Derivation and Evaluation of Land Surface Temperature from the Geostationary Operational Environmental Satellite Series

Abstract

The Geostationary Operational Environmental Satellites (GOES) have been continuously monitoring the earth surface since 1970, providing valuable and intensive data from a very broad range of wavelengths, day and night. The National Oceanic and Atmospheric Administration's (NOAA's) National Environmental Satellite, Data, and Information Service (NESDIS) is currently operating GOES-15 and GOES-13. The design of the GOES series is now heading to the 4th generation. GOES-R, as a representative of the new generation of the GOES series, is scheduled to be launched in 2015 with higher spatial and temporal resolution images and full-time soundings. These frequent observations provided by GOES Image make them attractive for deriving information on the diurnal land surface temperature (LST) cycle and diurnal temperature range (DTR). These parameters are of great value for research on the Earth’s diurnal variability and climate change. Accurate derivation of satellite-based LSTs from thermal infrared data has long been an interesting and challenging research area. To better support the research on climate change, the generation of consistent GOES LST products for both GOES-East and GOES-West from operational dataset as well as historical archive is in great demand. The derivation of GOES LST products and the evaluation of proposed retrieval methods are two major objectives of this study. Literature relevant to satellite-based LST retrieval techniques was reviewed. Specifically, the evolution of two LST algorithm families and LST retrieval methods for geostationary satellites were summarized in this dissertation. Literature relevant to the evaluation of satellite-based LSTs was also reviewed. All the existing methods are a valuable reference to develop the GOES LST product. The primary objective of this dissertation is the development of models for deriving consistent GOES LSTs with high spatial and high temporal coverage. Proper LST retrieval algorithms were studied according to the characteristics of the imager onboard the GOES series. For the GOES 8-11 and GOES R series with split window (SW) channels, a new temperature and emissivity separation (TES) approach was proposed for deriving LST and LSE simultaneously by using multiple-temporal satellite observations. Two split-window regression formulas were selected for this approach, and two satellite observations over the same geo-location within a certain time interval were utilized. This method is particularly applicable to geostationary satellite missions from which qualified multiple-temporal observations are available. For the GOES M(12)-Q series without SW channels, the dual-window LST algorithm was adopted to derive LST. Instead of using the conventional training method to generate coefficients for the LST regression algorithms, a machine training technique was introduced to automatically select the criteria and the boundary of the sub-ranges for generating algorithm coefficients under different conditions. A software package was developed to produce a brand new GOES LST product from both operational GOES measurements and historical archive. The system layers of the software and related system input and output were illustrated in this work. Comprehensive evaluation of GOES LST products was conducted by validating products against multiple ground-based LST observations, LST products from fineresolution satellites (e.g. MODIS) and GSIP LST products. The key issues relevant to the cloud diffraction effect were studied as well. GOES measurements as well as ancillary data, including satellite and solar geometry, water vapor, cloud mask, land emissivity etc., were collected to generate GOES LST products. In addition, multiple in situ temperature measurements were collected to test the performance of the proposed GOES LST retrieval algorithms. The ground-based dataset included direct surface temperature measurements from the Atmospheric Radiation Measurement program (ARM), and indirect measurements (surface long-wave radiation observations) from the SURFace RADiation Budget (SURFRAD) Network. A simulated dataset was created to analyse the sensitivity of the proposed retrieval algorithms. In addition, the MODIS LST and GSIP LST products were adopted to cross-evaluate the accuracy of the GOES LST products. Evaluation results demonstrate that the proposed GOES LST system is capable of deriving consistent land surface temperatures with good retrieval precision. Consistent GOES LST products with high spatial/temporal coverage and reliable accuracy will better support detections and observations of meteorological over land surfaces.