## Abstract

The radiometric conversion of the raw satellite sensor data is performed to calculate the top-of-the-atmosphere radiance and its associated uncertainty is estimated at first step. The next step of the satellite data processing is the atmospheric correction. This study focuses on the TPU created during the atmospheric correction procedure. The atmospheric correction consists of the angular geometry of the sun-sensor orientation and the atmospheric properties. The optically dominant atmospheric components are aerosol, ozone, water vapor. The atmospheric correction introduces additional uncertainties propagated through the atmospheric correction equation. In this study we propose the methodology to estimate the statistical TPU created by the atmospheric correction. The uncertainty of each atmospheric component is estimated first as a standard deviation. Next, the Jacobian matrix as a partial derivative of all surface reflectance bands with respect to several atmospheric parameters is formed and the TPU matrix is calculated using Jacobian matrix and the uncertainty covariance matrxi. The TPU matrix represents the uncertainty of the surface reflectance for each band of each pixel. Any downstream application product will be able to assess its uncertainty based on the associated TPU of the surface reflectance. For example, if a user computes NDVI, the associated uncertainty of the NDVI can be calculated based on the TPU of the surface reflectance. Having the NDVI uncertainty image will allow a user to see the area with larger or smaller NDVI uncertainty and help a decision making in an educated use of the NDVI product.

TPU from Atmospheric Correction of Landsat 8 OLI Imagery

The radiometric conversion of the raw satellite sensor data is performed to calculate the top-of-the-atmosphere radiance and its associated uncertainty is estimated at first step. The next step of the satellite data processing is the atmospheric correction. This study focuses on the TPU created during the atmospheric correction procedure. The atmospheric correction consists of the angular geometry of the sun-sensor orientation and the atmospheric properties. The optically dominant atmospheric components are aerosol, ozone, water vapor. The atmospheric correction introduces additional uncertainties propagated through the atmospheric correction equation. In this study we propose the methodology to estimate the statistical TPU created by the atmospheric correction. The uncertainty of each atmospheric component is estimated first as a standard deviation. Next, the Jacobian matrix as a partial derivative of all surface reflectance bands with respect to several atmospheric parameters is formed and the TPU matrix is calculated using Jacobian matrix and the uncertainty covariance matrxi. The TPU matrix represents the uncertainty of the surface reflectance for each band of each pixel. Any downstream application product will be able to assess its uncertainty based on the associated TPU of the surface reflectance. For example, if a user computes NDVI, the associated uncertainty of the NDVI can be calculated based on the TPU of the surface reflectance. Having the NDVI uncertainty image will allow a user to see the area with larger or smaller NDVI uncertainty and help a decision making in an educated use of the NDVI product.