Land surface emissivity is a key parameter in estimating the land surface radiation budget. The validation of the moderate-resolution imaging spectroradiometer (MODIS) land surface emissivity with field measurements is rarely performed. In this study, a field measurement was performed over the central part of the Taklimakan Desert for the validation of the MODIS land surface emissivity products (MOD11B1) Version 4 (V4.1) and Version 5 (V5). The homogeneity of two validation sites was verified using the advanced spaceborne thermal emission and reflection radiometer (ASTER) land surface temperature and emissivity acquired closely before and after the overpass of MODIS. MOD11B1 V4.1 and V5 emissivity data for bands 29, 31, and 32 were compared to the emissivity calculated from the field measured emissivity spectra convolved with the filter function of the MODIS bands 29 (8.52 μm), 31 (11.03 μm), and 32 (12.04 μm). The comparison results indicate that the V4.1 emissivity data agree well with the field measurements, with mean absolute differences of 0.017 and 0.007 for site 1 and site 2, respectively, and the mean absolute differences of the V5 emissivity data were 0.034 and 0.033 for site 1 and site 2, respectively. The data version used must be considered when MOD11B1 is used in real applications, especially for time series analysis.
Plant foliage density expressed as leaf area index (LAI) is an important parameter that is widely used in many ecological, meteorological and agronomic models. LAI retrieval using optical remote sensing usually requires the collection of surface calibration values or the use of image information to invert radiative transfer models. A comparison of LAI retrieval methods was conducted that included both empirical methods requiring ground based LAI calibration measurements and image based methods using remotely sensed data and literature reported parameter values. The empirical approaches included ordinary least squares regression with the Normalized Difference Vegetation Index (NDVI) and the Gitelson green index (GI) spectral vegetation indices (SVI) and a geostatistical approach that uses ground based LAI measurements and image derived kriging parameters to predict LAI. The image based procedures included the scaled SVI approach, which uses NDVI to estimate fraction of vegetation cover, and a hybrid approach that uses a neural network and a radiative transfer model to retrieve LAI. Comparable results were obtained with the empirical SVI methods and the scaled SVI method. The geostatistical approach produced LAI patterns similar to interpolated ground-based LAI measurements. The results demonstrated that although reasonable LAI estimates are possible using optical remote sensing data without in situ calibration measurements, refinements to the analytical steps of the various approaches are warranted.
An off-nadir canopy reflectance model, the Liang and Strahler algorithm for the coupled atmosphere and canopy (CAC) model, was used to simulate multi-angle reflectances based on various combinations of canopy biophysical parameters. Biophysical parameters such as leaf angle distribution and leaf area index were input to the CAC model along with reflectances of leaf, soil, and aerosol optical depth. The CAC model, however, can only be inverted through numerical iterations and it is extremely difficult to use for retrieval of those biophysical parameters with ordinary inversion methods. In order to retrieve those biophysical parameters, we employed an error back-propagation feed forward neural network program. We constructed a number of neural network models based on the simulated results from the CAC model. Ideally, through network training we would like to have a neural network model that uses the multi-angle reflectances as its inputs and output simultaneously all the biophysical parameters, the component reflectances of leaf and background soil, and the aerosol optical depth of the atmosphere. We have not yet reached this objective due to the requirement of an extremely large amount of calculation. In this paper, we report the results obtained from retrieving any individual parameter from multi-angle reflectances and results obtained from simultaneously retrieving some combinations of two parameters. We tested the use of a different number of multi-angle reflectances as input to the neural networks. This number varies in the range of 1 - 64. The test results show that a relative error between 1-5% or better is achievable for retrieving one parameter at a time or two parameters simultaneously
Conference Committee Involvement (4)
Land Surface and Cryosphere Remote Sensing IV
25 September 2018 | Honolulu, Hawaii, United States