Toward an imminent extinction of Colombian glaciers?

Toward an imminent extinction of Colombian glaciers? 1 2 3 Antoine Rabatel1, Jorge Luis Ceballos2, Natan Micheletti3, Ekkehard Jordan4, Michael 4 Braitmeier4, Javier González4, Nico Mölg5, Martin Ménégoz6, Christian Huggel5, Michael 5 Zemp5 6 7 (1) Univ. Grenoble Alpes, CNRS, IRD, Institut des Géosciences de l’Environnement (IGE), Grenoble, France8 (2) IDEAM, Bogota, Colombia9 (3) Institute of Earth Surface Dynamics, University of Lausanne, Switzerland10 (4) Heinrich-Heine-Universität, Düsseldorf, Germany11 (5) Department of Geography, University of Zurich, Switzerland12 (6) Barcelona Supercomputing Center, Edifici Nexus II, C/ Jordi Girona, 31, 08034 Barcelona, Spain.13 14 Published online: 10 Oct 2017, in Geografiska Annaler: Series A, Physical Geography 15 http://www.tandfonline.com/doi/abs/10.1080/04353676.2017.1383015?journalCode=tgaa20& 16 Citation: 17 Rabatel, A., J.L. Ceballos, N. Micheletti, E. Jordan, M. Braitmeier, J. Gonzalez, N. Mölg, C. Huggel, M. Zemp. 18 2017. Toward an imminent extinction of Colombian glaciers? Geografiska Annaler: Series A, Physical 19 Geography. doi: 10.1080/04353676.2017.1383015. 20 21 22 Corresponding author: 23 Dr. Antoine Rabatel 24 Institut des Géosciences de l’Environnement 25 54 rue Molière, 26 38400 Saint Martin d’Hères, France 27 E-mail: [email protected] 29 This is an Accepted Manuscript of an article published by Taylor & Francis in Geografiska Annaler: Series A, Physical Geography, on 2018, available online: https://www.tandfonline.com/doi/abs/10.1080/04353676.2017.1383015 ABSTRACT. 30 This study documents the current state of glacier coverage in the Colombian Andes, the 31 glacier shrinkage over the 20th century and discusses indication of their disappearance in the 32 coming decades. Satellite images have been used to update the glacier inventory of Colombia 33 reflecting an overall glacier extent of about 42.4±0.71 km2 in 2016 distributed in four 34 glacierized mountain ranges. Combining these data with older inventories, we show that the 35 current extent is 36% less than in the mid-1990s, 62% less than in the mid-20th Century, and 36 almost 90% less than the Little Ice Age maximum extent. 37 Focusing on Nevado Santa Isabel (Los Nevados National Park), aerial photographs from 1987 38 and 2005 combined with a terrestrial LiDAR survey show that the mass loss of the former ice 39 cap, which is nowadays parceled into several small glaciers, was about -2.5 m w.e. yr-1 during 40 the last three decades. Radar measurements performed on one of the remnant glaciers, La 41 Conejeras glacier, show that the ice thickness is limited (about 22 m in average in 2014) and 42 that with such a mass loss rate, the glacier should disappear in the coming years. 43 Considering their imbalance with the current climate conditions, their limited altitudinal 44 extent and reduced accumulation areas, and in view of temperature increase expected in future 45 climate scenarios, most of the Colombian glaciers will likely disappear in the coming decades. 46 Only the largest ones located on the highest summits will probably persist until the second 47 half of the 21st century although very reduced. 48 49 KEY WORDS. Glaciers, surface area changes, tropical Andes, Colombia 50 51 1. INTRODUCTION 52 Glaciers in Colombia are more than ice on mountains: they indeed are key components of the 53 landscapes lived by the Colombian society (Ceballos et al., 2012). For peasants, indigenous, 54 mountain climbers, artists, scientists and city dwellers, glaciers in Colombia fulfill different 55 functions within their territories and are part of their daily practices in different ways: from 56 sentinels of global climate changes, local water resources, unique ecosystems, to local-to-57 regional sources of mass flow hazards from glaciers on active volcanoes in the Cordillera 58 Central (Jordan et al., 1989; Thouret, 1990; Linder, 1991, 1993; Linder et al., 1994; Huggel et 59 al., 2007). The large ice loss of Colombian glaciers since the late 1970s, (Ceballos et al., 60 2006; Morris et al., 2006; Poveda and Pineda, 2009), like in most part of the tropical Andes 61 (Rabatel et al., 2013a), has strengthened the necessity of a glacier monitoring combining 62 repeated inventories at the national scale and in-situ measurements on benchmark glaciers 63 located in the two mostly glacierized mountain ranges of Colombia (e.g., Ceballos et al., 64 2012; Mölg et al., 2017). Such a monitoring strategy is in line with the international strategy 65 for glacier monitoring defined by the Global Terrestrial Network for Glaciers (GTN-G, gtn-66 g.org). 67 In-situ measurements were initiated after the eruption event of Nevado del Ruiz in 1985 and 68 are nowadays conducted by the Instituto de Hidrología, Meteorología y Estudios Ambientales 69 (IDEAM). These activities have been part of different international programs: started by the 70 Instituto Geográfico Agustín Codazzi (IGAC, Bogota) in 1988 supported by Deutsche 71 Forschungsgemeinschaft (DFG) and Volkswagen-Foundation, IDEAM has taken over the 72 task, presently cooperating with the joint international laboratory GREAT-ICE (Sicart et al., 73 2015) financed by the French Institut de Recherche pour le Développement (IRD), the World 74 Glacier Monitoring Service (wgms.ch), and the CATCOS project (Capacity Building and 75 Twinning for Climate Observing Systems) financed by the Swiss Agency for Development 76 and Cooperation (SDC). 