VICARIOUS CALIBRATION OF MICROWAVE RADIATION IMAGER ONBOARD FENG YUN SERIES SATELLITES IN YUNNAN FOREST

S. Wu¹, P. Yao¹, W. Xu¹

¹National Satellite Meteorological Center

As a key calibration technique, vicarious calibration, together with pre-launch laboratory calibration and in orbit real time calibration, forms the backbone of any microwave radiometer calibration strategy. Laboratory tests provide controlled measurements of the calibration target, receiver, and antenna, yet they cannot fully replicate the orbital environment. Calibration targets cannot reproduce the true in orbit radiative scene, cold sky calibration in vacuum does not capture receiver response below 10 K, and near field antenna measurements must be extrapolated to the far field with non-negligible error; moreover, antenna emissivity is difficult to measure on the ground. Real time in orbit calibration reveals the true system response on a second-by-second basis, but its accuracy is still limited by the laboratory-derived coefficients. To overcome these limitations, in-orbit vicarious calibration is required. Typical methods include rain-forest, ocean-surface, lunar, and cold-sky pitch-over calibrations. This paper reports on the rain-forest cross-calibration experiment carried out for China’s FengYun (FY) Microwave Radiation Imager (MWRI).

Between 2022 and 2025 the FengYun MWRI team conducted a series of joint ground–satellite campaigns in Yunnan Province, China. The objective was to exploit the tropical rain forest around Simao to develop and validate a rain-forest cross-calibration technique. Two kinds of observations were performed:

1. Ground-based radiometer measurements at 6.925, 10.65, 18.7 and 36.5 GHz in both V and H polarizations, mounted on a truck, have the ability to move around the experiment area. The radiometer was mainly located on the top of a small hill, platform and truck base together give an altitude difference of >10 m above the forest canopy, ensuring that the entire field of view was filled with rain forest.

2. In-canopy vertical temperature profiles measured at nine points. Up to five thermistors were deployed on each tower between 0 and 15 m height, yielding one temperature profile every 20 min. Nine vertical temperature-gradient sites are evenly distributed across a 25 km × 25 km area. Each site occupies a different underlying surface type: most are inside the forest, while a few are located in tea or coffee plantations. Within the forest every site is equipped with five temperature sensors at different heights; in the tea and coffee plots two sensors are installed at different levels.

Two dry-season radiometric data sets and more than one full year of vertical temperature profiles have been collected. Combined analysis shows that the forest emission exhibits pronounced diurnal variation, driven jointly by changes in ambient temperature and in the effective emissivity of the target. Because penetration depth of different frequencies in canopy is very different, we used the temperature-profile data together with the radiometer observations to quantify the true response of each channel within the forest medium. By coupling the temporal evolution of channel-dependent emissivity with the vertical temperature distribution, we constructed a preliminary RT model that uses physically based temperatures at multiple canopy heights as input and produces brightness temperatures for each channel as output.

Combined analysis of land-cover type and DEM indicates that the residual biases are dominated by (i) geolocation/co-registration errors and (ii) differences in local incidence angle arising from distinct azimuth angles of observation. Model refinement targeting these two error sources is on-going.