Laboratory measurements of a cell demonstrate that 1mm of liquid water strongly absorbs NIR radiation from 1000 to 3000 nm. An experiment with a spray of water droplets about 1 mm diameter demonstrates that water droplets strongly absorb the same NIR bands as a cell of water. We also have observed the spectrum of liquid water absorption in the transmission spectrum of drizzling fog, whereas dry fog does not show any significant absorption. Clouds have a composition similar to fog, suggesting that drizzle in clouds will cause absorption. The absorption fingerprint of drizzling cloud transmission spectra matches the spectrum of liquid water. The same spectral signature of liquid water in the cloud NIR absorption has been observed from an aircraft during the AIRS Project in January 2000. The spectral signature of ice was also observed on the same clouds. The presence of large droplets in precipitating clouds is evidenced by the formation of rainbows. Liquid water absorption features are not explained by Mie theory for cloud droplets in the size range from 10 to 20 microns. The absorption cannot be simulated using current radiation codes. We postulate that the liquid water in the form of drizzle in clouds absorbs NIR solar radiation. The effect is associated with precipitating clouds and includes Virga in many clouds which does not reach the ground. An explanation is that in clouds there is a population of water droplets with radii > 200 microns causing NIR absorption; drizzle consists of droplets around 500 microns. This large droplet population was observed in some of the in-situ particle size distribution measurements during the AIRS project.

Indeed, spectral measurements with FTIR spectroscopy of the transmission of solar infrared radiation through clear and cloudy skies has indicated that drizzling clouds absorb unexpectedly large amounts of near-infrared (NIR) radiation. This strong NIR absorption effect is not reproduced by the current radiation schemes in atmospheric models. Simulations with a radiative convective model and a mesoscale model indicate that surface temperature errors of several degrees C can be caused by this effect. Current climate and weather forecast models do not reproduce this strong absorption effect because large drops are missing from these models. Hence missing NIR radiation absorption of up to 150 W/m² by drizzling clouds in the cloud schemes of these models which may lead to significant uncertainties in the forecasts of these models, particularly on a regional basis.