MSG Maximum and Minimum Surface Temperatures Data: download version 1.0.1

Version History (current version shown in bold)

You are strongly advised to read this paper before using the data. Data are provided in NetCDF files and contain the following variables for each pixel on each day (click on links for further information about each variable):

• Minimum LST
• Maximum LST
• Minimum air temperature
• Maximum air temperature
• Error on minimum LST
• Error on maximum LST
• Observation fraction for minimum LST
• Observation fraction for maximum LST
• Error on Tmin
• Error on Tmax

Pixel latitudes and longitudes are not included in the NetCDF files in order to conserve space, as the geolocation grid is constant from one day to the next. The file msg_geo_euro.nc.gz (1.2 MB) contains latitude and longitude information for each SEVIRI pixel in the files below.

Data have been processed for the years 2009 to 2015 (up to 23/02/2015), inclusive. However, we have found some problems with the data between 2009 to 2011 so these data are not included here. If you would like to use these data, please email elizabeth.good@metoffice.gov.uk.

Minimum LST

This is the minimum cloud-free LST corresponding to the date of the file. This is obtained from the operational LSA SAF LST product by analysing all available slots between the local time period 23:00 - 08:00 (23:00 is the calendar day before the time stamp for the file). Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Maximum LST

This is the maximum cloud-free LST corresponding to the date of the file. This is obtained from the operational LSA SAF LST product by analysing all available slots between the local time period 11:00 - 16:00. Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Minimum air temperature

This is the estimated minimum air temperature for the day. This is calculated from a regression-based analysis using the minimum LST and other variables as predictors. Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Maximum air temperature

This is the estimated maximum air temperature for the day. This is calculated from a regression-based analysis using the maximum LST and other variables as predictors. Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Error on minimum LST

This is the error provided by the LSA SAF LST product used for the minimum LST. Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Error on maximum LST

This is the error provided by the LSA SAF LST product used for the maximum LST. Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Observation fraction for minimum LST

This is the % of the maximum number of possible observations that are cloud-free and valid during the analysis window used for the minimum LST extraction. (Maximum possible number of available slots between 23:00 and 08:00 is 37.) Units are in %, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Observation fraction for maximum LST

This is the % of the maximum number of possible observations that are cloud-free and valid during the analysis window used for the maximum LST extraction. (Maximum possible number of available slots between 23:00 and 08:00 is 21.) Units are in %, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Error on Tmin

This is the standard error on the calculated Tmin value (i.e. the square-root of the quadrature sum of the error on each predictor variable for each pixel multipled by it's regression coefficient + the uncertainty on that regression coefficient). These errors should be treated with care as they are unlikely to account all aspects on the uncertainty for Tmin. Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Error on Tmax

This is the standard error on the calculated Tmax value (i.e. the square-root of the quadrature sum of the error on each predictor variable for each pixel multipled by it's regression coefficient + the uncertainty on that regression coefficient). These errors should be treated with care as they are unlikely to account all aspects on the uncertainty for Tmax. Units are in K, with a scaling factor of 10 (i.e. divide by 10 to get correct numbers).

Difference between current and previous versions of this data set

Maintained by: Elizabeth Good Last updated: 16/03/2015 Hosted by: Met Office Hadley Centre

© Crown Copyright