77 The aims of this paper are: 1) to present and analyze the current state of glaciers in Colombia, 78 with the results of a new glacier inventory from 2016; 2) to draw a multi-decadal perspective 79 of changes in glacier surface-area using repeated glacier inventories since the mid-20th 80 Century and Little Ice Age maximum extent; and 3) to estimate the future evolution of 81 glaciers in Colombia on the basis of the up-to-date inventory, current surface-area and mass 82 loss rates, as well as future possible changes (until 2100) in air temperature according to 83 climate scenarios. 84 85 2. STUDY AREA 86 Glaciers in Colombia are located in four main areas (Fig. 1): from North to South: Sierra 87 Nevada de Santa Marta (about 10°50’ N; 73°40’ W), Sierra Nevada de El Cocuy (about 6°25’ 88 N; 72°20’ W), Cordillera Central: Los Nevados National Park (Ruiz-Santa Isabel-Tolima, 89 about 4°45’ N; 75°20’ W), and Cordillera Central: Nevado Huila (about 2°55’ N; 76°00’ W). 90 In the two northernmost areas, small slope glaciers can be found, whereas in the two 91 southernmost areas, small ice caps lying on more or less active volcanoes with significant 92 different slope angles (e.g., nevados del Ruiz, de Santa Isabel, de Tolima and del Huila) are 93 the dominant glacier type. 94 Table 1 lists the main characteristics of the glacierized areas of Colombia, with the glacier 95 cover in 2016, the maximum ice thickness estimate -where it existstogether with the year of 96 the estimate. One can note that the averaged maximum elevations of the glacierized summits 97 range in most parts between 5,100 and 5,400 m a.s.l. which is rather low in comparison with 98 the other glacierized areas in the tropical Andes of Ecuador, Peru and Bolivia where the 99 highest elevations frequently exceed 6,000 m a.s.l. (Rabatel et al., 2013a). 100 2.1. Climatic settings 101 From a climatological point of view, Colombia belongs to the inner tropics (Troll, 1941) with 102 continued humidity, homogeneous temperature (daily amplitude > annual amplitude) and 103 almost constant incident solar radiation throughout the year. At the seasonal scale, the 104 displacement of the inter-tropical convergence zone (ITCZ) strongly controls the annual 105 regime of precipitation which results to be contrasted from one region to the other at the 106 country scale (e.g., Poveda et al., 2005). The central and western parts of Colombia 107 (glacierized areas C and D on Fig. 1) experience a bimodal precipitation regime with two 108 periods of high precipitation (April-May and October-November) and two periods of less 109 precipitation (December-February and June-August). On the other hand, the Caribbean coast 110 (glacierized area A on Fig. 1) and the Pacific coast of the isthmus with Panama show a 111 unimodal precipitation regime (May-October), resulting from the northernmost position of the 112 ITCZ. The easternmost glacierized mountain range (glacierized area B on Fig. 1) also 113 experiences a single precipitation peak occurring during June-August which results from deep 114 convection of the moisture transported from the Amazon basin due to the orographic barrier 115 of the Andes. 116 The IDEAM maintains automatic weather stations (AWS) in the glacierized areas B and C (at 117 elevations up to 4700 m a.s.l.). Mölg et al. (2017) presented the data from the AWS located 118 on the Nevado Santa Isabel (glacierized area C) which show that over the monitoring period 119 (2009-2016), the average 0 °C isotherm was located at 4980 m a.s.l., higher than the summit 120 located at 4940 m a.s.l. In Sierra Nevada de El Cocuy (glacierized area B) the average 0 °C 121 isotherm over the period 2007-2016 was located at 5045 m a.s.l. It is worth noting that these 122 average 0 °C isotherm estimates may slightly vary within the considered glacierized areas and 123 in the other glacierized areas of Colombia due to local site effects. 124 The inter-annual variability of atmospheric conditions is dominated by the El Niño-Southern 125 Oscillation (ENSO). Although the climate characteristics of La Niña/El Niño events are not 126 uniform at the scale of a country, El Niño years (warm phase of ENSO) tend to be warmer 127 and drier, while La Niña years (cold phase of ENSO) are typically associated with colder and 128 wetter conditions in the mountains (e.g., Poveda et al., 2011). Poveda et al. (2011) underlined 129 that the ENSO effects are phase-locked to the above described seasonal cycle: i.e. stronger 130 during more intense precipitation months and vice versa. 131 2.2. Glacier surface processes 132 In terms of surface mass balance regime, the Colombian glaciers belong to the inner-tropics 133 (Kaser and Osmaston, 2002) as precipitation may occur all year long with one or two periods 134 of more intense precipitation depending on the glacierized region concerned. 135 Using the longest Colombian surface mass balance time series (since 2006) on La Conejeras 136 glacier on the Nevado Santa Isabel, Mölg et al. (2017) showed that there is no seasonal cycle 137 with ablation/accumulation processes that can occur all year long covering parts of or the 138 entire glacier surface area. They also mentioned that the impact of temperature and 139 precipitation on the surface mass balance relies on the phase of precipitation and the 140 subsequent albedo effect. This is in line with the former studies made on another glacier of the 141 inner-tropics located in the Ecuadorian Andes: Antizana 15 glacier, where both surface mass 142 and energy balance studies (Francou et al., 2004; Favier et al., 2004) revealed the strong 143 relationship between glacier surface albedo and melting. These studies showed that the 144 frequency and intensity of snowfalls, which can occur all year long, play a major role in 145 attenuating the melting processes and consequently, both precipitation and temperature are 146 crucial for the annual surface mass balance. 