Final Report Summary - SUBGRID-SCALE CLOUDS (Improving subgrid-scale cloud parameterization in global climate models using remote sensing data)
Project objectives
It is widely recognised that clouds represent the weakest link in future climate projections that use global climate models. Wide gaps remain in our ability to describe fine-scale cloud structures and processes in the generally coarse-resolution global models and to measure cloud macro- and microphysical properties from space. Our research group aimed at improving the model parameterisation of unresolved small-scale clouds, evaluating cloud-climate feedback and obtaining a better understanding of uncertainties in the remote satellite-sensing of clouds.
Work performed
In this endeavour, modelling and observation go hand in hand because satellite measurements serve both as inputs to cloud parameterisations and as benchmarks for evaluating the fidelity of the resulting model climate.
We implemented a novel statistical description of subgrid-scale clouds in a global climate model. Comparisons of modelled clouds with satellite measurements indicated that the new parameterisation approach not only improved the agreement between simulations and observations, but also revealed the remaining issues with the representation of cumulus and cumulonimbus clouds.
The modes of cloud-climate feedback on the global as well as regional scales were assessed. We found that the global climate is particularly sensitive to clouds in the lower atmosphere above the tropical oceans. On the regional scale, our results showed that the minimum Arctic sea ice extent in 2007 was in part due to the unusually low cloud levels that year. We also found deforestation in the Amazon region impacting clouds and, thus, radiation in a way that implies a positive (warming) feedback.
The project addressed the hotly debated issue of climate forcing due to anthropogenic aerosols that can serve as cloud condensation nuclei and, therefore, alter cloud properties and influence cloud-radiation interactions. Using satellite data to derive statistical correlations and to analyse shipping routes and the weekly cycle, we did find observational evidence for aerosol effects; however, this was generally less significant than that suggested by previous studies.
To guide the application of remote sensing data in the above modelling studies, we completed a detailed comparison of the current state-of-the-art microwave and visible-near-infrared satellite retrievals of cloud liquid water path - a crucial input parameter for our statistical parameterisation scheme. We identified the cloud types and geographic regions for which satellite measurements can be used with some certainty (non-precipitating, homogeneous marine stratocumulus areas) and for which the inconsistency of observational datasets indicate little retrieval skill (precipitating and/or heterogenous clouds, high latitude areas observed at oblique sun). We also catalogued potential deficiencies in the retrieval algorithms which can lead to large differences between the various satellite estimates.
To better understand the discrepancies in observational data, we then analysed synthetic visible-near-infrared retrievals of the liquid water path. The reflected radiance fields of hundreds of simulated cloud scenes were calculated in order to quantify the retrieval errors introduced by the assumption of plane-parallel flat-top clouds, which is the standard practice in operational algorithms. These error tables might be used to statistically correct existing datasets based on measurable parameters, such as cloud heterogeneity and sun/view geometry.
We also evaluated a unique dataset of cloud-motion winds and cloud-top heights derived from multi-angle stereo observations. Information on winds and cloud-top heights is important for numerical weather prediction, climate re-analysis and cloud feedback studies. The main advantage of the stereo technique lies in an accurate height assignment, the lack of which is the Achilles heel of traditional wind/height retrieval methods. We managed to identify several regions and cloud types where stereo winds and heights represent an improvement over the existing data. Our results can strengthen the arguments for dedicated stereo satellite platforms for the purposes of numerical weather prediction or cloud process studies; several such mission proposals are currently being formulated by various groups the world over.
Finally, we analysed a suite of active and passive cloud-top height retrieval techniques along stratocumulus-to-cumulus transition trajectories. Understanding such cloud transitions is a hot research topic due to their profound impact on the planetary and regional albedo; therefore, accurate cloud-top height measurements could serve as crucial constraints for modelling efforts. We found that stereo heights agreed remarkably well with the most accurate laser imaging detection and ranging (LIDAR) system measurements; both techniques showed a clear increase in mean cloud-top height along the main transition trajectories. In contrast, brightness-temperature-based retrievals were found to have large height errors in the case of low-level temperature inversions; such errors could be reduced but not completely eliminated by proposed corrections using various lapse rate formulations. These results further illustrate the potential of a dedicated wide-swathe-width cloud stereo platform that could compliment the accurate point measurements of LIDAR sensors.
