Derived LST: 10-day composites (DLST)

NRT Product Available since Nov 2015


Product Documentation

This operational product is documented in the Algorithm Theoretical Basis Document (ATBD), Product User Manual document (PUM) and the Product Output Format document (POF) The validation results for this product are available in the (VR) document.


Data Policy

The use of LSA SAF products in publications is kindly requested to be duly acknowledged:
DLST and DLST-TSP were provided by the EUMETSAT Satellite Application Facility on Land Surface Analysis (LSA SAF; Trigo et al., 2011)

Trigo, I. F., C. C. DaCamara, P. Viterbo, J.-L. Roujean, F. Olesen, C. Barroso, F. Camacho-de Coca, D. Carrer, S. C. Freitas, J. García-Haro, B. Geiger, F. Gellens-Meulenberghs, N. Ghilain, J. Meliá, L. Pessanha, N. Siljamo, and A. Arboleda, 2011: The Satellite Application Facility on Land Surface Analysis. Int. J. Remote Sens., 32, 2725-2744, doi: 10.1080/01431161003743199



Derived LST (DLST; LSA-003) provide a synthesis of the Meteosat Land Surface Temperature (MLST; LSA-001) products within a 10 day interval. The DLST consist of two sub-products: LSA-003A, the maximum and median MLST composites over the 10-day interval, and LSA-003B, a set of so-called Thermal Surface Parameters (TSP; LSA-003B), which are obtained by fitting a diurnal temperature cycle model (Göttsche and Olesen, 2009) to the corresponding MLST composites. Like the well-known Normalised Difference Vegetation Index (NDVI), MLST composites are simple subsets of the input quantity, i.e. the LSA-003A products. Temporal composites usually have considerably fewer missing data and are spatially more continuous than the input data. The LSA-003B sub-product contains sets of physically meaningful model parameters, i.e. the TSP, which summarize the thermal behaviour of the land surface for the 10 day interval and require about 10 times less storage space than the composites. TSP have physical meaning and direct relevance to (land surface) applications, e.g. minimum temperature (i.e. around sunrise), temperature amplitude and time of maximum temperature (Göttsche and Olesen, 2009; Duan et al. 2012; Holmes et al., 2013; Göttsche, 2016).


Product Description

The Derived LST (DLST; LSA-003) product is based entirely on Meteosat Land Surface Temperature (MLST; LSA-001) as input, which LSA SAF estimates from clear-sky TOA brightness temperatures of SEVIRIs split-window channels centred on 10.8 and 12.0 µm (Trigo et al., 2009). The LSA-003 products are generated at SEVIRI full spatial resolution (3 km sampling distance at nadir) within the area covered by the MSG disk. Each LSA-003A sub-product contains fields with the LST composite values for the respective SEVIRI slot, the number of valid input MLST and for uncertainty and Quality Control (QC) information. LSA-003B sub-products contain fields for each of the 7 TSP, the DTC modelling errors, and for Quality Control (QC) information on the fit and the input LSA-003A. There are generally 96 LSA-003A products (one for each SEVIRI slot) and one LSA-003B product for each compositing interval and type of composite (maximum and median). All LSA-003 products are disseminated in HDF5 format and the relevant information concerning the data fields is included in the HDF5 attributes (Göttsche, 2016).


Algorithm Description

Composited diurnal MLST cycles (LSA-003A) and Thermal Surface Parameters (TSP; LSA-003B) are both part of the DLST (LSA-003) product. Together the DLST sub-products provide a 10-day synthesis of the LSA SAF MLST product (LSA-001) and spatially smooth fields of maximum and minimum temperature are obtained. This is achieved in two steps consisting of the estimation of:
 (i) Maximum and Median LST (LSA-003A) within a compositing interval of 10 days, per time-slot and pixel, leading to a maximum/median value every 15 minutes (Trigo et al., 2009)
 (ii) Thermal Surface Parameters ‘TSP’ (LSA-003B) that summarize maximum and median diurnal temperature cycles given by the LSA-003A products (Göttsche, 2016).

Maximum composites, i.e. maximum LST per observation time-slot within a 10-day period, are well established and highly popular due to their simplicity. However, median composites (median per observation time slot) are computationally only slightly more demanding with the biggest difference being data handling and storage. An advantage of median LST is that they represent the typical situation within the composition interval. In contrast, maximum composites collect the highest observed LST and are, therefore, not necessarily representative. When composites of too few input data are formed (e.g. from only 2 valid measurements), they tend to be cloud contaminated: therefore, the number of measurements used to create a composite is kept for later interpretation. While minimum LST compositing is easily implemented, the resulting composites tend to be contaminated by undetected clouds, since the corresponding pixels are usually the coldest. However, modelling median LST composites with the DTC model described in (Göttsche and Olesen, 2009; Göttsche, 2016) yields stable estimates of representative minimum LST within the compositing period.

Thermal Surface Parameters (LSA-003B) for a 10-day period are obtained by fitting a model of the diurnal temperature cycle (Göttsche and Olesen, 2009) to the corresponding MLST composites (LSA-003A). The LSA-003B algorithm performs an effective compression of the 15 minute information (up 96 values per pixel, as contained in the sequence of LSA-003A products) through a small number of 7 (6 independent) physically meaningful parameters. Since the LSA-003A products are subsets of MLST, the LSA-003A algorithm is described together with the LSA-001 algorithm in the updated MLST ATBD, while the LSA-003B algorithm is described in detail in the DLST ATBD (Göttsche, 2016). An overview of the processing chains for the two DLST algorithms is provided in the figure below.


Data Characteristics



Product Uncertainties




Göttsche, Frank-M. , and Olesen, Folke-S. (2009), Modelling the effect of optical thickness on diurnal cycles of land surface temperature. Remote Sensing of Environment, Vol. 113(11), pp. 2306-2316. doi: 10.1016/j.rse.2009.06.006

Duan, S.-B., Li, Z.-L., Wang, N., Wu, H., Tang, B.-H. (2012). Evaluation of six land-surface diurnal temperature cycle models using clear-sky in situ and satellite data. Remote Sensing of Environment, Vol. 124, pp. 15-25. doi: 10.5194/hess-17-3695-2013

Holmes, T.R.H., Crow, W.T., and Hain, C. (2013), Spatial patterns in timing of the diurnal temperature cycle. Hydrology and Earth System Sciences, Vol. 17, pp. 3695–3706. doi: 10.1016/j.rse.2012.04.016

Hong, F., Zhan, W., Göttsche, F.-M., Liu, Z., Zhou, J., Huang, F., Lai, J. and Li, M. (2018), Comprehensive assessment of four-parameter diurnal land surface temperature cycle models under clear-sky. ISPRS Journal of Photogrammetry and Remote Sensing, vol. 142, pp. 190-204.

Example of Product