Land Surface Temperature

Land surface temperature (LST) is the mean radiative skin temperature of an area of land resulting from the mean balance of solar heating and land-atmosphere cooling fluxes. It is a basic determinant of the terrestrial thermal behaviour, as it controls the effective radiating temperature of the Earth’s surface. The sensitivity of LST to soil moisture and vegetation cover means it is an important component in numerous applications. With the demand for LST data from Earth Observation currently experiencing considerable growth it is important that the users of this data are appropriately engaged by the LST data providers.

The GlobTemperature project under the Data User Element of ESA's 4th Earth Observation Envelope Programme (2013-2017) aims to promote the wider uptake of global-scale satellite LST by the research and operational user communities; the key to success depending on the coherence and openness of the interactions between the LST and user communities. By incorporating detailed user input into the specifications, their subsequent testing of the LST data sets, and sustained access to data in a user-friendly manner through common data formats GlobTemperature is enhancing the portfolio of LST products from Earth Observation, while concurrently breaking down the barriers to successful application of such data through its programme of dialogue between the data providers and data users.

Here we present current outcomes from the project, which is achieving some innovative developments: a globally representative and consistent matchup database enabling validation and intercomparison of multi-sensor LST data sets; a combined geostationary earth orbit (GEO) and low earth orbit (LEO) global data set for LST to resolve the diurnal cycle which is a key request from users of LST data; a first Climate Data Record for LST from the ATSRs; the dissemination of LST data sets via a dedicated Data Portal in harmonised data format; and the establishment, in collaboration with international colleagues of a first working group (ILSTE-WG) dedicated to LST and Emissivity, whereby user evaluation of products by climate services aims to provide a thrust to sustained operational support of this group meeting a critical need amongst users of LST data.

Passive microwave measurements are less affected by clouds than IR observations. The Land Surface Temperature (LST) can be derived from microwave measurements, regardless of the cloud conditions. However, retrieving LST from microwave is still challenging. Contrarily to the infrared emissivities, the microwave emissivities vary strongly with surface properties (soil moisture, vegetation, snow) and show spatial and temporal patterns. In addition, the spatial resolution of the microwave observations degrades with decreasing frequencies, whereas the atmospheric contribution to the signal increases with frequencies. Lastly, the microwave radiation can emanate from the subsurface, not from the surface skin: the lower the frequency, the larger the penetration into the subsurface.

A methodology has been developed to estimate the LST, along with atmospheric water vapor, cloud liquid water, and surface emissivities over land, from passive microwave imagers (Aires et al., 2001). It is based on a neural network inversion, trained on a large data set of simulated radiances, using real atmospheric and surface information over the globe. It makes use of first guess information about microwave emissivity (Prigent et al., 2006), coincident surface skin temperature (derived under clear sky condition from IR observations, through the ISCCP program, Rossow and Schiffer, 1999), as well as first guess information from meteorological reanalysis. The theoretical root mean square error of the LST retrieval over the globe is 1.3 K under clear-sky conditions and 1.6 K in cloudy conditions. These LSTs have been compared with in situ LST measurements (Catherinot et al., 2011). Under clear sky conditions, the quality of the microwave neural network retrieval is equivalent to the infrared ISCCP products, for most in situ LST stations, with errors ~4 K as compared to in situ measurements. The performance of the algorithm is similar under clear and cloudy conditions. However, this neural network inversion requires a large range of ancillary observations that are difficult to collect in near-real time.

Here, we will present a new retrieval approach, limiting the number of ancillary inputs that are required while maintaining the LST accuracy. An algorithm, easily and broadly applicable to past and current microwave imagers is developed, based on measurements provided by the microwave conical scanners, from 19 to 90 GHz, and pre-calculated atlases of microwave land surface emissivities, at the same frequencies. The inversion is trained on the LST retrievals obtained with the method previously described. Results will be presented and evaluated with in situ observations, over a large range of environments. The microwave LST will also be compared to different sources of infrared satellite estimates, under clear sky conditions.  

The objective of this project is to produce a long climate record of an “all weather” LST, applying the methodology to all conical scanners in the microwaves (SSM/I, SSMI/S, AMRS-E) and limiting the inconsistency related to inter-calibration problems and to overpassing local times. This new dataset will complement the available IR datasets for the cloudy cases, and better describe the diurnal cycle of the LST regardless of the cloud conditions.

Aires, F., C. Prigent, W.B. Rossow, and M. Rothstein, J. Geophys. Res., 106, 14,887-14,907, 2001.

Catherinot, J., C. Prigent, R. Maurer, F. Papa, C. Jimenez, F. Aires, and W. B. Rossow, J. Geophys. Res., 116, D23, doi:10.1029/2011JD016439, 2011.

