Canadian Climate Stripes

Its been busy around here, and summer is too nice to be blogging. But one thing I’ve been meaning to pull together for a while now is climate stripes (or bar codes) for Canadian cities.

I’ve adapted Ed Hawkins climate stripe visualizations for Canadian cities using the homogenized temperature data from Environment Canada (here) and a few lines of Python code. Each bar represents a mean annual temperature anomaly, or the departure from the overall mean, with dark blue being the coldest and dark red being the warmest.

I’ve also plotted the temperatures versus year and scaled the colour of the points the same way, and given an example for the Vancouver data. To me, this is actually more intuitive. But chacun son gout!

Vancouver, BC (-1.9 to +1.6 C, 1896 to 2017):


Same Vancouver data shown as a scatterplot:


Prince George, BC (-2.9 to 2.2 C):


Calgary, AB (-2.7 to +3.2 C, 1885 to 2017):


Toronto, ON (-3.3 to +3.3 C, 1840 to 2017):


Hay River, NT (-4.3 to +3.3 C, 1893 to 2017):


Sidney, NS (-2.3 to +2.0 C, 1870 to 2017):



Peyto Glacier and Wapta Icefield: 100* years of change

Okay, 98 years technically. I’ve recently obtained high-quality scans of the Alberta/British Columbia Boundary Commission Survey maps, courtesy of Bob Sandford. They are real, and they are amazing. So after doing a quick georeferencing I’ve combined one of the maps with some recent imagery and datasets to produce this:


Peyto Glacier and the Wapta Icefield, from historical topographic maps (1919), GLIMS glacier outlines (purple), and a recent Landsat 8 scene. 

The survey maps from 1919 show glacier boundaries that line up well with the maximum glacier extents. Look for the trimlines that mark boundaries between vegetated areas and deglaciated areas. Significant retreats of Peyto Glacier, Bow Glacier, and Baker Glacier are clear from the animation. What’s less obvious is the change in surface height: not only are these glaciers melting back faster than they are advancing, but they are essentially deflating. Surface lowering is probably reducing the total volume of this ice mass faster than the retreat itself.

The topographic maps were constructed A.O. Wheeler, a pioneer of photogrammetry, and his survey team. Traversing up the divide between Alberta and British Columbia, Wheeler and his team set up and surveyed benchmark positions, and collected photographs from passes and high points all the way from the US border to north of Mt. Robson. Their expeditions took place over multiple seasons, with a median year of around 1919.  Christina Tennant and Brian Menounos published a neat study in 2012 on historical glacier area change in the Rockies using these maps.


Mt. St. Nicholas above the Bow Hut, part of the Wapta Icefield ski traverse (my photo!)

Having skiied the Wapta traverse a few years ago (okay, more like 15 years ago) the rate of glacier change that these historical data show really puts things in perspective. As in, you might be able to do the traverse without worrying about crevasses by the middle of the century. Thanks, #climatechange.

Western Canada, 2017: a hot and smoky preview

Overnight yesterday and today (12 September), a wildfire that had been burning near Waterton Lakes National Park exploded in size and triggered mass evacuations. This post is motivated in part by the stories that have continuously emerged from western Canada in the summer of 2017. Stories of people evacuating in the middle of the night with 20 minutes notice; stories of people losing everything they owned, their livestock, their pets. Stories of the fire crews and support crews and volunteers who have worked incredibly hard throughout the summer to minimize the damage and support those affected by the fires.


Overnight growth of the Waterton Lake wildfire. Fire activity detected from MODIS satellites, and shown in a Google Earth .kml layer (available at



In western Canada, from the Prairies to the Central Interior of British Columbia (BC), the summer of 2017 was hot by any standard. How hot was it? I coded up a simple Google Earth Engine script to calculate average daytime and nighttime surface temperatures for July and August 2017, and compared these with the average between 2000 and 2016 (Figure 1). The code uses surface temperatures measured from the MODIS Aqua satellite operated by NASA (MYD11A1).