It is widely recognised that clouds represent the weakest link in future climate projections that use global climate models. Wide gaps remain in our ability to describe fine-scale cloud structures and processes in the generally coarse-resolution global models and to measure cloud macro- and microphysical properties from space. Our research group aimed at improving the model parameterisation of unresolved small-scale clouds, evaluating cloud-climate feedback and obtaining a better understanding of uncertainties in the remote satellite-sensing of clouds.
Work performed
In this endeavour, modelling and observation go hand in hand because satellite measurements serve both as inputs to cloud parameterisations and as benchmarks for evaluating the fidelity of the resulting model climate.
We implemented a novel statistical description of subgrid-scale clouds in a global climate model. Comparisons of modelled clouds with satellite measurements indicated that the new parameterisation approach not only improved the agreement between simulations and observations, but also revealed the remaining issues with the representation of cumulus and cumulonimbus clouds.
The modes of cloud-climate feedback on the global as well as regional scales were assessed. We found that the global climate is particularly sensitive to clouds in the lower atmosphere above the tropical oceans. On the regional scale, our results showed that the minimum Arctic sea ice extent in 2007 was in part due to the unusually low cloud levels that year. We also found deforestation in the Amazon region impacting clouds and, thus, radiation in a way that implies a positive (warming) feedback.
The project addressed the hotly debated issue of climate forcing due to anthropogenic aerosols that can serve as cloud condensation nuclei and, therefore, alter cloud properties and influence cloud-radiation interactions. Using satellite data to derive statistical correlations and to analyse shipping routes and the weekly cycle, we did find observational evidence for aerosol effects; however, this was generally less significant than that suggested by previous studies.
To guide the application of remote sensing data in the above modelling studies, we completed a detailed comparison of the current state-of-the-art microwave and visible-near-infrared satellite retrievals of cloud liquid water path - a crucial input parameter for our statistical parameterisation scheme. We identified the cloud types and geographic regions for which satellite measurements can be used with some certainty (non-precipitating, homogeneous marine stratocumulus areas) and for which the inconsistency of observational datasets indicate little retrieval skill (precipitating and/or heterogenous clouds, high latitude areas observed at oblique sun). We also catalogued potential deficiencies in the retrieval algorithms which can lead to large differences between the various satellite estimates.
To better understand the discrepancies in observational data, we then analysed synthetic visible-near-infrared retrievals of the liquid water path. The reflected radiance fields of hundreds of simulated cloud scenes were calculated in order to quantify the retrieval errors introduced by the assumption of plane-parallel flat-top clouds, which is the standard practice in operational algorithms. These error tables might be used to statistically correct existing datasets based on measurable parameters, such as cloud heterogeneity and sun/view geometry.
We also evaluated a unique dataset of cloud-motion winds and cloud-top heights derived from multi-angle stereo observations. Information on winds and cloud-top heights is important for numerical weather prediction, climate re-analysis and cloud feedback studies. The main advantage of the stereo technique lies in an accurate height assignment, the lack of which is the Achilles heel of traditional wind/height retrieval methods. We managed to identify several regions and cloud types where stereo winds and heights represent an improvement over the existing data. Our results can strengthen the arguments for dedicated stereo satellite platforms for the purposes of numerical weather prediction or cloud process studies; several such mission proposals are currently being formulated by various groups the world over.
Finally, we analysed a suite of active and passive cloud-top height retrieval techniques along stratocumulus-to-cumulus transition trajectories. Understanding such cloud transitions is a hot research topic due to their profound impact on the planetary and regional albedo; therefore, accurate cloud-top height measurements could serve as crucial constraints for modelling efforts. We found that stereo heights agreed remarkably well with the most accurate laser imaging detection and ranging (LIDAR) system measurements; both techniques showed a clear increase in mean cloud-top height along the main transition trajectories. In contrast, brightness-temperature-based retrievals were found to have large height errors in the case of low-level temperature inversions; such errors could be reduced but not completely eliminated by proposed corrections using various lapse rate formulations. These results further illustrate the potential of a dedicated wide-swathe-width cloud stereo platform that could compliment the accurate point measurements of LIDAR sensors.