part of the Zongo hydrological measurement system with the Talata glacier in the background:


Glacier mass balance depends on the relationship between ablation (melting and sublimation) and accumulation which, respectively, depend largely on complex radiative energy balances and precipitation.  (Sicart 2005, Sicart 2008, Rabatel 2013.)  At Zongo, the tropical glacier undergoes three distinct seasons:  a rainy (mostly snow over Zongo's elevation range) austral summer season from roughly January-April where increased precipitation offsets fairly high melt rates; a dry and colder winter season where minimal long wave atmospheric emission occurs due to the thin, dry, cloudless skies and therefore there is minimal melting despite large insolation; and a transition period between September and December where peak melting occurs due to decreased albedo, increased potential for insolation when clouds are absent, and increased long wave radiation from high cirrus clouds.  The timing of the onset of the heart of the rainy season is critical to the annual mass balance at Zongo because any delay results in significant melting without countervailing accumulation.  This delay will occur during a strong El Nino. (Rabatel 2013.)  

According to the literature, surface temperature over short time periods is poorly correlated with the ablation of tropical glaciers due to the thin dry air and cold temperatures at elevation compared to mid and high latitude glaciers.  (Sicart 2005, Sicart 2008, Rabatel 2013.) Thus, the degree day melting models which relate surface temperatures over short periods to melt in mid and high-latitude glaciers are inappropriate for use with tropical glaciers on short time scales.  Nevertheless, over longer periods of time, temperature integrates the ablation and accumulation factors and correlates with mass balance.    

Finally, the strong decline in the mass balance from 1991-2010 without a rising temperature trend suggests Zongo is out of balance with the prevailing climate in the tropical Andes.  Glaciers operate as low pass climate filters which take a significant amount of time to equilibrate to the prevailing climate.  Given the absence of a temperature trend in Zongo's vicinity over this two decade period, the remarkable loss of mass indicates that the glacier was, and likely still is, out of balance with the prevailing climate in the tropical Andes.  On somewhat different time scales, both average temperature and mass balance act as low pass filters to remove noise and express the signal in the underlying climatic processes even though they may not capture the details of the physics on a small time scale.


Annual mass balance of Zongo has a strong negative correlation with nearby annual temperature anomalies and Zongo will continue to melt as it is out of balance with the prevailing climate.

The summit ridge of Huayna Potosi (6088m), near the top of the Zongo glacier:

Citations and Other Relevant Articles Reviewed for Project Zongo

Blard, P-H., et al. "Degree-day melt models for paleoclimate reconstruction from tropical glaciers: calibration from mass balance and meteorological data of the Zongo glacier (Bolivia, 16° S)." Climate of the Past Discussions 7.3 (2011): 2119-2158.

Carvalho, Leila MV, Charles Jones, and Brant Liebmann. "The South Atlantic convergence zone: Intensity, form, persistence, and relationships with intraseasonal to interannual activity and extreme rainfall." Journal of Climate 17.1 (2004): 88-108.

Carvalho, Leila MV, et al. "Precipitation characteristics of the South American monsoon system derived from multiple datasets." Journal of Climate 25.13 (2012): 4600-4620.

Favier, Vincent, Patrick Wagnon, and Pierre Ribstein. "Glaciers of the outer and inner tropics: A different behaviour but a common response to climatic forcing." Geophysical Research Letters 31.16 (2004).

Jeschke, M.L., "Glacier retreat in the Bolivian Andes as a consequence of global climate change."  Dissertation, Potsdam University (2009), available here.

Jomelli, Vincent, et al. "Irregular tropical glacier retreat over the Holocene epoch driven by progressive warming." Nature 474.7350 (2011): 196-199.

Rabatel, A., et al. "Review article of the current state of glaciers in the tropical Andes: a multi-century perspective on glacier evolution and climate change." The Cryosphere Discussions 6.4 (2012): 2477-2536.

Rabatel, A., et al. "Current state of glaciers in the tropical Andes: a multi-century perspective on glacier evolution and climate change." The Cryosphere 7.1 (2013): 81-102.

Ramallo, Cinthya, et al. "Rainfall analysis and its implication on the mass balance and hydrological regime of the Zongo Glacier (Bolivia)." EGU General Assembly Conference Abstracts. Vol. 15. 2013.

Seiler, Christian, Ronald WA Hutjes, and Pavel Kabat. "Climate Variability and Trends in Bolivia." Journal of Applied Meteorology and Climatology 52.1 (2013): 130-146.

Seiler, Chrisitian, Ronald WA Hutjes, and Pavel Kabat. "Likely Ranges of Climate Change in Bolivia." Journal of Applied Meteorology and Climatology 2013 (2013).

Sicart, J. E., P. Wagnon, and P. Ribstein. "On the relation between meteorological conditions and the melting of outer tropics' glaciers." EGS-AGU-EUG Joint Assembly. Vol. 1. 2003.

Sicart, Jean Emmanuel, Patrick Wagnon, and Pierre Ribstein. "Atmospheric controls of the heat balance of Zongo Glacier (16 S, Bolivia)." Journal of Geophysical Research: Atmospheres (1984–2012) 110.D12 (2005).

Sicart, Jean Emmanuel, et al. "Glacier mass balance of tropical Zongo glacier, Bolivia, comparing hydrological and glaciological methods." Global and Planetary Change 59.1 (2007): 27-36.

Sicart, Jean Emmanuel, Regine Hock, and Delphine Six. "Glacier melt, air temperature, and energy balance in different climates: The Bolivian Tropics, the French Alps, and northern Sweden." Journal of Geophysical Research: Atmospheres (1984–2012) 113.D24 (2008).

Sicart, Jean Emmanuel, et al. "Sky longwave radiation on tropical Andean glaciers: parameterization and sensitivity to atmospheric variables." Journal of Glaciology 56.199 (2010): 854-860.

Sicart, Jean Emmanuel, et al. "Analysis of seasonal variations in mass balance and meltwater discharge of the tropical Zongo Glacier by application of a distributed energy balance model." Journal of Geophysical Research: Atmospheres (1984–2012) 116.D13 (2011).

Soruco, Alvaro, et al. "Mass balance of Glaciar Zongo, Bolivia, between 1956 and 2006, using glaciological, hydrological and geodetic methods." Annals of glaciology 50.50 (2009): 1-8.

Vuille, Mathias, et al. "Climate change and tropical Andean glaciers: Past, present and future." Earth-Science Reviews 89.3 (2008): 79-96.

Vuille, Mathias. Climate Change and Water Resources in the Tropical Andes. Technical Note No. IDB-TN-515. Inter-American Development Bank, 2013.

No comments:

Post a Comment