147 In a review paper about the state of glaciers in the tropical Andes, Rabatel et al. (2013a) 148 concluded that the sensitivity of inner tropical glaciers to climate is closely linked to the 149 absence of temperature seasonality and to the fact that the 0 °C isotherm constantly oscillates 150 through the glaciers. As a consequence, a minor variation in air temperature can influence the 151 melt processes by determining the phase of precipitation and consequently affects the surface 152 albedo and mass balance. 153 154 3. METHODS AND DATA 155 3.1. Quantification of glacier surface-area 156 Former studies have documented the glacier surface-area changes since their maximum extent 157 during the Little Ice Age and until the early 2000s (e.g., Jordan et al., 1989; Florez, 1992; 158 Pulgarin et al., 1996; Ceballos et al., 2006; Poveda and Pineda, 2009; Herrera and Ruiz, 159 2009). Note that the Little Ice Age maximum extent has not been so systematically dated in 160 the Colombian Andes as it was the case in the other countries of the tropical Andes (e.g., 161 Rabatel et al., 2005, 2008; Jomelli et al., 2009), even if the link between moraines and 162 reliably dated Ruiz eruptions on 1595/03/12 and 1845/02/18 locally provides good indicators 163 (Jordan et al., 1987; Jordan and Mojica, 1987). The former studies on the extent of Colombian 164 glaciers in the past are based on moraines (reflecting the Little Ice Age maximum extent), 165 aerial photographs from the late 1940s to the mid-1990s, and Landsat TM and ETM from the 166 mid-1990’s to the early 2000s. In the current study, an update of the glacial coverage across 167 all Colombian glacierized mountain ranges has been realized using images from the following 168 satellites: QuickBird (2007, spatial resolution of 2.5 m in multispectral mode = visible + near-169 IR), ALOS (2007, 2008, 2009, spatial resolution of 10 m in multispectral mode = visible + 170 near-IR), RapidEye (2010, spatial resolution of 5 m in multispectral mode = visible + near-IR) 171 and Landsat-8 OLI (2016, Fig. 1). 172 On the basis of Landsat-8 images from late January-early February 2016, a detailed inventory 173 was produced and a database was generated according to the design of the GLIMS glacier 174 relational database. For an extensive description of the database content, the reader will refer 175 to the GLIMS website (http://www.glims.org/MapsAndDocs/db_design.html). The 2016 176 Landsat-8 images provide the perfect conditions for a glacier inventory: no snow cover 177 outside the glaciers and no cloud cover on the mountains, a particular challenge in Colombia 178 due to often persistent cloudy weather conditions. These images have a spatial resolution of 179 30 m in multispectral mode and 15 m in the panchromatic mode (the spectral bands “green”, 180 “NIR-IR” and “MIR” available at 30 m have been pansharpened at 15 m). Due to the small 181 size of the glaciers and their limited number, the delineation of the glacier outlines has been 182 made manually. Manual delineation can have advantages over automatic detection of glacier 183 ice in shadowed areas (Gardent et al., 2014). Note that debris-covered glacier areas are 184 limited in Colombia, either because the glaciers are small ice caps, or remnants of ice caps, or 185 slope glaciers; and in every case, rock walls overhanging the glaciers are limited or absent. 186 However, ashes resulting from eruptions can cover some parts of the ice caps located on 187 active volcanoes. On the 2016 satellite images, it was the case on the north-western side of the 188 Nevado del Ruiz, but because the ash cover was not homogeneous and ice free areas can be 189 seen, it did not prevent an accurate delineation of the glacier margin. For older data sources, 190 aerial photographs from 1959, 1987 and 2005 used on Nevado del Ruiz allowed an accurate 191 delineation of glacier contour due to their high spatial resolution. 192 Regarding the uncertainties, they largely depend on the data sources (moraines, aerial photos, 193 satellite images). It is noteworthy that the estimated values for the Little Ice Age are probably 194 associated with the highest uncertainty compared with inventories performed using aerial 195 photos or satellite images. Indeed, for the Little Ice Age the surface area reconstruction is 196 based on the moraine ridges which are not always continuous over the glacier foreland. 