Today, the most up-to-date climate models provide a valuable but still limited understanding of climate system processes. New Columbia Water Center research, however, could spark fundamental advances in our ability to comprehend the complexities of climate by using probabilistic cloud modeling to properly characterize precipitation and solar radiation, and by developing a realistic representation of how plants respond to water stress.
Over the past two decades, scientific understanding and modeling of the climate has improved immensely. However, significant uncertainties remain. Chief among these are the representation of convection (clouds and precipitation) and vegetation’s physiological reaction to water stress.
Convection is poorly characterized in numerical models, especially regarding its frequency and intensity. The so-called “cumulus parameterization problem” continues to plague climate and weather model development and, more fundamentally, limits understanding and prediction of phenomena such as monsoons, El Niño and many others.
In addition, many numerical models predict a peak of photosynthesis and evapotranspiration in the wet season in wet tropical forests; however, observation shows the opposite to be true. Without a correct model of photosynthesis and evapotranspiration, we are unable to predict changes in the tropical carbon and hydrologic cycles. Given that tropical forests are the “lungs of the planet,” providing its main CO2 sink, they are thus a key piece of the human-caused climate change puzzle; understanding and predicting their fate is critical to our future, and the failure to properly model their dynamics represents a major gap in much-needed understanding.
The erroneous representation of the continental hydrological cycle arises from 1) the inaccurate timing of radiation and precipitation at the surface and 2) our inability to represent the physiological response of plants to water stress.
To address the first challenge, Columbia Water Center researchers used a probabilistic approach of turbulence and cloud modeling that will result in improved predictions and correction of most observed precipitation biases (frequency/intensity), compared to conventional deterministic models of convection. This work will help better predict rainfall across multiple temporal and spatial scales, and most importantly help better simulate how a changing cloud cover affects (forces) rainfall and radiation at the Earth’s surface. Results of this research were published in three papers in the Journal of the Atmospheric Sciences.
In addition to obtaining the correct surface forcing, the CWC research group is now working to improve the scientific description of how plants react to water stress, which in turn will help accurately predict the continental patterns of the hydrologic cycle. The team’s new, physiologically-based water stress model will accurately simulate the way plants redistribute water in the soil—a fundamental change from current ad-hoc representations of vegetation water stress. Starting from first principals, the new model is based on the simple fact that water flows from high to low moisture regions. Preliminary results are promising, replicating the observed way that plants use near-surface water in the wet season and deeper soil moisture from groundwater in the dry season. As a result, estimates of carbon and water fluxes are drastically improved and exhibit correct seasonality and response to water stress. These improvements are essential to better predict ecosystems’ response to droughts and climate extremes, which are likely to become more frequent under climate change. In the tropics, this understanding will be key to determining whether tropical forests remain CO2 sinks in the future or if droughts transform these forests into sources of CO2, leading to further increases of atmospheric CO2 concentration.
Finally, the team is constraining and validating its models through new applications of global satellite remote sensing. Using cutting-edge, machine-learning tools, the group is developing new algorithms to observe vegetation water stress, soil moisture and precipitation. Combined with data assimilation, this innovative approach can improve predictions and refine model parameters globally.
All together, this research holds the promise of pushing fundamental advances in the scientific representation and prediction of weather and climate forecasting, as well as the future of state of carbon in the atmosphere.Today, the most up-to-date climate models provide a valuable but still limited understanding of climate system processes. New Columbia Water Center research, however, could spark fundamental advances in our ability to comprehend the complexities of climate by using probabilistic cloud modeling to properly characterize precipitation and solar radiation, and by developing a realistic representation of how plants respond to water stress.
Related Publications:
A Probabilistic Bulk Model of Coupled Mixed Layer and Convection. Part I: Clear-Sky Case
A Probabilistic Bulk Model of Coupled Mixed Layer and Convection. Part II: Shallow Convection Case
An Idealized Prototype for Large-Scale Land–Atmosphere Coupling
Earth, Water and Sky –A Conversation with Pierre Gentine, a new Columbia Water Center Scientist