• There is an error in the version 2 near-real-time (NRT) SPIReS data. The dust concentration in the V2 netcdf files in the dust_concentration variable is correct; however, other variables that use the retrieved concentration are incorrect. This includes the albedo with darkening (albedo_dirty_terrain_corrected and albedo_dirty_flat), visible snow darkening (deltavis), and radiative forcing (radiative_forcing). We are working on correcting the issue. These variables should not be used until this problem has been corrected. Read more...

Snow Today

Daily images of snow data and seasonal analyses

Get Data

Snow Today data files can be accessed via FTP at 

dtn.rc.colorado.edu/shares/snow-today

To access the data using an FTP client (e.g. FileZilla, Cyberduck, FireFTP, WinSCP, etc.), please use the following information as input:

Host Name: dtn.rc.colorado.edu
Username: anonymous 
Password: pwd 
Directory: /shares/snow-today

Known data issues

6 November 2026: NRT SPIReS V2 data albedo, snow darkening, radiative forcing error

We found a bug in the section of the code that ingests the dust_concentration values to calculate dirty albedo (albedo_dirty_terrain_corrected and albedo_dirty_flat), visible snow darkening (deltavis), and radiative forcing (radiative_forcing). The dust values are incorrectly scaled by 1e3 instead of 1e6. The result is that the dust values used in the calculation of the other variables are 1000x dustier than they should be. 

There are two reasons why it is not possible to fix these errors with a simple dust scaling correction. The first is dirty albedo, snow darkening, and radiative forcing are also influenced by illumination angle, grain size, etc, which interact nonlinearly with dust concentration. Second, when the dust, illumination, and grain size information is passed to the lookup table (LUT) to calculate the downstream variables, the LUT clips out-of-range values to the min/max value in the table. This clipping behavior, combined with the incorrect scaling, transforms even tiny amounts of dust into the maximum dust value in the LUT, erasing any spatial patterns of dust in those variables that were correctly detected over the landscape.        

For example,  we compare two scenes: a clean snow example for tile h09v05 from 2020-03-16 and a dusty snow example from 2022-04-22 as determined by the Colorado Dust on Snow Program (CODOS) records. For these dates, we recalculate the albedo using the correctly-scaled dust concentrations and compare the results to the variables with the incorrect dust. The table below shows the results of this comparison.  Note that the ”cleaner” day (less dust) is more impacted than a “dirty” day.

DateSceneErroneously introduced 
Dust (dust_concentration)
Effect on AlbedoEffect on Delta VIS 
(snow darkening)
Effect on Radiative Forcing
2020-03-16Clean+620 ppm−6.42 pp+8.65 pp+51.55 W m⁻²
2022-04-22Dusty+317 ppm−3.91 pp+4.26 pp+25.74 W m⁻²

Effectively, we introduced a dark bias of about 6.5% in clean-example albedo and 4% in dusty-example albedo, thus overestimating radiative forcing by more than 50 W/m2 in the clean scene. The effect is more pronounced in the clean example because the correctly retrieved dust concentration is low, so there's a larger change when it gets clipped to 1000 ppm in the albedo LUT. In the dusty snow example, much of the correctly retrieved pixels have dust in the 400-600 ppm range already, so it's a smaller change when clipped to 1000.

We are working to reprocess the entire time series with the correction. Until we do so; we do not recommend using the albedo_dirty_terrain_corrected, albedo_dirty_flat, deltavis, or  radiative_forcing variables.

Learn more about these data sets

Historical MODIS/Terra L3 Global Daily 500m SIN Grid Snow Cover, Snow Albedo, and Snow Surface Properties. (SPIRES_HIST, Version 1) 

Near Real-Time MODIS/Terra L3 Global Daily 500m SIN Grid Snow Cover, Snow Albedo, and Snow Surface Properties. (SPIRES_NRT, Version 2).