197 However, the uncertainty is not given in all the related studies. Regarding the aerial photos 198 and the satellite images, to compute a margin of uncertainty on the delineation of the glacier 199 outline Rabatel et al. (2011) considered different sources related to: 200 (i) the pixel size of the image or digital photograph, which has an influence on the 201 digitization; 202 (ii) the process of geometric correction and georeferencing of the images, orthophotos 203 and numerical maps, which affects the geometry of the used data source; 204 (iii) the errors associated with visual identification and manual delineation of the 205 glacier outline; which depend on the ability and experience of the operator. After a 206 test of multiple digitization, this error was set a ±1pixel for the Landsat satellite 207 images used for 2016, and ±2 pixels for the orthophotos or very high resolution 208 satellite images like Quickbird used for 2007 and RapidEye used for 2010; 209 (iv) the possible residual snow cover, which compromises the accurate visual 210 identification of the border of the glacier. This error has a huge spatial variability, 211 but is always limited in our case because the images were selected to have a 212 minimum snow cover outside the glaciers. 213 The total uncertainty is the root of the quadratic sum of the different independent errors. 214 Uncertainty in surface area can be considered as the horizontal uncertainty of the position of 215 the margin times its length (Rabatel et al., 2011). 216 3.2. Quantification of glacier volume 217 Ice thickness measurements have been acquired on La Conejeras glacier (Nevado Santa 218 Isabel) using an ice penetrating radar (IPR) during field campaigns in January-February 2014. 219 Our IPR is a geophysical instrument specially designed by the Canadian company Blue 220 System Integration Ltd in collaboration with glaciologists to measure the thickness of glacier 221 ice (Mingo and Flowers, 2010). It comprises a pair of transmitting and receiving 5 MHz 222 antennas that allow continuous acquisition, georeferenced with a GPS receiver. 223 the balacance-budget equilibrium line altitude (ELA0, cf. Cogley et al., 2011) derived from 374 the surface mass balance measurements for La Conejeras glacier is about 4920 m a.s.l., thus 375 160 m above its mean altitude computed from the area-altitude distribution (i.e. 4760 m a.s.l.), 376 and 120 m above the mean altitude of the glaciers located on the Nevado Santa Isabel. This is 377 the same for the Sierra Nevada de El Cocuy where the mean altitude of the glaciers is 4910 m 378 a.s.l., 120 m below the ELA0 derived from in situ measurements on Ritacuba glacier. 379 Reanalyzing the 10-yr monthly mass balance time series of La Conejeras glacier, Mölg et al. 380 (2017) have shown that the mean annual mass balance has been close to -3 m w.e. yr-1 over 381 the period 2006-2015 (cf. Fig. 7 where annual mass balance are plotted). Mölg et al. (2017) 382 also showed that the annual ELA was on average close to the glacier maximum altitude 383 during the monitoring period, with an accumulation-area ratio of about 4%, i.e. almost no 384 accumulation zone. 385 On the other hand, the comparison between the 2014 Lidar DEM and the photogrammetric 386 DEM from 1987 (see section 3.3) showed that the glacier-surface elevation has lowered by 80 387 m at 4700 m a.s.l. (altitude of the glacier surface close to the front of the glacier in 2014) 388 between the two dates; 50 m between 1987 and 2005 (Fig. 6). The geodetic mass balance was 389 -2.56 m w.e. yr-1 for the period 1987-2005 and -2.46 m w.e. yr-1 for the period 2005-2014 390 (Fig. 7). Note that the in situ surface mass balance averaged over the closest period (i.e. 2007-391 2014) was -2.45 m w.e. yr-1. This very good agreement between the two independent methods 392 shows that the well distributed network of in situ measurements at the surface of La Conejeras 393 glacier allows an accurate quantification of the mass balance using the glaciological method. 394 Computed at the scale of the entire Santa Isabel ice cap, the geodetic mass balance between 395 1987 and 2005 was -2.69 m w.e. yr-1, i.e. slightly more negative than considering La 396 Conejeras glacier only. 397 398 4.5. Future changes of Colombian glaciers 399 The strong shrinkage of the Colombian glaciers since the mid-20th century and in particular 400 the constant increase in the rate of shrinkage at the country scale over the past four decades is 401 an indication of the strong imbalance of glaciers with current climate. Figure 7 shows the 402 volume loss of La Conejeras glacier computed on the basis of the ice thicknesses measured in 403 2014, the annual changes in surface-area and the surface mass balances in situ measured since 404 2006. A linear extrapolation of the glacier volume changes of the last decade for the future 405 would result in the disappearance of La Conejeras glacier in the first years of the 2020s, likely 406 in concert with the other remaining glaciers of Nevado Santa Isabel. 407 The mass balances measured on La Conejeras and Ritacuba glaciers cannot be directly 408 extrapolated to the scale of all other glaciers in Colombia, as neighboring glaciers under 409 similar climate conditions can show different mass balances in relation with the dynamic 410 response of glaciers to a change in climate forcing (e.g., Rabatel et al., 2016). Estimates of 411 future changes and disappearance of Colombian glaciers based on decadal trends in glacier 412 surface-area loss therefore imply some uncertainty. Nevertheless, as a first approximation a 413 linear trend extrapolation from the observed glacier surface-area shrinkage rates in the 414 different glacierized areas of Colombia during the last decades (Fig. 5A) allows a rough 415 estimation of their future changes and disappearance. Accordingly, glaciers on the Nevado de 416 Tolima will likely disappear before 2030, and most of the glaciers in the Sierra Nevada de 417 Santa Marta and Sierra Nevada de El Cocuy before 2050. Only the few largest glaciers with 418 the highest maximum elevations on Nevado del Huila, Nevado del Ruiz and in the Sierra 419 Nevada de Santa Marta and Sierra Nevada de El Cocuy will probably persist after the mid-21st 420 century although strongly reduced. Our results suggest that glacier extinction in Colombia 421 happens much faster than the corresponding estimates in the 4th Assessment Report of the 422 Intergovernmental Panel on Climate Change (IPCC) (Magrin et al., 2007: “within the next 423 100 years”), but not as dramatic as suggested by Poveda and Pineda (2009: “by the late 2010-424 20 decade”). The latter estimates are based on Landsat TM and ETM+ images from 1989-425 2007 and result in slightly smaller total areas for 2004-07, and correspondingly higher loss 426 rates, than the present study. 