Abstract
Collection of spectroradiometric measurements with associated biophysical variables is an essential part of the development and validation of optical remote sensing vegetation products. However, their quality can only be assessed in the subsequent analysis, and often there is a need for collecting extra data, e.g., to fill in gaps. To generate empirical-like surface reflectance data of vegetated surfaces, we propose to exploit emulation, i.e., reconstruction of spectral measurements through statistical learning. We evaluated emulation against classical interpolation methods using an empirical field dataset with associated hyperspectral spaceborne CHRIS and airborne HyMap reflectance spectra, to produce synthetic CHRIS and HyMap reflectance spectra for any combination of input biophysical variables. Results indicate that: (1) emulation produces surface reflectance data more accurately than interpolation when validating against a separate part of the field dataset; and (2) emulation produces the spectra multiple times (tens to hundreds) faster than interpolation. This technique opens various data processing opportunities, e.g., emulators not only allow rapidly producing large synthetic spectral datasets, but they can also speed up computationally intensive processing routines such as synthetic scene generation. To demonstrate this, emulators were run to simulate hyperspectral imagery based on input maps of a few biophysical variables coming from CHRIS, HyMap and Sentinel-2 (S2). The emulators produced spaceborne CHRIS-like and airborne HyMap-like surface reflectance imagery in the order of seconds, thereby approximating the spectra of vegetated surfaces sufficiently similar to the reference images. Similarly, it took a few minutes to produce a hyperspectral data cube with a spatial texture of S2 and a spectral resolution of HyMap.
Highlights
The acquisition of spectroradiometric data and associated biophysical variables are an essential part of the development and validation of imaging spectroscopy vegetation products [1,2,3]
Based on the calculated statistics and recorded run-time in Table 1, the following trends can be observed: (1) kernel ridge regression (KRR) and Gaussian processes regression (GPR) emulation approximated the surface reflectance data considerably more accurately than the two interpolation methods, who perform with linear interpolation performing slightly better than nearest interpolation
The performance of neural networks (NNs) to reconstruct surface reflectance data tended to be more unstable; for the CHRIS dataset, the NN emulator performed to the other emulators, while, for the HyMap dataset, the NN emulator performed on the same order as the interpolation methods
Summary
The acquisition of spectroradiometric data and associated biophysical variables are an essential part of the development and validation of imaging spectroscopy vegetation products [1,2,3]. The obvious option is returning to the field to collect new measurements This is not always feasible: repeating a campaign is costly and time-consuming, it may be that too much time has passed and the vegetation conditions have changed. It implies that an alternative way to collect extra data has to be considered In this respect, the second option involves generating new data based on the already existing empirical data, e.g., by means of interpolation or extrapolation techniques. The second option involves generating new data based on the already existing empirical data, e.g., by means of interpolation or extrapolation techniques This approach will never replace the collection of original field data, yet it can provide an adequate approximation and is without costs
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