Prigent, C., F. Aires, and W.B. Rossow, Bull. Am. Meteorol. Soc., doi:10.1175/BAMS-87-11-1573, pp. 1572-1584, 2006.

Rossow, W. B., and R. A. Schiffer, Bull. Am. Meteorol. Soc., 80, 2261–2288, doi:10.1175/1520-0477, 1999.

The Visible Infrared Imaging Radiometer Suite (VIIRS) is the primary component of the Suomi National Polar-orbiting Partnership (NPP) satellite --  launched on Oct. 28, 2011, from Vandenberg Air Force Base in California. The Joint Polar Satellite System (JPSS), a NOAA/NASA program, and the Defense Weather Satellite System, a U.S. Air Force program, were tasked with the next generation requirements for environmental research, weather forecasting and climate monitoring measurement capabilities that Suomi NPP represents. Since 2004, scientists funded by NOAA and NASA, have been evaluating and validating the JPSS VIIRS Land operational products. Prior to launch, these products were created by applying VIIRS code to Moderate Resolution Imaging Spectroradiometer (MODIS) data.

In 2014, based on this long period of product evaluation, NASA embarked on a program to produce and evaluate the Earth System Data Records (ESDRs) needed to ensure continuity with NASA EOS data products. A number of these products have their heritage in the MODIS heritage algorithms, and in some cases early versions of the MODIS code were used by the operational VIIRS algorithm development teams. Based on these two criteria, the VIIRS Land Discipline Team has established priorities and phasing plans detailing the production of Suomi NPP Land Science Products for the current round of investigations (2014-2017).

In this talk, we will present the functional description and capabilities of the NASA Suomi-NPP Land Science Investigator Processing System (SIPS), including product development phases and production schedules, timeline for processing, and delivery of Land science products in close coordination with the Suomi-NPP Land science team. The Land SIPS processing stream is expected to be operational by December 2016, generating land products either using the NASA science team delivered algorithms, or the “best-of” science algorithms currently in operation. In addition to generating the standard land science products through processing of the NASA’s VIIRS Level 0 data record, the Land SIPS processing system is also used to produce a suite of near-real time products for NASA’s applied science community. Land SIPS will also deliver the standard products, ancillary data sets, software and supporting documentation (ATBDs) to the assigned Distributed Active Archive Centers (DAACs) for archival and distribution.

Quality assessment and validation will be an integral part of the Land SIPS processing system -- with the latter being performed at Land Data Operational Product Evaluation (LDOPE) facility, and the former conducted under the auspices of the Committee of Earth Observation Satellites (CEOS) Working Group on Calibration & Validation (WGCV) Land Product Validation (LPV) Subgroup; adopting the best-practices protocols and tools needed to assess the quality of heritage EOS-MODIS products generated at the MODIS Adaptive Processing System (MODAPS).

One of the purposes of the GlobTemperature project is to provide a product of global Land Surface Temperature (LST) based on Geostationary Earth Orbit (GEO) and Low Earth polar Orbit (LEO) satellite data. The objective is to use existing LST products, which are obtained from different sensors/platforms, combining them into a harmonized product for a reference view angle. In a first approach, only infra-red based retrievals are considered, and LEO LSTs will be used as a common denominator among geostationary sensors.

We consider LST data provided by a wide range of sensors to optimize spatial coverage, namely: (i) 2 LEO sensors - the Advanced Along Track Scanning Radiometer (AATSR) series of instruments on-board ESA’s Envisat, and the Moderate Resolution Imaging Spectroradiometer (MODIS) on-board NASA’s TERRA and AQUA; and (ii) 3 GEO sensors - the Spinning Enhanced Visible and Infrared Imager (SEVIRI) on-board EUMETSAT’s Meteosat Second Generation (MSG), the Japanese Meteorological Imager (JAMI) on-board the Japanese Meteorological Association (JMA) Multifunction Transport SATellite (MTSAT-2), and NASA’s Geostationary Operational Environmental Satellites (GOES).

The merged LST product is generated in two steps: 1) calibration between each LEO and each GEO that consists in the removal of systematic differences (associated to sensor type and LST algorithms, including calibration, atmospheric and surface emissivity corrections, amongst others) represented by linear regressions; 2) angular correction that consists in bringing all LST data to reference (nadir) view.

We present a methodology that allows the effective estimation of angular effects on LST. It relies on a kernel model of the surface thermal emission, which describes the angular dependence of LST as function of viewing and illumination geometry. The model was adjusted to MODIS and SEVIRI/MSG LST estimates and validated against LST retrievals from those sensors obtained for other years (not used in the calibration). It is shown that both models lead to a reduction of LST differences between the two sensors, indicating that these may be used to effectively estimate/correct angular dependence in LST. A global set of kernel model parameters is finally obtained by adjusting the model to either a GEO and a LEO or the two LEOs (poles).