427 Taking into account the influence of temperature changes on glacier surface processes (see 428 2.2.), an alternative to the extrapolation of surface-area changes can be made from the 429 relationship between the 0 °C isotherm and the maximum elevation and/or the ELA of the 430 glaciers, and considering the future projections of temperature using different climate 431 scenarios. Figure 8 shows the Hadcrut4 observations (Morice et al., 2012) as well as historical 432 and future CMIP5 experiments following the two extreme radiative concentration scenarios 433 (RCPs) RCP 2.6 and 8.5 (Taylor et al., 2012) for the near-surface air temperature. Data from 434 different global climate models (see Fig. 8 caption) are averaged over the region defined as 435 the box 2°-10°N, 72°-77°W to encompass the different glacierized areas in Colombia. Over 436 the reference period extended from 1961-1990, both model and observations show a 437 temperature increase within the range of 0.5 °C, an increase smaller than the inter-annual 438 variability over this period. The scenarios RCP 2.6 and RCP 8.5 show an increase in air 439 temperature reaching respectively 1.6 °C [0.5 to 2.7 °C] and 6.3 °C [5.4 to 7.2 °C] by the end 440 of the 21st century considering a 10-year average of the multi-model ensemble experiments. 441 Assuming that the current vertical gradient of air temperature remains unchanged, such an 442 increase in temperature would raise the 0 °C isotherm by 320 m (ranging from 100 to 540 m) 443 for RCP 2.6 and by 1260 m (ranging from 1080 to 1440 m) for RCP 8.5; i.e. reaching the 444 elevation of 5320 and 6260 m a.s.l., respectively. Note that these estimates are in close 445 agreement with the results found by Schauwecker et al. (2017) for the Peruvian Andes. In 446 such conditions, 75% (100%) of the Colombian glaciers would be entirely located below the 0 447 °C isotherm by the end of the 21st century considering RCP 2.6 (8.5). 448 In addition, although the time-series are short (~10 years) the meteorological and 449 glaciological data from La Conejeras glacier (Mölg et al., 2017) allow quantifying the 450 sensitivity of the ELA to air temperature and elevation of the 0 °C isotherm. The significant 451 correlation between the ELA and the 0 °C isotherm (r = 0.9, p < 0.002) shows that a 100-m 452 increase in the 0 °C isotherm leads to an increase in the ELA by 160 m. As a consequence, the 453 above mentioned increases in the 0 °C isotherm by the end of the 21st century would place the 454 ELA 500 and 2000 m above its current location for the RCP 2.6 and 8.5. Assuming that these 455 estimates made from the data available on La Conejeras glacier can be transposed to the other 456 glacierized areas in Colombia, the projected ELA would be above the maximum elevation of 457 80% (100%) of the Colombian glaciers. In such conditions, glaciers in Colombia would 458 constantly be in ablation over most or the totality of their surface-area (very limited or no 459 accumulation zone would persist) and their shrinkage/disappearance looks ineluctable. 460 It is worth noting that considering the RCP 2.6, the increase in air temperature during the 461 coming decades would mainly occur before 2040-2050, meaning that the remaining 462 glacierized surface areas in Colombia would stabilize during the second half of the 21st 463 century. 464 Finally, it must be reminded that even with the use of "anomaly" approaches applied to 465 remove the biases of climate models, large uncertainties remain when using CMIP5 scenarios, 466 in particular because of the potential non-stationarity of the model bias. Global climate 467 models show also weaknesses to simulate regional atmospheric circulation changes 468 (Shepherd, 2014) and their coarse resolution does not allow to simulate correctly the 469 feedbacks strengthening the warming with the altitude (MRIEDW, 2015) and the local impact 470 of particle deposition on glacierized areas (Hansen and Nazarenko, 2004). Even with 471 significant improvements in terms of ENSO modeling from CMIP3 to CMIP5 (Bellenger et 472 al., 2014), it is very challenging to anticipate the potential ENSO changes over the next 473 decades, and these ones may have strong impacts on the Colombian climate. Nevertheless, the 474 use of CMIP5 model projections is currently one of the unique ways to anticipate the future 475 changes in temperature and precipitation. Retrospective validations show that CMIP models 476 reproduce the main features of the current climate in Southern America (e.g., Vera et al., 477 2006; Sillmann et al., 2013) and can be used to estimate the future trends of temperature over 478 this continent, whereas the uncertainties in terms of precipitation are very high (Blazquez et 479 al., 2013). By setting up calibration approaches, Marzeion et al. (2014) and Réveillet et al. 480 (2015) demonstrated the possibility to use CMIP outputs to simulate glaciers future evolution. 