Enhanced Temporal Resolution of Observations of Thermodynamic Variables, Clouds and Precipitation: Recent Scientific Results from the Temporal Experiment for Storms and Tropical Systems (TEMPEST) Missions

Passive microwave sensors provide some of the most valuable observed quantities for initialization and validation of regional and global weather forecasting models. Although substantial progress has been made in global climate modeling, extreme weather events are still misrepresented in parameterizations, largely due to a lack of sufficient observations. At the same time, passive microwave sensors perform Earth observation exclusively from low Earth orbit (LEO), so their limited number leads to substantial temporal sampling gaps from the tropics to the mid-latitudes, where cyclones and other precipitating storms cause the greatest damage to life and property.

In the context of low-cost microwave sensors, the Temporal Experiment for Storms and Tropical Systems – Demonstration (TEMPEST-D) mission demonstrated the first global (up to 58 degrees latitude) observations from a multi-frequency microwave radiometer on a CubeSat, operating for nearly three years from 2018 to 2021. This NASA Earth Venture Technology mission produced a high-quality scientific data, instrument calibration, radiometer stability, and long mission duration. TEMPEST-D brightness temperatures were validated using scientific and operational microwave sensors, including GPM/GMI and four MHS sensors, operating at similar frequencies to TEMPEST-D channels at 87, 164, 174, 178 and 181 GHz. TEMPEST-D performance was shown to be comparable to or better than much larger operational sensors, in calibration accuracy, precision, stability and instrument noise, during its nearly 3-year mission.

A nearly identical TEMPEST instrument produced by JPL alongside TEMPEST-D for risk reduction was made available to the U.S. Space Force under a mandate to demonstrate low-cost space technologies to improve global weather forecasting. The TEMPEST flight-spare instrument was then integrated with the Compact Ocean Wind Vector Radiometer (COWVR) that had been produced by NASA/JPL for the U.S. Air Force. COWVR and TEMPEST were launched together on December 21, 2021, as part of the Space Test Program – Houston 8 (STP-H8) mission and deployed on the Japanese Experiment Module on the ISS for at least 3 years of operations. COWVR and TEMPEST have performed coordinated, complementary observations of Earth’s oceans and atmosphere from the ISS since January 7, 2022. Retrievals of water vapor profiles, clouds, and precipitation from COWVR/TEMPEST-H8 are being performed in collaboration between JPL and Colorado State University.

Atmospheric inversion techniques have been developed to retrieve water vapor altitude profiles, as well as single-layer cloud liquid water and cloud ice water, from TEMPEST brightness temperatures, using ECMWF Reanalysis v5 (ERA5) data as an initial guess. These retrievals are enhanced through the inclusion of geostationary infrared data from GOES-16 ABI channels, increasing the number of levels and reducing the error of water vapor retrieval, particularly in the upper troposphere.

Previous studies have validated the accuracy and precision of TEMPEST-D brightness temperatures using only clear-sky oceanic observations. Cross-validation of TEMPEST-D passive microwave and RainCube active microwave CubeSat sensors has demonstrated high qualitative and quantitative correlation over storm systems on a global basis. Recent studies have extended the quantitative validation of both TEMPEST-D and TEMPEST-H8 to include observations over tropical cyclones, hurricanes, and typhoons using both passive microwave GPM-GMI brightness temperatures and active microwave GPM-DPR vertical cumulative reflectivity. These active/passive microwave observations provide the basis for development of surface rain rate estimates and retrieval of the vertical structure of precipitation from combined TEMPEST and DPR observations.

Previous machine learning models have demonstrated rain rate estimation from TEMPEST-D brightness temperatures over the continental U.S., with very good quantitative comparison with ground-based weather radar rainfall estimates. This work has been extended to a global basis using machine learning models based on a random forest classifier and regression models for multiple precipitation rates. These global models have been developed to estimate surface rain rates based on TEMPEST-D and TEMPEST-H8 brightness temperature observations. Independent validation has demonstrated consistently high performance across all machine learning models, as compared with IMERG rain rate products, showing high values of correlation.

The accuracy, precision, and long-term stability of TEMPEST-D microwave radiometer operation on a 6U CubeSat throughout the three-year mission demonstrated the potential for substantial enhancement of temporal observations from the tropics to the mid-latitudes. Future constellations of passive microwave sensors on small satellites in LEO are expected to provide greatly enhanced temporal observations for atmospheric sounding and remote sensing of clouds and precipitation.