Radiometers play a crucial role in providing accurate geo-physical information, relying heavily on precise calibration for both radiometric accuracy and spectral consistency. Radiometers consistently allocate time and hardware resources to calibration, resources that could otherwise be utilized for environmental sensing. In addition, calibration faces challenges such as frequency dependence and environmental influences, requiring to the need for innovative solutions. In this study, advancements in deep learning (DL) techniques are utilized, using NASA’s Soil Moisture Active Passive (SMAP) satellite data to create a DL-based radiometer calibrator. The use of 2-D spectral features as input in a convolutional neural network shows promising results with high correlation and low error. Notably, ancillary features like internal thermistor temperature prove accurate for estimating antenna temperature. This compensates for changes in receiver noise temperature and short-term gain fluctuations, even when there’s no reference load or noise diode power. The proposed calibration technique, emphasizing reduced reference information, holds significant potential for a higher number of antenna scene observations within a footprint.

Passive remote sensing through microwave radiometry has been utilized in Earth observation by estimating several geophysical parameters. Because of the low noise floor associated with the instrument (i.e., radiometer), the received geophysical emission is sampled in a protected band dedicated to remote sensing. This protected L-band occupying 1400-1427 MHz is also exciting and ideal for science because of lower attenuation from the atmosphere. This reason has also made this microwave region ideal for next-generation (xG) wireless communication. 5G cellular systems support two frequency ranges FR1 (0.45 GHz–6 GHz) and FR2 (24.45 GHz-52.6 GHz). Although operating bands are prohibited from conducting any up-link or down-link operations in the protected portion of the L-band, out-of-band (OOB) emissions can still have a significant impact on passive sensors because of the high sensitivity requirements related to science. This study will demonstrate a unique physical testbed that has the capability to observe in-band and OOB emissions in a protected anechoic chamber. Flexibility on transmitted waveforms and the potential to analyze raw measurements (IQ samples) of radiometers will help in designing onboard radio frequency interference (RFI) processing along with the coexistence of communication and passive sensing technologies.

The success of microwave radiometry depends on how accurately it can measure the natural emission of the Earth without the effects of unwanted signals. The consequence of unwanted signals in radiometers is known as radio frequency interference (RFI). The high intensity of these corrupted signals, along with wider bandwidth and longer duration, may jeopardize the overall success of a mission. These reasons resulted in a need for a robust RFI detection algorithm that will enable the mitigation of the contaminated portions of the measurements. Attributes related to RFI could be very dynamic, making it very difficult to detect with a particular algorithm. To address this issue, deep learning (DL) could be an attractive solution to detect RFI with the help of time-frequency analysis, i.e., spectrograms of the received measurement. This study aims to detect and localize RFI in a particular time-frequency bin of spectrograms with the help of DL to retrieve the non-contaminated portion of the measurements.

Passive microwave remote sensing plays an essential role in providing valuable information about the Earth’s surface, particularly for agriculture, water management, forestry, and other environmental fields. One of the key requirements for precision agricultural applications is the availability of field-scale high-resolution remote sensing data products. With the recent development of reliable unmanned aircraft systems (UAS), airborne deployment of remote sensing sensors has become more widespread to provide such products. With this in mind, we developed a UAS-based dual H-pol (horizontal) and V-pol (vertical) polarized radiometer operating in L-band (1400-1427 MHz). The custom dual-polarized antenna acquires surface emission response through a software-defined radio (SDR). This SDR-based system provides full control over the data acquisition parameters such as bandwidth, sampling frequency, and data size. Radio frequency interference (RFI) poses a significant challenge in radiometric measurements, requiring post-processing of the full-band radiometer data to identify and eliminate RFI-contaminated measurements, thus ensuring accurate Earth emission readings.. In this paper, we implemented near-real-time RFI detection onboard during the flight to accelerate the post-processing. The altitude and the speed of the UAS can be varied to achieve desired ground resolution for the measurement. This paper presents the full custom design and development of a lightweight SDR-based UAS-borne radiometer for precision agriculture. Additionally, we introduce the concept of an agile radiometer implemented from a small UAS that can serve as a testbed for both current and future spaceborne missions.

Soil Moisture Active Passive (SMAP) is a NASA's earth observing satellite which is used for global scale soil moisture measurement and differentiating frozen/thawed state. It is employed in 1400–1427 MHz protected band which uses L-Band radiometer for the quantification. But increasing number of wireless equipment such as air surveillance radar signals and 5G communication are making it harder to protect the radiometer microwave sensing in this secured spectrum. These technologies are responsible for the Radio Frequency Interference (RFI) in SMAP's passive observation. In this study, a novel deep learning architecture is developed that uses convolutional neural network (CNN) to predict RFI. Our model uses SMAP's level 1A raw antenna counts as well as level 1B quality flags to dynamically label these antenna raw measurements as RFI contaminated and RFI free footprints. This example study shows around 94% accuracy in detecting RFI and such result may recommend a lucrative technique in detecting RFI.

This paper explores the retrieval of snow water equivalent (SWE) through the use of machine learning techniques and active radar data collected over the 2020 SnowEx campaign. The retrieval makes use of active radar measurements provided by NASA's SWESARR instrument for direct sensing of snowpack sensitivity to SWE. The example results show that an RMSE of 1.93 cm can be obtained through a combined use of SAR data with sufficient ancillary data. Such results may indicate successful SWE estimation by means of pairing spaceborne SAR measurements with sufficient auxiliary information.