481 We describe here a potential evolution for the Colombian glaciers that follows two scenario 482 based on different societal evolutions. A limitation of our study relies on the regional or local 483 forcing and feedbacks described previously that could modulate these future evolutions. 484 4.6. Potential impacts of future glacier changes 485 In other regions of the tropical Andes glaciers represent an important source of water for 486 domestic, agricultural or industrial use, for example in La Paz - Bolivia where the water 487 coming from the glaciers represents up to 30% of the runoff during the dry season (Soruco et 488 al., 2015), and recent studies have shown the negative impacts of current glacier shrinkage on 489 the biodiversity of the proglacial areas (e.g., Dangles et al., 2017; Zimmer et al., 2017). In 490 Colombia, the potential impact of glacier shrinkage mainly relates to the páramo ecosystems 491 (Brown et al., 2007), as well as for local agriculture and tourism. However, the glacierized 492 volcanoes in Colombia remain – at least for the next few decades – a natural hazards, as 493 dramatically shown with the example of the post-eruption lahars of the Nevado del Ruiz in 494 1985 (e.g., Jordan et al., 1987; Thouret, 1990). Because the Nevado del Ruiz presents the 495 largest single ice coverage in Colombia (10.11 km² in 2016) with an estimated ice volume of 496 484 x106 m3 back in 2003 (measured maximum and mean thickness of 190 and 47 m in 1999, 497 Huggel et al., 2007), the risk of lahars generated from the interaction of volcanic activity and 498 snow and ice will still persist for several decades. As a consequence, to better estimate the 499 potential water release resulting from an eruption of the Nevado del Ruiz and to prepare 500 potential impact scenarios, an accurate mapping to the ice thickness and distribution, as we 501 presented here for La Conejeras glacier, is urgently recommended. A similar mass of ice (8.0 502 km² in 2016 and 648 x106 m3 of ice estimated for 2001) persists on Nevado del Huila which 503 produced several far-reaching (up to 150 km) lahars in 2007 and 2008 when Nevado del Huila 504 erupted and large amount of water were produced (Worni et al., 2012). 505 506 5. CONCLUSION 507 In this study we presented the results of a new glacier inventory of the Colombian Andes 508 using 2016 Landsat images, in combination with in situ measurements of glacier thickness 509 using radar and of glacier surface topography using LiDAR and aerial photogrammetry on the 510 well studied La Conejeras glacier located on the Nevado Santa Isabel in Los Nevados 511 National Park. 512 The main results showed that: 513 - The glacier surface area is nowadays very reduced in Colombia, with a total ice covered 514 area in 2016 of 42.4 km². The mean glacier size was 0.43 km², and small size glaciers 515 largely predominate (70 % < 0.5 km²). 516 - The glacier shrinkage is strong since the mid-1970s and, remarkably, almost constantly 517 increasing reaching a mean annual area loss rate of -3 % yr-1 during the last years, which 518 points to a continued climatic forcing, possibly in addition to local topographic and 519 geometric effects. 520 - Mass loss on the Santa Isabel ice cap has been strong over the last three decades with an 521 average annual mass balance of about -2.5 m w.e. yr-1 since 1987 quantified using aerial 522 photogrammetry and terrestrial LiDAR. 523 Considering the imbalance of the glaciers in Colombia with the current climate conditions, the 524 relative low altitude of the Colombian glaciers, and the expected changes in air temperature 525 for the 21st century, most of them will most likely disappear in the coming decades and only 526 the largest ones located on the highest summits will persist until the second half of the 21st 527 century. 528 529 AUTHOR CONTRIBUTION STATEMENT 530 A. Rabatel conducted the GPR monitoring on La Conejeras glacier, realized the 2016 glacier 531 inventory, analyzed the data, wrote the manuscript and produced the figures and tables. N. 532 Micheletti conducted the LiDAR measurements on La Conejeras glacier and produced the 533 DEM. J.L. Ceballos (with colleagues from IDEAM) performed ten years of mass balance in-534 situ measurements on La Conejeras glacier and analyzed the data with N. Mölg. J.L. Ceballos 535 and C. Huggel made the glacier mapping from the satellite images from 2007 to 2010. E. 536 Jordan, M. Braitmeier and J. González realized the photogrammetric DEMs of Nevado Santa 537 Isabel for 1987 and 2005 and completed the sources in Table 3. M. Ménégoz analyzed the 538 CMIP5 temperature data. M. Zemp led the Andean part of the CATCOS project and together 539 with J.L. Ceballos managed the 2014 field campaign. All the co-authors revised the 540 manuscript. 