Daytime surface temperature anomalies for July and August 2017, from MODIS TERRA. Anomaly is calculated as the difference between mean temperatures in 2017 and the mean temperatures observed between 2003 and 2016

Daytime surface temperatures in July and August were up to +5 C warmer than the 2002-2016 average (Figure 1) over a large part of the prairies, and between 0 and +2 C warmer through much of southern BC.  More than 1200 wildfires burned over 11,000 square kilometers in BC alone through most of the summer, leading to the evacuation of more than 45,000 people from their homes.  Interestingly, nighttime temperatures on the prairies were close to normal, but 2-3 C above normal in the mountains.


Nighttime surface temperature anomalies for July and August 2017, from MODIS TERRA.

Is the summer of 2017 a preview of the new normal? Average annual temperature increases of 4-7 C are projected for much of western Canada by the end of the century, depending on which emissions scenario the world follows. While moisture and forest management techniques are also an important part of the wildfire equation, the way human activities are changing the atmosphere will result in increased extreme fire weather conditions, and a longer fire season.


Projected summer temperature change by 2080-2100 under RCP8.5 (high emission scenario), from Environment and Climate Change Canada. Changes are relative to average temperatures between 1986 and 2005. Details here:

The Year of the Floods: S Asia and SE Texas in 2017

[Thanks to a reader suggestion, I’ve updated the title and my terminology to ‘south Asia’ to better reflect the regional geography.]

[update 2: corrected the precipitation maps to account for the fact GPM data is provided half-hourly in units of [mm/hr]. Estimated precipitation totals are half of what was originally mapped.]

The scale of the disaster unfolding in the wake of Hurricane Harvey is truly impossible to fathom. Rapid intensification of the storm as it headed towards the Texas coast led to Category 4 hurricane-force winds, and as the center of Harvey stalled out just inland from the Gulf of Mexico some regions received more than 1000 mm of rain in a few days. Some stations are now reporting storm totals over 49″ (1250 mm). The evacuations continue, the reservoirs upstream of Houston are being pumped into the bayous in the hopes of preventing an end-run of storm waters around the back of the dam, and the human toll continues to rise. (update: the reservoirs have overtopped their spillways)

The Global Precipitation Measurement (GPM) mission uses satellites to estimate rainfall rates over most of the globe. These data have already been used elsewhere to examine Harvey, and a short Google Earth Engine script can be used to map preliminary rainfall totals.


Updated GPM precipitation totals for SE Texas, 22 – 29 August 2017.  GPM IMERG data processed in Google Earth Engine and mapped in QGIS

The magnitude of the accumulated rainfall in the Houston area is mind-boggling, but the GPM estimate appears to be about half of what has been observed on the ground. Whats more astounding is that recent rain events in Nepal, India, and Bangladesh are of a similar magnitude and cover an even larger area. You might have missed the events that have taken place in South Asia in the past few weeks, and accurate tallies of the precipitation totals are slow to come out from the region. But the impacts of the floods are being felt by hundreds of millions.

Using Google Earth again, with the same spatial scale and the same colour scale, I mapped the accumulated precipitation between 8 August and 15 August, when the flooding peaked. Total estimated precipitation amounts more than 500 mm cover a large swath of the northern plains, along the borders of Nepal, India, Bangladesh, and Bhutan, with peak precipitation amounts over 1000 mm in Bangladesh.


Updated precipitation maps for south Asia, 8 – 15 August 2017. GPM IMERG data processed in Google Earth Engine and plotted in QGIS.

For now, the focus needs to be on the people who are still in trouble or just starting to pick up the pieces. As with most natural disasters, the poorest will be hit the hardest: 80% of those hit by Harvey have no flood insurance, and in thousands of remote villages in SE Asia basic food, water, and sanitation needs are not being addressed.

The role that climate change may or may not have played in both Hurricane Harvey and in the south Asia floods will be examined long after the flood-waters recede. But both events have taken place against the backdrop of a warming atmosphere and warming oceans, and may represent a new reality.


Anomalous snowfall: did it amplify the earthquake-triggered Langtang avalanche?

