GK2A Infrared Calibration

      The Advanced Meteorological Imager (AMI) of the GEO-KOMPSAT-2A (GK2A) has 10 infrared channels and performs radiometric calibration during the level 1A data processing stage. This calibration is achieved by measuring the radiances of warm and cold targets, blackbody onboard GK2A, and space-look, respectively. In addition to onboard calibration, the NMSC examines inter-calibration as part of the Global Space-based Inter-Calibration System (GSICS) activities. This is useful for both checking the quality of operational measurements and monitoring sensor degradation over time. The GSICS inter-calibration system compares two observation values measured by an instrument. The goal is to calibrate GK2A AMI with reference instruments, which are known to be relatively accurate.

Infrared Inter-Calibration of Imagers on GK2A Using High-Spectral- Resolution Sounders as References

      The NMSC performs inter-calibration between the AMI on GK2A and high-spectral-resolution sounders on LEO satellites. Data from the Infrared Atmospheric Sounding Interferometers (IASI) on Metop–B and -C satellites, as well as Cross-track Infrared Sounder (CrIS) on SNPP and NOAA-20 satellite, are used for this inter-calibration. The infrared inter-calibration results are generated once a day.

Algorithm (ATBD)

      The inter-calibration system for GK2A’s infrared channels is based on the GSICS Coordinate Center (GCC) Algorithm Theoretical Basis Document (ATBD). To conduct the calibration system, a collocation dataset is first required to be created for the measured pixels of GK2A and LEOs. The field of view (FOV) of hyper-spectral sounders is approximately 12 km in diameter at nadir, while the FOV of AMI infrared channels is 2 km. The radiance values from the sounder are compared with the average values of AHI radiances over a 7×7 pixel FOV box corresponding to the sounder FOV. Several channel measurements from the hyper-spectral sounder are converted into simulated radiances based on the spectral response function (SRF) of GK2A. This is called the constraint method, which generates a super channel consisting of numerous channel combinations to imitate a broadband channel. More details on the methods of collocation and spectral simulation are described below.

Collocation Method

      The collocation algorithm used in inter-calibration is determined by the GSICS Research Working Group. GEO imagers and hyper sounder data that meet the collocation criteria are selected to perform the following check.

Difference check for observation time

     | Time LEO − Time GEO | < 300s ( =5 min)

Difference check for satellite zenith angle

     | cos( SZA LEO ) / cos( SZA GEO ) − 1 | < 𝜀1

Uniformity check for environment

      To reduce the differences between the observation conditions of two satellites, such as especially time, navigation, optical path, and cloud advection, inter-calibration only selects measurements over uniform scenes. A uniformity test is conducted for environmental pixels over a 21x21 area (called as the ENV box) of GEO.

     STDV(GEO radiances, ENV BOX) < 𝜀2

Normality check

      To check the normality of the GEO radiance data within the FOV BOX, the following condition is used, given that the length of FOV BOX is 7.

     | MEAN(GEO radiances, FOV BOX) − MEAN(GEO radiances, ENV BOX) | × 7 / STDV(ENV BOX) < 𝜀3(Gaussian)

      The tables below represent the criteria used in inter-calibration between GEO imagers and AIRS/IASI/CrIS. The values differ according to channels and weather conditions. If the brightness temperature of IR105 exceeds 275 K, the scene condition is categorized as clear; otherwise it is categorized as cloudy.

data result
AMI channel condition 𝜀1 𝜀2 𝜀3
SW038 Clear 0.01 0.0238 2
Cloudy 0.03 0.0476
WV063 All 0.01 0.371 1
WV069 All 0.01 0.561 1
WV073 All 0.01 0.661 1
IR087 Clear 0.01 1.18 2
Cloudy 0.03 2.36
IR096 Clear 0.01 1.46 2
Cloudy 0.03 2.92
IR105 Clear 0.01 1.62 2
Cloudy 0.03 3.24
IR112 Clear 0.01 1.77 2
Cloudy 0.03 3.54
IR123 Clear 0.01 1.91 2
Cloudy 0.03 3.82
IR133 Clear 0.01 2.03 2
Cloudy 0.03 4.06

Spectral response compensation method

      CrIS employed a gap-filling method that predicts CrIS gap channels based on principal component-based regression. For more information, please refer to Hui Xu (2018).