Abstract
This PhD-thesis investigates the directional effects in land surface temperature (LST) estimates from the SEVIRI sensor onboard the Meteosat Second Generation (MSG) satellites. The directional effects are caused by the land surface structure (i.e. tree size and shape) interacting with the changing sun-target-sensor geometry. The directional effects occur because the different surface components, e.g. tree canopies and bare soil surfaces, will in many cases have significantly different temperatures. Depending on the viewing angle, different fractions of each of the components will be viewed by the sensor. This is further complicated by temperature differences between the sunlit and shaded parts of each of the components, controlled by the exposure of the components to direct sunlight. As the SEVIRI sensor is onboard a geostationary platform, the viewing geometry is fixed (for each pixel), while the illumination geometry changes both over the course of the day and with the seasons. In the present study, the directional effects are assessed at different scales using a modeling approach. The model applied, the Modified Geometry Projection (MGP) model, represents the surface as a composite of four components; shaded and sunlit canopy and background, respectively. Given data on vegetation structure and density, the model estimates the fractions of the four components as well as the directional composite temperature in the view of a sensor, given the illumination and viewing geometry.
The modeling results show that the magnitude of the directional effects mainly depends on the tree cover, with moderate tree covers (20-40 %) causing the largest directional effects but with significant effects also at much sparser tree cover. The magnitude is also highly dependent on the temperature difference between the surface components, which is often largest in semi-arid areas that have relatively cool tree canopies and relatively hot soil/grass background. The largest amplitude in the directional effects occurs at “hot spot" geometry, which for geostationary sensors is at the equinoxes. Furthermore, the directional effects have varying magnitude and sign on both diurnal and seasonal scales, which will have implications if using LST products in downstream applications like hydrological or soil vegetation atmosphere transfer (SVAT) models. The directional effects will cause uncertainties in LST estimates that are different in terms of timing than the uncertainties in data from polar orbiting sensors, which will cause discrepancies between measurements from the two types of sensors. An assessment of the performance of current LST algorithms from MSG SEVIRI for semi-arid West Africa was carried out, using data from two field sites in Senegal and Mali. The agreement between the satellite and ground data for the rainy season was generally discouraging with biases exceeding 5 K, while there were indications that performance is much better during the dry season. The large discrepancies are thought to be caused by insufficient correction for the influence of the very moist atmosphere. For the period studied, the uncertainties found in the current LST products are likely to make it infeasible to identify the directional effects in the satellite data, as the uncertainties will mask the directional effects.
The modeling results show that the magnitude of the directional effects mainly depends on the tree cover, with moderate tree covers (20-40 %) causing the largest directional effects but with significant effects also at much sparser tree cover. The magnitude is also highly dependent on the temperature difference between the surface components, which is often largest in semi-arid areas that have relatively cool tree canopies and relatively hot soil/grass background. The largest amplitude in the directional effects occurs at “hot spot" geometry, which for geostationary sensors is at the equinoxes. Furthermore, the directional effects have varying magnitude and sign on both diurnal and seasonal scales, which will have implications if using LST products in downstream applications like hydrological or soil vegetation atmosphere transfer (SVAT) models. The directional effects will cause uncertainties in LST estimates that are different in terms of timing than the uncertainties in data from polar orbiting sensors, which will cause discrepancies between measurements from the two types of sensors. An assessment of the performance of current LST algorithms from MSG SEVIRI for semi-arid West Africa was carried out, using data from two field sites in Senegal and Mali. The agreement between the satellite and ground data for the rainy season was generally discouraging with biases exceeding 5 K, while there were indications that performance is much better during the dry season. The large discrepancies are thought to be caused by insufficient correction for the influence of the very moist atmosphere. For the period studied, the uncertainties found in the current LST products are likely to make it infeasible to identify the directional effects in the satellite data, as the uncertainties will mask the directional effects.
| Originalsprog | Engelsk |
|---|
| Forlag | Museum Tusculanum |
|---|---|
| Antal sider | 101 |
| Status | Udgivet - 3 dec. 2010 |
FN’s Verdensmål
Dette resultat bidrager til følgende verdensmål
