Tracking Climate Changes with Satellite Imagery

I was just reading a NASA article about massive glacial melting in Greenland. What caught my attention, more than anything, was the incredibly short time scale over which major increases in warming and melting were observed – 4 days! Rarely do we get such a drastic change over a continental mass in such a short time. It just goes to show how precarious things are and how closely some climate systems teeter on the brink of stability (others, like the middle of the Sahara, are arguably more in the center of their stability zone).

Melting snow and ice in Greenland. Time difference between the two images is only 4 days. Image courtesy: NASA.

 

Researchers used a combination of satellite imagery including Indian Space Research Organizations (ISRO) OceanSat-2 and MODIS from NASA’s Terra and Aqua satellites. Results were confirmed with passive microwave data aboard a USAF meteorological satellite. Here’s a quick rundown of how these satellites can detect the glaciers melting: MODIS is a keystone instrument for global studies of atmosphere, land, and ocean processes. It has 36 bands with bands 1-19 and band 26 in the visible and near infrared range, and remainder bands in the thermal range from 3 to 15 mm. It provides daylight reflection and day/night emission spectral imaging of any point on the Earth every 1-2 days. Bands 1-19 are used in measuring solar reflectance, giving us properties like albedo and helping us assess land cover classes (vegetation, open water, desert, etc.) and land cover changes. The remainder of the bands in the mid- and thermal-IR portion of the electromagnetic spectrum are useful in determining land cover temperature, sea surface temperature and for measuring and correcting for atmospheric effects.

Click here if you want to know the gory details of the science behind the MODIS algorithms – it’s pretty interesting!

 

Using specific thermal and IR bands, MODIS can measure Land Surface Temperatures (LST) Oceansat-2 has a scanning scatterometer. Looking at the backscatter of the Ku-band operating at 13.515GHz (active microwave) you can observe different properties between dry and wet snow. When the snow becomes wet (in this case due to melting) the backscatter decreases significantly because water is highly absorbent in radar frequencies. The figure below illustrates nicely what the difference between wet and dry snow looks like.

Two views of a portion of Greenland ice sheet showing contrast in radar backscatter between wet and frozen conditions. Image courtesy: NASA-JPL

 

The drastic melting in Greenland was discovered by Son Nghiem of NASA's Jet Propulsion Laboratory who was analyzing Radar data from Oceansat-2 and noticed almost the entire region had undergone some degree of melting. After double checking to make sure there was no data error, he contacted his colleague at NASA Goddard, Dorothy Hall, who took a look at some MODIS Land Surface Temperature data which also showed some abnormally high temperatures in the region. This also coincided with a large heat dome that had parked itself over Greenland for the better part of a week at that time, following several earlier anomalously high heat events earlier in the summer. The Radar results were verified using another Air Force meteorology satellite for a third data point. Melting events like this have occurred in Greenland in the past, about every 150 years according to ice cores. The last such event was back in 1889, so this event is roughly on the correct time frame, and will hopefully be an isolated event. Continued episodes like this could have enormous impacts on the North Atlantic climate, impacting ocean currents and atmospheric weather patterns. These sensors provide regional, large scale coverage on the order of kilometer scale pixel sizes. This type of coverage is excellent for global and continental scale global monitoring. In fact, it’s necessary for applications of that scale – if we used higher resolution sensors we’d have an unnecessary amount of data to deal with. When we look at hurricanes through AVHRR, it still takes a lot of coverage to capture the whole thing. If we did that with meter scale sensors, the data volume would be almost unmanageable for most systems.

Different resolutions of the same area. But, there are definitely applications for the high resolution imagery.

 

While understanding continental scale climate patterns like in Antarctica or Greenland, it is necessary to understand and measure changes at a much smaller scale. Commercial imagery like from GeoEye and DigitalGlobe are excellent for this type of application – they can provide us details ready for analysis in a local GIS. A great example of this is the South Cascade Glacier in northern Washington. This is a small alpine glacier that has retreated significantly in the last 50 years and is a site of detailed in situ and remotely sensed analysis.

Photographs of the South Cascade Glacier, which has lost nearly half of it's volume since 1958.

 

GeoEye imagery showing satellite images of the South Cascade Glacier from 2002 (upper right), 2004 (UL), 2007 (LR), and 2011 (LL).

 

 

Using Image Analysis and GIS it is easy to map area extent changes in the glacial toe over the 9 year time span. Using standard remote sensing techniques for measuring albedo and land cover properties, it is very easy to map exactly fast how the glacier his retreating.  This meter scale data is far too high resolution to be useful over the scale of an area like Greenland, but it is perfect for using it on areas like the South Cascade Glacier in which changes are happening on the order of 1, 5, or 10 meters.  So, different data types have different value depending on the nature of the study.

 

References:

https://earthobservatory.nasa.gov/images/78607/satellites-observe-widespread-melting-event-on-greenland

https://modis.gsfc.nasa.gov/data/atbd/atbd_mod11.pdf

http://www2.bren.ucsb.edu/~dozier/Class/esm236/Reading/Konig.pdf

https://glaciers.research.pdx.edu/south-cascade-glacier