The disaster that unfolded in Langtang Valley in the moments after the M7.8 Nepal Earthquake on 25 April 2017 is difficult to contemplate. The earthquake-triggered collapse of glaciers high above the serene village of Langtang Valley contributed to a devastating avalanche that swept down from the mountains with a leading edge of straight-line winds that approached major tornado strength.  New research led by Koji Fujita from Nagoya University, published in the open-access journal Natural Hazards and Earth System Science (link), suggests that the avalanche may have actually been composed primarily of snow. The intersection of a rare (1 in 100 to 1 in 500 year) snowpack and a major earthquake (1 in 80 years) led to one of the biggest mass casualty events of the earthquake.

Perspective views of Langtang Village avalanche deposit. (a) orthoimage, and (b) pre- and post-event elevation difference.

In the paper, we use a combination of data collected from helicopter and UAV surveys in the days and months after the earthquake to quadruple previous estimates of the total volume of the avalanche deposit (Figure 1 above). Koji, who has worked in the valley since the 1980s and is well known by the villagers, also interviewed locals who witnessed the blast to help constrain the sequence of events. Finally, we pieced together weather station data from the valley (ICIMOD/DHM) and regional precipitation fields to determine the return period for a winter snowpack of this magnitude.

The winter of 2014-2015 was particularly wet: the passage of Cyclone HudHud in October 2014 dropped upwards of 300 mm of precipitation in the Langtang region, and caused numerous avalanche fatalities on the famed Annapurna trekking circuit.  When I visited Langtang Valley after the earthquake, most of our stations appeared to be destroyed by earthquake-triggered avalanches.  However,after recovering the dataloggers and downloading the data, it became clear that some were toppled in the middle of winter by avalanches that also took out entire yak herds (according to the locals I spoke to).

The stations that remained standing during the winter told the story of near-continuous snowfall from early March to 25 April (Figure 2). If snowfall during the typhoon is included, snowpacks may have been up to 3.5 m at high elevations. This snowpack only needed a trigger to go off, and on 25 April at 11:57, it got two: one from the ground motion, and one from the glacier ice cliff collapses.


Our results have been shared with both local and government officials, but hotels have already started to appear in the rubble that lies over the avalanche.

Press release (English)


APECS webinar on work + life in the Himalayas

On the second anniversary of the Gorkha Earthquake, I gave a webinar on my time based in Kathmandu for the Association of Polar Early Career Scientists (APECS). It was primarily a research talk, but targeted towards early career researchers who may be contemplating a non-traditional position (i.e. not a post-doc or a tenure-track position). To keep the talk interesting I also got to talk about mountain biking, being in a Kathmandu cover band, and my personal experiences in dealing with the earthquake.

Its also my first webinar, and I have to admit that its a bit disconcerting to talk to 40 or so people without any feedback or visual clues as to how you’re doing. If you’re interested, the presentation and audio can be seen here:

UAV work in the Canadian Rockies

Believe it or not, its not as easy as UAV work in the Himalayas. My UAV partner-in-crime Philip Kraaijenbrink (PhD candidate at Utrecht University) recently joined the Coldwater Lab in Canmore, Alberta, for a few weeks. Here is his honest assessment:

Our goals with these repeat surveys during the melt season (April – ????) are to develop estimates of snow melt rates from a variety of slope aspects and angles, to look at thermal aspects of snow melt (enhanced melt rates and heat advection from vegetation, rocks), and to build datasets that can be used to validate distributed snow hydrology models.

We’ve only managed to sneak in one flight so far, but the new S.O.D.A. camera from senseFly seems to work very well with the eBee RTK.


A rare moment of relative calm, jacket thrown over the laptop to try and cut the glare from the snow.


From our first flights of 2017: a closeup of the digital elevation model with shading (left) and the same scene shown as an orthoimage (right). Sleds show up nicely, and you can even see our footprints in the snow heading to the upper right corner of the scene.


As of 1 December 2016, I will be joining the Coldwater Laboratory in Canmore, Alberta, as a Research Scientist!


