Improving satellite estimates of global snowfall

The dawn of satellite meteorology in the 1960s substantially improved scientists’ ability to collect atmospheric data from the whole globe and led to better understanding and forecasting of weather events. One might assume that modern satellites can detect every rain or snow event on Earth, but no satellite is perfect. High-latitude snow, especially light snowfall over the ocean or from near-surface clouds, poses a problem for satellites.

Such snowfall may be light, but it’s not insignificant. Detecting and quantifying all global precipitation, including very light snow, gives us a more accurate picture of the water cycle and Earth’s energy budget.

Today meteorologists are getting a clearer picture than ever with the help of a new generation of satellites. Those Earth-orbiting probes are aiding day-to-day forecasts, especially when storms move onto land from bodies of water. They also afford more accuracy in climate predictions, which is critical for understanding the future of precipitation, coastlines, and ice sheet loss.

The key to detecting and estimating precipitation is good radar coverage. Radar devices emit microwave pulses that bounce off the rain or snow particles in the atmosphere. The intensity of pulses that return to the radar, known as reflectivity (Z), tells us about the size, shape, and concentration of rain or snow particles in the radar’s line of sight. Then, through complex relationships that are research topics in their own right, reflectivity can be converted to a precipitation rate.

Ground-based radar networks around the world provide crucial information for tracking storm systems, creating short-term forecasts, and estimating precipitation amount. Unfortunately, ground-based radar devices are generally confined to land, and each one’s range is limited to just a few hundred kilometers. For a global picture of precipitation, we must turn to spaceborne satellite radar and other remote sensing tools.

CloudSat , a NASA satellite launched in April 2006, demonstrates the essential capabilities of spaceborne radar and has enabled the first studies of near-global snowfall from a spaceborne instrument. With an orbit from 82° S to 82° N latitude, CloudSat‘s Cloud Profiling Radar (CPR) can probe the atmosphere over nearly the whole globe, including remote high-latitude regions. It also provides detailed information on the vertical structure of what it observes. The CPR’s low reflectivity threshold (–29 dBZ) allows the instrument to identify light rain and snowfall events.

Because of CloudSat‘s unique capabilities, atmospheric scientists have begun to use the satellite to survey different snowfall types. For instance, shallow convective snow―say, lake-effect snow―occurs frequently in many parts of the world (figure 1), including over oceans where ground-based observations are limited. That type of snowfall is nonnegligible: Up to 50% of snowfall in certain oceanic regions comes from shallow convection. Using CloudSat, scientists can globally quantify the frequency of shallow convective snowfall occurrence for the first time.

Because CPR is very sensitive, it also excels at measuring snowfall events occurring at very cold 2 m temperatures (temperatures 2 m above the surface) of 255 K or less (figure 2), where snowfall is often very light and difficult to detect.

CloudSat‘s many years of data collection, paired with data from the Advanced Microwave Scanning Radiometer–Earth Observing System (AMSR-E ) instrument on the nearby Aqua satellite, has set the stage for the new Global Precipitation Measurement (GPM) satellite mission. Launched in February 2014 as a joint mission of NASA and the Japan Aerospace Exploration Agency, GPM has a primary goal of providing more sophisticated precipitation observations to help scientists better understand the global hydrologic cycle, including snowfall at latitudes as high as 65° N and 65° S.

The satellite determines precipitation by measuring the atmosphere at multiple frequencies. Satellite radiometers take passive measurements by simply observing the radiation emitted or scattered toward the satellite at multiple frequencies. Those measurements are known as brightness temperatures. Radar devices, on the other hand, take active measurements by sending out pulses at specific frequencies and detecting how the return pulses have been altered by particles in the atmosphere. The Goddard Profiling (GPROF) precipitation retrieval algorithm compares the GPM multifrequency readings with an a priori database of millions of measurements of brightness temperature and coincident radar-derived precipitation rates.

Immediately postlaunch, the GPROF algorithm used snowfall rates from CloudSat‘s CPR and brightness temperatures from Aqua‘s AMSR-E for 2 m temperatures of 255 K or less. In the future, GPROF will be updated to use a database of coincident brightness temperatures and radar reflectivities measured by its own instruments. GPM‘s scanning radar and high-frequency radiometer—a new combination that makes for better sampling of the atmosphere—will enable atmospheric scientists to sharpen the focus on global snowfall already gleaned from CloudSat.

Using CloudSat data, my colleagues and I have analyzed how frequently and in what ambient environment snowfall occurs within GPM‘s latitude boundary. The joint histogram in figure 3 shows snowfall event occurrence with respect to 2 m temperature and total precipitable water (TPW; the depth of water in a column of air, if it all rained out).

The shape of the histogram can be partly explained by the Clausius–Clapeyron relationship: As the air temperature increases, so does the water vapor content. There are instances of snowfall at relatively high temperatures and low TPW, though not vice versa. Snowfall events do occur at 2 m temperatures below 255 K. Though they are only a small fraction of total snowfall events, that type of snowfall is critically important to regions like Antarctica and Greenland, which are covered in ice.

The very cold snowfall events that CloudSat has identified within the GPM latitude boundary occur mostly in the Northern Hemisphere, which is also where much of the human population resides. GPM might be able to identify those cold snowfall events using the postlaunch algorithm. In the future, when the algorithm starts to use the satellite’s own observations exclusively, researchers will investigate how effectively GPM observes snowfall in cold environments, especially since snowfall reflectivities may be much smaller than the minimum detectable reflectivity (12 dBZ) of GPM‘s radar.

The next step, then, is to analyze the GPM snowfall retrievals derived from the postlaunch algorithm. Creating a global map of GPM-observed snowfall will help determine how well the GPROF algorithm identifies and estimates snowfall, especially for snowfall occurring over very cold surfaces and shallow convective over-ocean snowfall. The synergistic use of the CloudSat and GPM data sets will foster a new era of observation-based global snowfall measurements, which will help improve scientists’ understanding of Earth’s water cycle and energy budget and improve forecasting.

Marian Mateling is a PhD student working with Mark Kulie and Tristan L’Ecuyer at the University of Wisconsin–Madison. She focuses on assessing and improving satellite detection of frozen precipitation. Additionally, she uses both models and observations in her research of the energy budget in polar regions.