So, we’ve finally begun the process of transitioning over fully to the new CONUS version of the SPoRT LIS. This “new” version of the SPoRT LIS has been under development actually for several years now, and underwent initial testing and evaluation at the Huntsville WFO in spring 2015, followed by an evaluation by several WFOs and RFCs in summer 2015. Image 1 below shows the differences in the domains. The new version of the SPoRT LIS encompasses the entire CONUS and surrounding areas of southern Canada and northern Mexico, albeit with some anticipated degradation especially in the border regions due to lack of consistent radar/precipitation coverage.
Image 1. The CONUS SPoRT LIS (left) and the approximate domain of the old Southeast CONUS version (right). Note: the images are from different periods.
Not only does the CONUS version offer a geographic expansion over the previous version of the LIS, but new variables are a part of the new SPoRT LIS, including 0-200 cm relative soil moisture changes on several timescales (weekly, bi-weekly, monthly, seasonal, semi-annual and annual) soil moisture percentiles and soil temperatures. The soil moisture percentiles and change values can be especially useful for the drought designation and analysis process, and have been used in this capacity at the Huntsville office since their inception. Of course, there are other applications for hydrology, fire weather and blowing dust. We’re planning to explore more of these latter unique and interesting applications with several of SPoRT’s collaborative Western CONUS WFOs next spring and summer. The SPoRT LIS soil temperature data have shown promising application for impacts during winter weather events during evaluation of a few events in the previous winter, with more evaluation expected during the upcoming winter. In addition to the new variables, the new version of the SPoRT LIS is using NSSL’s Multi-Radar Multi-Sensor data for precipitation forcing in the near term and is also solely incorporating the VIIRS GVF over the legacy MODIS GVF.
Image 2. Examples of SPoRT LIS 0-200 cm relative soil moisture weekly change (left) and 0-200 cm relative soil moisture percentile (right)
Users of the SPoRT LIS and GVF data for their local modeling purposes will need to make the appropriate changes to their EMS/UEMS model runs to properly incorporate these new data sets. Please contact Jon Case at SPoRT or me (Kris White) if you have any questions. Thanks for reading!
During the afternoon and early evening hours on 2/15/2016, a large area of rain covered much of northeast Kentucky and southeast Ohio as well as the western half of West Virginia.
An upper level disturbance then moved across the area during the evening and overnight hours with the rainfall mixing with and then transitioning to all snow.
I wanted to show how the SFR image performed during this transition. The image above is from 0220 UTC on 2/16/2016. At that time, much of the precipitation across West Virginia was still in the form of rain…with an area of snow extending from northwest Pennsylvania across central Ohio into southwest portions of that state.
There appears to be several observations of rain across Ohio with surface temperatures of 32 to 35 DegF where the SFR product indicated snow in the clouds. It does appear that where surface temperatures were warmer than 35 DegF, the SFR product did not indicate any snow in the clouds.
From an earlier post, I believe the SFR throws out snow when the model-based 10-m temperatures exceeded 33 DegF. Is this filter working in this situation?
On Jan 26 2014, an upper level shortwave caused an area of light snow across Ohio, western Pennsylvania and the northern counties of West Virginia. Surface temperatures were quite cold with readings generally in the teens. Even at these cold temperatures, the SFR product did indicate snowfall across the far northern counties of our forecast area.
The maximum snowfall rates indicated on the 1605 UTC product was about 0.3 to 0.4 inches per hour. Based on reports, these numbers appear to be representative of what actually was occurring.
While this is just one case, the SFR product appears to work reasonably well at temperatures below 22 DegF.
On Jan 25 2014, a mid-level shortwave moved across the region generating light to moderate snow. I have included screen captures of the 1118 UTC regional radar mosaic and surface observations…along with a 1120 UTC Snowfall Rate Product and surface observations.
It looks like the SFR product did not detect all of the snow that was falling around 11 UTC. But the misses can generally be described as either (1) the surface temperatures being too cold or (2) the probabilistic model, that is part of the calulations, indicating probabilities that were too low to determine if there was snow.
Once you know all of the details on how the product is calculated, I think this product did a good job at detailing where the snowfall was occurring.