541 542 ACKNOWLEDGEMENTS 543 This study was conducted in the context of the project Capacity Building and Twinning for 544 Climate Observing Systems (CATCOS) supported by the Federal Office of Meteorology and 545 Climatology MeteoSwiss, contract no. 7F-08114.1, between the Swiss Agency for 546 Development and Cooperation (SDC) and MeteoSwiss, by the Swiss State Secretariat for 547 Economic Affairs (SECO). NASA/METI/AIST/Japan Space systems, U.S./Japan ASTER 548 Science Team are acknowledged for the release of the ASTER GDEM V2. Ekkehard Jordan 549 wishes to thank Luis Miguel Vélez, leader of AEROESTUDIOS, Medellin, for the 550 cooperation in the realization of aerial flights over Columbian glaciers in the adequate 551 conditions for the delivery of the first line of images of the Santa Isabel glaciers that is fully 552 stereoscopically valuable for photogrammetry. M. Ménégoz is supported by the project 553 VOLCADEC funded by the Spanish programme Retos (MINECO/FEDER, ref. CGL2015-554 70177-R) and thanks Pierre-Antoine Bretonnière (BSC, Barcelona, Spain) who downloaded 555 the CMIP5 data. Antoine Rabatel acknowledges the contributions of the SNO GLACIOCLIM 556 (use of the IPR), the LMI GREAT-ICE (Institut de Recherche pour le Développement, IRD) 557 and the Labex OSUG@2020 (Investissements d'avenir – ANR10 LABX56). Finally, we thank 558 Lothar Schrott (scientific editor), Wilfried Haeberli and an anonymous referee for their 559 constructive comments used to improve the paper. 560 561 REFERENCES 562 Basantes Serrano, R., Rabatel, A., Francou, B., Vincent, C., Maisincho, L., Cáceres, B., 563 Galarraga, R., Alvarez, D., 2016. Slight mass loss revealed by reanalyzing glacier mass 564 balance observations on Glaciar Antisana 15 (inner tropics) during the 1995-2012 period. 565 Journal of Glaciology, 62(231), 124-136. doi: 10.1017/jog.2016.17 566 Baumann, S., 2006. Aufbau eines Gletscherinventars für Kolumbien und Abschätzung 567 glaziologischer Parameter. Diploma thesis, TU München, Deutschland. 568 Bellenger, H., Guilyardi, É., Leloup, J., Lengaigne, M., Vialard, J., 2014. ENSO 569 representation in climate models: from CMIP3 to CMIP5. Climate Dynamics, 42(7-8), 570 1999-2018. 571 Blázquez, J., Nuñez, M.N., 2013. Analysis of uncertainties in future climate projections for 572 South America: comparison of WCRP-CMIP3 and WCRP-CMIP5 models. Climate 573 dynamics, 41(3-4), 1039-1056. 574 Borrero, C.A., Hincapie, G., Guarnizo, L.F., Ramirez, J., Coral, A., Valla, F., Thouret, J.C, 575 Jordan, E., Funk, M., 1996. Proyecto de Investigacion: IMIP – Procesos de interaccion 576 hielo-magma en el Volcan Nevado del Ruiz. In: Geologia y medio ambiente para el 577 desarrollo, II seminario sobre el Cuaternario en Colombia. INGEOMINAS, Bogotá, 452-578 462. 579 Braithwaite, R.J., 2015. From Doktor Kurowski’s Schneegrenze to our modern glacier 580 equilibrium-line altitude (ELA). The Cryosphere, 9, 2135-2148, doi:10.5194/tc-9-2135-581 2015 582 Braitmeier, M., 2003. Die Energiebilanz an der Oberfläche des Nevado Santa Isabel, 583 Kolumbien. PhD Thesis, Heinrich Heine Universität, Düsseldorf, Germany, 116 p. 584 Brown, S., Roa, C., Roa, C., Yepes L.D., 2007. Protocol for the characterization of carbon 585 and water cycles in high-elevation ecosystems of the Andes. Mountain Research and 586 Development, 27 (4), 372-375, doi: 10.1659/mrd.0953 587 Ceballos, J.L., Euscategui, C., Ramirez, J., Canon, M., Huggel, C., Haeberli, W., Machguth, 588 H., 2006. Fast shrinkage of tropical glaciers in Colombia. Annals of Glaciology, 43, 194-589 201. 590 Ceballos, J.L., Rodriguez Murcia, C.E., Real Nuñez, E.L., 2012. Glaciares de Colombia más 591 que montañas con hielo. IDEAM, Bogotá, Colombia, 344 p. 592 Cogley, J.G., 2009. Geodetic and direct mass-balance measurements: comparison and joint 593 analysis. Annals of Glaciology, 50, 96-100. 594 Cogley, J.G., Hock, R., Rasmussen, L.A., Arendt, A.A., Bauder, A., Braithwaite, R.J., 595 Jansson, P., Kaser, G., Möller, M., Nicholson L., Zemp, M., 2011. Glossary of glacier 596 mass balance and related terms. IHP-VII Technical Documents in Hydrology No. 86, 597 IACS Contribution No. 2, UNESCO-IHP, Paris. 598 Dangles, O., Rabatel, A., Kraemer, M., Zeballos, G., Soruco, A., Jacobsen, D., Anthelme, F., 599 2017. Ecosystem sentinels for climate change? Evidence of wetland cover changes over 600 the last 30 years in the high Bolivian Andes. PLoS ONE, 12(5), e0175814. doi: 601 10.1371/journal.pone.0175814 602 Favier, V., Wagnon, P., Chazarin, J.-P., Maisincho, L., Coudrain, A., 2004. One-year 603 measurements of surface heat budget on the ablation zone of Antizana glacier 15, 604 Ecuadorian Andes. Journal of Geophysical Research, 109, D18105, doi: 605 10.1029/2003JD004359 606 Francou, B., Vuille, M., Favier, V., Caceres, B., 2004: New evidence for an ENSO impact on 607 low-latitude glaciers: Antizana 15, Andes of Ecuador, 0°28’S. Journal of Geophysical 608 Research, 109, D18106. doi: 10.1029/2003JD004484 609 Vera, C., Silvestri, G., Liebmann, B., González, P., 2006. Climate change scenarios for 753 seasonal precipitation in South America from IPCC-AR4 models. Geophysical Research 754 Letters, 33, L13707, doi: 10.1029/2006GL025759 755 WGMS. 2016. Glacier Thickness Database 2.0. Gärtner-Roer, I., Andreassen, L.M., Bjerre, 756 E., Farinotti, D., Fischer, A., Fischer, M., Helfricht, K., Huss, M., Knecht, T., Kutuzov, S., 757 Landmann, J., Lavrentiev, I., Li, H., Li, Z., Machguth, H., Naegeli, K., Navarro, F., 758 Rabatel, A., Stentoft, P., Zemp, M. (eds.), World Glacier Monitoring Service, Zurich, 759 Switzerland. doi: 10.5904/wgms-glathida-2016-07 760 Worni, R., Huggel, C., Stoffel, M., Pulgarín, B., 2012. Challenges of modeling recent, very 761 large lahars at Nevado del Huila Volcano, Colombia. Bulletin of Volcanology, 74, 309-324. 