C028-0019
SHADOWS AS TOOL TO EXTEND ALTIMETRY RECORDS TO THE PAST: NEW REMOTE SENSING METHOD
SHADOWS AS TOOL TO EXTEND ALTIMETRY RECORDS TO THE PAST: NEW REMOTE SENSING METHOD
Thursday, 10 December 2020
Poster
Abstract:
Our ability to monitor altimetric changes in the cryosphere have increased dramatically in the last few decades. However, many of the processes driving these changes have time-scales larger than a few decades. Therefore, current research can significantly benefit from extending the altimetry time series into the past.
Here we present a novel methodology to recover altimetric information from non-stereo historical satellite and aerial images using the length of shadows projected by topographic features. Such shadows are often present on satellite and aerial imagery, especially on high-latitude terrain.
Provided a know solar elevation, we can calculate the height difference between a topographic feature and its shadow using simple geometrical principles. To date, such analysis has not been used to recover past thicknesses of ice masses due to the errors induced by unconstrained viewpoint geometries, atmospheric refraction, and the fuzziness of the shadow edge.
Here we test this conceptually straightforward idea using advanced tools to minimize such error sources. Among these tools we consider: precise Sun position calculations, refraction computation based on meteorological reanalysis data, precise viewpoint geometry, and shadow profile modelling.
To test the proposed methodology, we have obtained 108 shadow-derived elevations differences from ASTER imagery (ignoring stereo information). We then compared these measurements to geodetic GNSS field measurements (for mountains in the Central Andes of Argentina, and the Sentinel Range in Antarctica), or ICESat elevation data (for freeboard heights of Abbot ice shelf).
We found that shadow-derived height differences have a typical error of 3.5 m +- 0.24% (at 95% confidence level). Such accuracy levels suggest that this methodology could become a useful tool to estimate the surface height of ice bodies from aerial or satellite images much older than modern altimetric measurements.
Here we present a novel methodology to recover altimetric information from non-stereo historical satellite and aerial images using the length of shadows projected by topographic features. Such shadows are often present on satellite and aerial imagery, especially on high-latitude terrain.
Provided a know solar elevation, we can calculate the height difference between a topographic feature and its shadow using simple geometrical principles. To date, such analysis has not been used to recover past thicknesses of ice masses due to the errors induced by unconstrained viewpoint geometries, atmospheric refraction, and the fuzziness of the shadow edge.
Here we test this conceptually straightforward idea using advanced tools to minimize such error sources. Among these tools we consider: precise Sun position calculations, refraction computation based on meteorological reanalysis data, precise viewpoint geometry, and shadow profile modelling.
To test the proposed methodology, we have obtained 108 shadow-derived elevations differences from ASTER imagery (ignoring stereo information). We then compared these measurements to geodetic GNSS field measurements (for mountains in the Central Andes of Argentina, and the Sentinel Range in Antarctica), or ICESat elevation data (for freeboard heights of Abbot ice shelf).
We found that shadow-derived height differences have a typical error of 3.5 m +- 0.24% (at 95% confidence level). Such accuracy levels suggest that this methodology could become a useful tool to estimate the surface height of ice bodies from aerial or satellite images much older than modern altimetric measurements.