New horizons! Canmore, from Ha Ling Peak. Our new office is somewhere down there.

The Coldwater lab is part of the Centre for Hydrology at the University of Saskatchewan. I am really excited about the opportunity to work with Dr. John Pomeroy and his crew of students, post-docs, technicians, and colleagues. Our work will focus on snow and ice hydrology, mountain hydrometeorology, forest hydrology and climate change. 

I am still involved with the International Centre for Integrated Mountain Development (ICIMOD, Kathmandu) as a Visiting Scientist, and look forward to continuing the collaborations and friendships I developed during my four years in Nepal.

My new position will keep me closer to family and brings me back to the mountains I cut my teeth on during my university years and my M.Sc. It’s great to be back home!





IGS 2015 Poster: On-glacier weather station observations

Here’s a poster I presented last year at the ICIMOD-International Glaciological Society symposium in Kathmandu. The time series plots were done in R ggplot, the study area map in QGIS, and everything was put together in Inkscape.

The goal was to clearly and simply present some initial results from our high-altitude on-glacier meteorological station. Standing beside the poster, it was easy to point readers to the interesting highlights (surface albedo, monsoon onset and withdrawal). As chair of the local organizing committee, it was great to just stand beside the poster and talk science for a few hours – but I think the poster design provides opportunities for discussions.  71A1604.png

Glacier Collapse, Tibetan Plateau

[Put this together with Simon Gascoin (CNRS – Toulouse) after a request from Brian Kahn, at Climate Central, @blkahn]

Recent posts by NASA and Nature describe a mysterious glacier collapse on the Tibetan Plateau:


Figure 1: Sentinel-2 imagery acquired 24 July 2016, after the event (NASA; S. Gascoin)

This event really has no precedent, and no apparent trigger. Possible mechanisms for the collapse could be earthquake, rockfall on the glacier, or excessive melt or precipitation leading to extremely high water pressures at the glacier bed. There are no reported earthquakes in the region around the timing of the event (17 July; A small mass movement on the southeast slopes above the glacier appear as a dark streak in the post-event imagery, but it doesn’t appear to be big enough to trigger the ice avalanche, and may have occurred because of the avalanche.

Is there a climate signal in this event? Right now the answer appears to be no. Excessive melt does not appear to be likely, as surface temperatures based on NCEP/NCAR reanalysis are only slightly above normal (+1 to +2C) in the region from June 2016 – July 2016 (Figure 2). Much higher temperature anomalies occurred in the Pamirs and Karakoram region (but no glacier collapses were reported there).  The Global Precipitation Mission satellites show decent precipitation in the region in the 24-48 hours prior to the event (Figure 2), but not at the site. Local measurements would help confirm if the event was triggered by precipitation.


Figure 2: Surface air temperature anomalies [C] for May-June-July 2016. Departures from 1981-2010 climatology (NCEP/NCAR)


Figure 3: Mean precipitation rate [mm/hour], 16-17 July 2016, from GPM (source:

The glacier is heavily crevassed, or broken up, above the detachment point (Figure 4), and it appeared to be in a similar shape before the event. Beyond this basic information there are no clues. But low-angle glaciers generally don’t just slide off mountains, so the field investigations and local data will be really important for determining the cause of this disaster.

The risk of natural hazards is amplified in the mountains and by the mountains. And climate change generally acts to enhance these risks even further. In the Himalayas, glacier changes are leading to the formation of lakes that can pose downstream flood risks (e.g. Dig Tsho, Nepal); the loss of glaciers can reduce the stability of mountain slopes and lead to landslides; and extreme precipitation events can cause severe, rapid, and widespread flooding (Pakistan 2010 and Uttarakhand 2013). And as we saw in Nepal in 2015, and now here on the Tibetan Plateau, glaciers can release deadly snow and ice avalanches.

If this event is climate related, its another ominous sign for the future.


Figure 4: Sentinel-2 imagery of the Aru Co slide, acquired 21 July 2016, and enhanced for better contrast in glacierized areas (S. Gascoin)