Abstract
This paper presents a single-channel atmospheric correction method for remotely sensed infrared (wavelength of 3–15 μm) images with various observation angles. The method is based on basic radiative transfer equations with a simple absorption-focused regression model to calculate the optical thickness of each atmospheric layer. By employing a simple regression model and re-organization of atmospheric profiles by considering viewing geometry, the proposed method conducts atmospheric correction at every pixel of a numerical weather prediction model in a single step calculation. The Visible Infrared Imaging Radiometer Suite (VIIRS) imaging channel (375 m) I4 (3.55~3.93 μm) and I5 (10.50~12.40 μm) bands were used as mid-wavelength and thermal infrared images to demonstrate the effectiveness of the proposed single-channel atmospheric correction method. The estimated sea surface temperatures (SSTs) obtained by the proposed method with high resolution numerical weather prediction models were compared with sea-truth temperature data from ocean buoys, multichannel-based SST products from VIIRS/MODIS, and results from MODerate resolution atmospheric TRANsmission 5 (MODTRAN 5), for validation. High resolution (1.5 km and 12 km) numerical weather prediction (NWP) models distributed by the Korea Meteorological Administration (KMA) were employed as input atmospheric data. Nighttime SST estimations with the I4 band showed a root mean squared error (RMSE) of 0.95 °C, similar to that of the VIIRS product (RMSE: 0.92 °C) and lower than that of the MODIS product (RMSE: 1.74 °C), while estimations with the I5 band showed an RMSE of 1.81 °C. RMSEs from MODTRAN simulations were similar (within 0.2 °C) to those of the proposed method (I4: 0.81 °C, I5: 1.67 °C). These results demonstrated the competitive performance of a regression-based method using high-resolution numerical weather prediction (NWP) models for atmospheric correction of single-channel infrared imaging sensors.
Highlights
Earth surface applications of mid-wavelength (3–5 μm) and thermal infrared (8–14 μm) (MWIR and TIR, respectively) remote sensing, including surface temperature retrieval, essentially require atmospheric correction
We provide an idea of a single-channel atmospheric correction method, with a combination of high-resolution (1.5 km and 12 km) numerical weather prediction (NWP) models from the Korea Meteorological Administration (KMA)
The parameters were calculated for Visible Infrared Imaging Radiometer Suite (VIIRS) imaging channels I4 and I5 by the single-channel method proposed in this study
Summary
Earth surface applications of mid-wavelength (3–5 μm) and thermal infrared (8–14 μm) (MWIR and TIR, respectively) remote sensing, including surface temperature retrieval, essentially require atmospheric correction. The split-window (SW) algorithm is widely used for TIR sensors, as it is easy to apply and can be used without extensive computation of radiative transfer models (RTMs) or explicit information of atmospheric conditions [1,6,7,8,9]. These multi-channel methods have limited application in other types of single-channel-based IR sensors, such as LANDSAT-7 [10]. Depending on the radiative transfer models, the number of calculations should be increased, or even require calculations for every pixels of NWP data, in order to minimize the effects of different viewing angles, as well as atmospheric spatial inhomogeneity and to obtain a uniform atmospheric correction quality within a scene
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