762 Zimmer, A., Meneses, R., Rabatel, A., Soruco, A., Anthelme, F., 2017. Time lag between 763 glacial retreat and upward migration alters tropical alpine communities. Perspectives in 764 Plant Ecology Evolution and Systematics, in press. doi: 10.1016/j.ppes.2017.05.003 765 766 Glacierized area Glacier cover (km²) in 2016 Highest elevation of the area (m a.s.l.) Average max. elevation of glaciers (m a.s.l.) Average mean elevation of glaciers (m a.s.l.) Average min. elevation of glaciers (m a.s.l.) Max thickness (m) and year of estimate S.N. de Santa Marta 7.2±0.27 5678 5340 5170 5000 // S.N. de El Cocuy 15.5±0.33 5346 5055 4910 4750 // Los Nevados National Park 11.8±0.52 5314 5170 5060 4900 190, 1999 V.N. del Huila 8.0±0.23 5390 5250 5020 4710 // 767 Table 1: Colombian glacierized areas with the most up-to-date glacier cover surface-area and 768 topographic features of the glaciers. Note that the elevation data are computed from ASTER 769 GDEM V2 and may differ from other sources due to differences in the accuracy of used data. 770 Maximum thickness estimates are taken from Huggel et al. (2007) and Ceballos et al. (2012), 771 but the original data have been provided by J. Ramirez (Servicio Geologico de Colombia). 772 S.N. = sierra nevada, V.N. = volcán nevado. 773 774 Size class (km²) Colombia Sta Marta El Cocuy P.N. Los Nev. Huila <0.1 Number 30 13 6 11 Number (%) 30 46 24 35 Area (km²) 1.73 0.74 0.37 0.62 Area (%) 4 10 2 5 0.1-0.5 Number 40 11 11 10 8 Number (%) 40 39 44 32 53 Area (km²) 10.02 2.70 2.54 2.00 2.77 Area (%) 24 38 16 17 35 0.5-1 Number 19 3 3 7 6 Number (%) 19 11 12 23 40 Area (km²) 13.72 2.61 2.28 4.74 4.10 Area (%) 32 36 15 40 51 1-5 Number 10 1 5 3 1 Number (%) 10 4 20 10 7 Area (km²) 16.92 1.13 10.28 4.39 1.12 Area (%) 40 16 66 37 14 Total Number 99 28 25 31 15 Area (km²) 42.4 7.2 15.5 11.8 8.0 775 Table 2: Summary statistics (number and area) on glaciers in Colombia for the 2016 776 inventory. 777 778 Santa Marta El Cocuy Ruiz Santa Isabel Tolima Huila Total (km²) (km²) (km²) (km²) (km²) (km²) (km²) LIA max 82.60 c/* 148.70 c/* 47.50 c/* 27.80 c/* 8.60 c/* 33.70 c/* 348.9+23.7c/* 1939 21.40 c/* 1946 10.80 c/* 3.10 c/* 1954 19.40 c/* 110.6 1955 38.90 g 1958 2.7 c/* 1959 21.40 a/+ 20.70 b/+ 21.00 c/* 9.78 a/+ 9.50 b/+ 9.40 c/* 2.22 a/+ 1961 18.86 d/+ 1965 19.77 a/+ 16.30 c/* 19.06 d 1970 18.21 d 1973 14.1 e/°° 28.0 e/°° 1974 16.26 a/+ 1975 19.60 c/* 1976 21.3 e/°° 10.8 e/°° 3.8 e/°° 26.0 e/°° 1978 39.12 a/+ 38.80 c/* 1981 16.10 c/* 15.40 c/* 87.95 1985 35.70 c/* 18.70 c/* 1986 31.45 g 17.00 c/* 1987 17.70 b/+ 6.50 b/+ 6.40 f/+ 6.56 h/+ 2.10 c/* 1.60 g 1989 12.00 c/* 14.72 d/+ 1990 14.10 c/* 1994 23.70 g 66.43 1995 11.10 g 13.39 d/+ 1996 5.30 g 1997 11.76 g 1.18 g 2001 12.95 g 2002 8.40 g 10.32 g 3.33 g 1.03 g 53.33 2003 19.8 g 2005 2.78 h/+ 2007 7.70 /°+ 18.60 /°+ 2.60 /°+ 0.93 /°+ 10.80 /°+ 2008 17.70 /° 2009 7.40 /° 17.40 /° 2010 16.00 /° 1.80 /° 0.74 /° 9.70 /° 2016 7.20±0.27 /° 15.46±0.33 /° 10.11±0.26 /° 1.0±0.08 /° 0.65±0.06 /° 8.00±0.23 /° 42.42±0.71 ~2005-2016 -6 % -17 % -2 % -62 % -30 % -25 % -20 % ~1995-2016 -35 % -35 % -14 % -81 % -45 % -40 % -36 % ~1985-2016 -55 % -51 % -46 % -84 % -59 % -48 % -52 % ~1955-2016 -63 % -60 % -53 % -90 % -71 % -58 % -62 % LIA-2016 -91 % -90 % -79 % -96 % -92 % -76 % -88 % 779 Table 3: Surface area changes since the Little Ice Age maximum. For each glacierized area 780 the surface area (km²) for each date is presented as well as the loss for different periods (in % 781 of the initial surface area for the considered period). Data before 2007 were taken from 782 previous studie, the letter indicates the original study: a = Jordan et al., 1989; b = Linder 1991, 783 1993; c = Florez, 1992; d = Pulgarin et al., 1996; e = Hoyos-Patino, 1998; f = Braitmeier, 2003; 784 g = Ceballos et al., 2006; h = Gonzalez et al., 2010. Symbols indicate the method: * = 785 planimetry on aerial photos; + = photogrammetric restitution with uncertainty estimate; ° = 786 planimetry on satellite ortho-images (pixel size between 0.5 and 15 m); °° = planimetry on 787 Landsat MSS (pixel size of 79 m). Regarding the total glacier cover computed for the gray 788 shaded lines, when several surface-areas are available for a glacierized area, the average is 789 considered. The uncertainty for the 2016 inventory have been computed from the quadratic 790 sum of the uncertainties of each glacier of the considered area. 791 792 Figure 1: Glacierized areas in Colombia. The spectral bands combination used for the 793 Landsat-8 images provided by USGS-EDC involves the bands #6 (middle infra-red: MIR), #5 794 (short-wave infra-red: SWIR) and #3 (green). 795 796 797 Figure 2: IPR measurements acquired in Jan-Feb 2014 on La Conejeras glacier. 798 799 800 Figure 3: A) Ortho-photo of Nevado Santa Isabel with glacier extent in 1987 and 2016. B) 801 Ortho-photo from 2005, with the outline of the ice cap in 1987 (in dark grey) and the outline 802 of the remnant glaciers in 2005 and 2016 (light blue and red respectively). The horizontal 803 scale shown on A is the same for B. Sources: Braitmeier (2003) for the 1987 ortho-photos and 804 González et al. (2010) for the 2005 ortho-photos. 805 806 807 Figure 4: A) RIEGL VZ-6000 operating at the front of La Conejeras glacier. B) Scanned 808 point cloud of La Conejeras glacier. 809 810 811 Figure 5: A) Glacier surface area changes in the different glacierized areas of Colombia since 812 the 1940s. B) Rates of mean annual area loss in percentage per year for each glacierized area. 813 The black curve and grey area represent the average with 1 st-dev. interval. The average has 814 been smoothed using a polynomial fit. 815 816 817