LAND SURFACE TEMPERATURE VALIDATION WITH IN SITU MEASUREMENTS Group 7 Juan Manuel González Cantero Irene Grimaret Rincón Alex Webb Advisor: Darren Ghent
Research problem The project task is to design an experiment to validate clear-sky land surface temperature (LST) observations from space using ground-based measurements.
What is land surface temperature? Land surface temperature is the radiative skin temperature of the land. Land surface temperature is not the same as air temperature and can vary up to 10K over a few meters. It is influenced by conditions such as precipitations and albedo.
Motivations LST is important for assessing the surface energy balance between the atmosphere and the ground. LST is required for a wide variety of climatic, hydrological, ecological and biochemical studies.
Aim The aim is to validate satellite data with ground-based measurements so we can trust on that data. Ground measurements are point measurements and satellites measure areas, so we need to relate them.
Satellite Data SEVIRI can measure LST from geostationary orbit with 15min resolution. In the UK has ~6km spatial resolution.
Satellite Data ATSR and MODIS have 1km of spatial resolution. They have a lower temporal resolution than SEVIRI.
Case-study Area A 1km 2 region has been arbitrarily chosen because ATSR and MODIS have this resolution. This region has been chosen near to the University of Leicester for a hypothetical analysis.
Methodology - Overview Three IR radiometers were available to measure LST on the ground, so it was necessary to divide our region into three land types. The LST of each type of land needed to be calculated as well as the percentage of each land type.
Methodology Land Classification In a Google Earth picture, the principal land types are: Trees Urban Grass A Landsat image has been downloaded and opened in the program ENVI.
Methodology Land Classification
Methodology Land Classification Blue zone: Deciduous trees Total area: 7.5%
Methodology Land Classification Red zone: Urban (Paving concrete) Total area: 63.2%
Methodology Land Classification Green zone: Green Grass Total area: 29.3%
Methodology Field work
Methodology Field work
Methodology Field work
Methodology Field work
Methodology Field work Need to measure the brightness temperature of all surfaces at the same time when the satellite passes. Measurements are taken every second and averaged over a minute.
Methodology Field work A clear sky brightness temperature is also measured.
Methodology Problems The sky wasn t clear and it was raining.
Methodology Problems The sky wasn t clear and it was raining
Methodology Problems The sky wasn t clear and it was raining
Methodology Problems The sky wasn t clear and it was raining. The satellite cannot measure below clouds so will measure an incorrect LST if it is cloudy.
Methodology Problems The sky wasn t clear and it was raining. The satellite cannot measure below clouds so will measure an incorrect LST if it is cloudy. All urban zones were assumed with the same properties.
Methodology Problems The sky wasn t clear and it was raining. The satellite cannot measure below clouds so will measure an incorrect LST if it is cloudy. All urban zones were assumed with the same properties. Shadow regions have different LST s to sunny regions, but we assumed all zones were in sunlight.
Methodology Problems The sky wasn t clear and it was raining. The satellite cannot measure below clouds so will measure an incorrect LST if it is cloudy. All urban zones were assumed with the same properties. Shadow regions have different LST s to sunny regions, but we assumed all zones were in sunlight. Landsat has 30m resolution which makes the classification of the individual trees difficult.
Instrument calibration Each radiometer degrades over time so a correction needs to be applied. y 1 = 0.9634x 1 + 2.6206 y 2 = 1.0273x 2 + 2.2821 y 3 = 1.0862x 3 + 2.4422 Where x i is the uncalibrated BT in the instrument i and y i is the calibrated BT.
Converting BT to LST B c T c = ε c B c T sss + 1 ε c B c T sss Where B c T c is the emitted radiance given by the Planck function for an effective BT in the channel c. B c T sss is the emitted radiance for surface temperature. ε c is the emissivity of the surface. B c T sss is the radiance for the effective BT of the sky.
Converting BT to LST The emissivity values for each surface were taken from the ASTER Spectral Library. The Library gives the emissivity in each wavelength and it is necessary to calculate an average in the range of the instruments.
Results Once the LST for each region has been calculated they need to be weighted by the surface fraction in the region: T sss = W tttt T sss,tttt + W ggggg T sss,ggggg +W uuuuu T sss,uuuuu Using this method the LST of the 1km 2 region is calculated as: T sss = 29.1
Interpretation of results The urban LST was calculated as 32.2 and the grass and trees were 24.2 and 22.4. Looking at the Landsat LST image we can see that the park is colder (darker) than the urban zones. This agrees with our results. Victoria Park
Conclusion We have successfully demonstrated this method. This method is applicable to other places as well as Leicester. The LST calculated could not be compared with satellite data in this study because it was cloudy, but the method would work on a clear day.
Further work To repeat this experiment with more land classification types on a clear day and compare this results with actual satellite data.
References Prata, 1994. Trigo, 2008. Wan, 1996. ASTER spectral library, http://speclib.jpl.nasa.gov/
Thank you!! Any question?