Race To the Bottom¶

2026 ARM Summer School Project¶

Three-layer Neural Net with Long Short Term Memory designed to extend the Microwave Radiometers range of precipitable water vapor (PWV) detection.

Project Scope¶

• Water vapor strongly absorbs at wavelength of ~900 nm. High PWV reduces the downwelling direct beam irradiance near that frequency. Therefore we are expecting that the shortwave irradiance decreases with higher water vapor concentrations.

• However shortwave (SW) irradiance is also affected by clouds and by Solar zenith angle

• A model that leverages concurrent microwave and solar radiometry to extend the radiometric retrievals (to the low end) where microwave is signal challenged.

Final Presentation¶

• Link to Final Presentation