The highest snowfall rates indicated by this image was around 0.3 to 0.5 inches which seems to be representative of what was occurring.
When I examined the 1522 UTC SFR product, I noticed there was an absence of snow across our forecast area. Radar and surface observations indicated that light to moderate snow was continuing across most of our counties.
Per the Quick Guide, I checked the surface observations to see if the temperatures were about 22 DegF or colder. The temperatures across our northern and western counties were actually 22 DegF or colder. So the SFR product was behaving as it should across those counties.
However, the temperatures across the remainder of our region were above 22 DegF. The snow is definitely not lake effect as the current snow was still related to a shortwave which had pushed to our east.
What could be causing the lack of indicated snow across the portions of our area that still had surface temnperatures above 22 DegF?
I have attached a screen capture of the SFR product from 1024 UTC on 1/21/14. The label on the image is wrong. It states the units of the product are in/hr. But they are actually mm/hr.
During this time, we were having widespread light to moderate snow as an upper level disturbance moved across our forecast area. Reports around 2 inches of snow were common around the time of the product. We had received reports of snow coming down around an inch per hour. The maximum SFR detected in the product was 1.6 mm/hr…or 0.06 in/hr. Using a ratio of 15:1 yields a maximum snowfall rate around 0.9 inches per hour.
While we had several surface observations from which we could estimate precipitation rates, our WSR-88D was not operating correctly. The legacy precipitation were okay. But the Dual Pol precipitation products were not totally reliable due to equipment issues. So the additional information from the SFR product should have helped estimate the precipitation rates.
Effective 10 September, the real-time SPoRT-LIS running over much of the southern and eastern U.S. was upgraded with several improvements.
The upgrade is transparent to Environmental Modeling System (EMS) end-users, since file and data formats are the same and the EMS processing with the “lis” land surface model (LSM) option operates the same as before. However, it is highly recommended that EMS end-users currently running the “lis” option consider changing to the land-use database described in the 2nd bullet below.
The most noteworthy modifications and improvements are:
- Updated LIS software to support an upgrade from Noah LSM version 2.7.1 to version 3.2. This upgrade includes an improved look-up table methodology for some static fields and improved handling of heat fluxes over snow-covered regions.
- Changed land-use classification (vegetation type) from the U.S. Geological Survey (USGS) 24-class database to the newer International Geosphere Biosphere Programme (IGBP)/MODIS 20-class database. The IGBP/MODIS database is more up-to-date than the USGS database, especially with urban classifications.
- Switched from a coarse-resolution surface albedo climatology to a look-up table methodology for surface albedo based on (a) input Green Vegetation Fraction (GVF) data from the high-resolution SPoRT-MODIS real-time product and (b) the newer IGBP/MODIS land-use database. A sample real-time SPoRT-MODIS GVF map projected onto the 3-km LIS domain is given in Figure 1, showing a comparison between the monthly climatological GVF and the real-time MODIS GVF data from 30 August. An example comparison between the original climatological specification of surface albedo and the newer look-up table methodology using real-time SPoRT-MODIS vegetation data is given in Figure 2 from the same day. Both of these upgrades will improve the surface energy budget in the real-time LIS.
- Modified the long-term atmospheric forcing (excluding precipitation) that drives the LIS-Noah LSM integration from the North American Land Data Assimilation System (NLDAS) to NLDAS phase 2 (NLDAS-2).
Contact SPoRT for the official upgrade documentation for further details.
Figure 1. Comparison between the default monthly climatological Green Vegetation Fraction (GVF, in percent) time-interpolated to 30 August (left), and real-time SPoRT-MODIS GVF on 30 August 2012 (right). Note the much lower GVF over the Midwest in the SPoRT-MODIS dataset corresponding to the substantial drought. (Click image twice for full size)
Figure 2. Comparison between climatological surface albedo (%) time-interpolated to 30 August in the former LIS configuration (left), and surface albedo as a function of the real-time SPoRT-MODIS GVF in the upgraded LIS configuration (right). Note the higher surface albedo corresponding to lower SPoRT-MODIS GVF in the Plains and Midwest regions. (Click image twice for full size)