Publications

You, Dongqin; Wen, Jianguang; Liu, Qiang; Liu, Qinhuo; Tang, Yong (2014). The Angular and Spectral Kernel-Driven Model: Assessment and Application. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 7(4), 1331-1345.

Abstract
Land surface albedo is a critical parameter in the earth's energy budget. Multiple-sensor data contain more information than single-sensor data, enabling us to retrieve albedo more accurately. The Angular and Spectral Kernel driven model (ASK model), which introduces component spectra into a kernel-driven model, provides a way to combine multiple-sensor data to retrieve BRDF/albedo. The construction of the ASK model is detailed in Liu's paper. As a follow-up, this paper provides an extensive assessment of the ASK model and its application using multi-sensory data. The assessment is described in both angular and spectral dimensions using simulated datasets from ProSail, 5-Scale, and RGM. With the ability to combine information from the spectral and angular domains, the inversion of the ASK model requires fewer angular observations than the traditional model. Four angles are sufficient when combining seven MODIS bands. In the spectral dimension, the model performance reveals high numerical correlations among bands: the red and NIR bands are generally required to make a good spectra fitting, and adding an additional SWIR band can improve the performance. The synergistic retrieval of albedo combining FY3/VIRR, AVHRR, and MODIS shows a satisfactory agreement with in situ measurements, where the RMSE is 0.013 in the 4-day composited temporal resolution retrieval. The results show that the ASK model is promising for BRDF/albedo inversion using multi-sensor data, although it shows some dependence on the accuracy of the component spectra.

DOI:
10.1109/JSTARS.2013.2271502

ISSN:
1939-1404; 2151-1535