MRMS: Observations and Radar Reimagined

MRMS: Observations and Radar Reimagined

November 20, 2018 0 By ThielWx

If you have seen the Projects Page recently, you might have noticed that the research I did at Ohio University, in addition to work that I will be doing at the University of Oklahoma, both involve the Multi-Radar/Multi-Sensor (MRMS) System. Due to its recent deployment in NWS offices, many have yet to hear of this system, what it does, and the suite of products that are offered within operations. This article will hopefully shed some light on MRMS, and show why I’m excited to continue working with it!

Picture This….

You’re working a shift on radar at your favorite office NWS Podunk, watching a line of thunderstorms on the edge of your warning area. Naturally your radar is further away than you would like, and you think you see some a storm that has the potential to produce damaging hail.  Uncertainty is high due to degraded resolution horizontally and vertically. “What’s the freezing level again?” you think as seconds tick away, eating into the lead time and potential effectiveness of a warning. While this is a highly simplified example, it speaks to several key advantages to MRMS.

Multiple Observations-One Platform

As the name suggests, the MRMS system takes in data from multiple sources, including surface observation stations, lightning data sensors, NWS radars, and the Rapid Refresh (RAP) weather model. The data is “fit” to a 1km by 1km grid (approximately), which covers the entire Continental United States (CONUS). Above the surface, grid points extend upward to 20km at 33 predefined levels. After some multiplication, this means there are over 800 million grid points within the MRMS domain that update every two minutes, leading to a high degree of resolution spatially and temporally.


A snippet from the MRMS Factsheet, which shows how integrating multiple radars into a single domain can improve the quality of data available to forecasters. (Click the image to access the MRMS Factsheet)

Of course not all observations are continuous in space, radar is an easy example. Between each elevation scan, there are gaps in the radar data that need to be filled in to accurately represent the environment. Data near each grid point are then weighted based upon their values and distance from the parent radar. Once you have a spatially continuous grid with all of your observations, this is where the fun begins…

Creative and Useful Products Galore!

Due to the scope of products involved with MRMS, we will shift our focus to radar-derived products that have a special advantage. When looking at radar data, it is important to not only look at the values, but also how high up they are with respect to the near storm environment. One key level is called the freezing level, where the atmosphere becomes colder than 0 degrees Celsius, can be useful for forecasting hail. But how do we determine this level? This is where temperature profiles from the RAP can be used. By knowing the height of the freezing level from these temperature profiles at each location, we can pull the reflectivity values from only the freezing level; creating a quick and easy product for NWS forecasters to use.

An example of an MRMS radar image at the freezing level.

During my undergrad research, we took advantage of this capability of isothermal reflectivities with a special product called Vertically Integrated Ice (VII). This product adds up the reflectivity values vertically from -10C to -40C, creating an estimation of the amount of ice within a vertical column. One key indicator for lightning production is the presence of ice within thunderstorms, and VII has shown promise in forecasting the onset of lightning production (Mosier et al. 2011).

An example of an MRMS VII image, corresponding to the isothermal reflectivity image above.


The MRMS system has a wide variety of applications beyond those explained here, such as aviation, rainfall estimates, etc., but hopefully this article was able to show what makes MRMS different from conventional radar and its advantages, including:

  • Higher resolution that is spatially continuous
  • Rapidly updating products for severe weather forecasting
  • Mitigation of the limitations involving of single-radar observations (cone of silence, terrain blockage, distance from radar)
  • Development of new products using multiple data types

In order to be fair, there are also limitations to MRMS, such as:

  • Data delays during active events
  • Model errors from the RAP negatively influencing radar-derived products
  • Potential for poor fitting of data to the grid

For more info on MRMS, make sure to visit the NSSL MRMS webpage. For an even further in depth look at MRMS and it’s uses in the field, an article by Smith et al. 2016 is an excellent place to start, as it’s where the foundation of my MRMS knowledge comes from!