Research Features

Clouds in Midlatitude Storms

Image: See caption following
A midlatitude storm dominating the Eastern half of the continental US. The picture is an IR image from the GOES-East satellite.

Clouds are a major source of uncertainty in scientists' efforts to understand and predict climate change. The problem lies in large part with the fact that the relationships between cloud properties and atmospheric conditions, while well understood in the microphysical scale of a cloud droplet, are not well known in the large scale of a cloud system. This is particularly important in the middle latitudes, where clouds are organized along large systems known as midlatitude storms. Climate models, the computer tools scientists use to predict climate change, use representations of microphysical processes to produce large-scale cloud systems. These representations produce instantaneous cloud fields that often are not realistic in their structure and properties. This makes it highly uncertain that, in a climate change scenario, the model clouds will respond appropriately to the changing atmospheric conditions. The improvement of the representation of midlatitude storm clouds in climate models is recognized as an issue of immediate priority and is one of the major goals of the Global Energy and Water Cycle Experiment (GEWEX) Cloud System Study (GCSS) project.

Working Group 3 of GCSS (WG3) was formed to study the properties and formation mechanisms of clouds formed in midlatitude storms. The approach taken by the working group is to first use a suite of atmospheric models to simulate the clouds formed during field experiment campaigns that observed midlatitude storms in locations throughout the world. Then, comparisons between the model outputs and validations against observations will be used by the group's scientists to improve model representation of midlatitude storm clouds.

The ISCCP/WG3 Datasets

As part of the working group's efforts, a dataset was constructed at NASA/GISS that uses satellite retrievals from the International Satellite Cloud Climatology Project (ISCCP) dataset to resolve the cloud properties during the period and over the location of the field experiments studied by the group. This dataset is available on-line in the ISCCP/WG3 pages, which give a visitor the opportunity to examine the cloud properties in a storm located in the middle of the continental USA (WISP), another crossing the southern coast of Australia (CFRPIII), several storms located over the eastern coast of Canada (CASPII) and a number of arctic storms that cross the basin of the Beaufort Sea (BASE). Some images and information about these datasets are presented below.


Sample WISP image Sample CFRP image
The Winter Icing and Storms Project (WISP) field studies were conducted in the Colorado Front Range area over a period of two years. The above image shows Cloud Optical Thickness on 3/14/90. The Australian Cold Front Research Project (CFRP) Phase III case study focused on a cool change passage that took place over the south coast of Australia. The above image shows Cloud Top Temperature on 11/19/84.
Sample CASP image Sample BASE image
The Canadian Atlantic Storms Program (CASP) II was conducted near the Avalon Peninsula of Newfoundland, Canada, to improve understanding and prediction of the mesoscale structure of east coast storms. The above image shows Cloud Optical Thickness on 2/26/92. The Beaufort and Arctic Storms Experiment (BASE) study was a Canadian-led international field campaign to study the weather systems occurring in southern Beaufort Sea and surrounding areas of the Arctic. The above image shows Cloud Top Pressure on 9/8/94.

A Study Of The CFRPIII Storm Case

[Sample image from animation]
An animation (1.0 MB MPEG) of the lifetime of the storm's cloud top temp.

The Australian CFRP III storm case was picked as the first to be studied through the use of atmospheric model simulations and through comparisons with observations. The storm developed over the southern coast of Australia and was followed for three days of its life cycle (Nov. 17-19). The atmospheric models used in the study simulated, among other fields, the top pressure, top temperature, and optical thickness of the clouds, which makes possible direct comparisons between the model outputs and the ISCCP satellite retrievals.

The CFRPIII storm case was simulated by several atmospheric models. In this discussion, we will use the DARLAM model as an example to illustrate the use of model/observation comparisons to validate and eventually improve model cloud representations. We will examine cloud fields from the model simulation during two days in the life of the storm, one at its beginning and one at its mature stage, and we will compare the model cloud properties to the ones retrieved from the ISCCP satellite observations.

The DARLAM model is a limited area model that was run over a 30x20° domain centered on the experiment, with a resolution of 30 km. Its output was sampled at 0.5° resolution and is compared to ISCCP CX retrievals also sampled at 0.5° resolution. The parameters discussed are:

Cloud Top Pressure
The atmospheric pressure at the level of the top of the cloud. In the ISCCP dataset it is derived from the infrared radiances measured by the satellites and the vertical temperature profile of the atmosphere. In the models it represents the average top pressure of the clouds in a grid box.
Cloud Optical Thickness
An index of the cloud's ability to reflect solar radiation. In the ISCCP data it is derived from the reflected visible radiances measured by the satellites. In the models it represents the column optical thickness of the clouds in a grid box.
Histogram
A plot that shows the number of clouds in the study area which have a particular optical thickness and cloud top pressure. The combination of these two parameters defines the type of cloud (e.g., cumulus, cirrus, etc.).

The CFRPIII Storm Case: Beginning Stage

A comparison of the model cloud optical thickness field with the satellite retrieved values at the beginning stage of the storm (Nov. 17) is shown on the right.

The ISCCP DX data is on the upper left side, the DARLAM model output on the upper right side and a difference plot of the two below. On the difference plot, red indicates that the model simulates optically thinner clouds than those observed by the satellite. Blue indicates that the simluated clouds are optically thicker than those observed.

This comparison shows that the model forms clouds along the storm that are generally optically thicker than the observed ones.

[photo]

A comparison of the model cloud top pressure field with the satellite retrieved values for Nov. 17 is shown at right.

In this case, red in the difference plot indicates lower cloud top pressures (higher clouds) in the model output than in the ISCCP DX data.

This comparison shows that the model cloud tops are generally too high compared to the observations.

[photo]

When cloud top pressure and optical thickness are put together to form a cloud type histogram (see below), it reveals that at the begining stage of the storm the model makes mostly cirrostratus and cirrus clouds while the satellite observes mostly altostratus and altocumulus clouds.

A Study Of The CFRPIII Storm Case: Mature Stage

A comparison of the model cloud optical thickness field with the satellite retrieved values at the mature stage of the storm (Nov. 19) is shown on the right.

The ISCCP DX data is on the upper left side, the DARLAM model output on the upper right side and a difference plot of the two is below. On the difference plot, red indicates that the model simulates optically thinner clouds and blue optically thicker clouds than those observed by the satellite.

This comparison shows the model making clouds optically thicker than the observed along the front and optically thinner behind it.

[photo]

A comparison of the model cloud top pressure field with the satellite retrieved values for Nov. 19 is shown on the right.

In this case, red in the difference plot indicates lower cloud top pressures (higher clouds) in the model output than in the ISCCP DX data.

This comparison shows the model producing clouds that are somewhat lower behind and somewhat higher in front of the storm than the satellite observed ones.

[photo]

When the cloud type histograms of the model and the satellite are compared for the mature stage of the storm (see below), it can be seen that both produce the primary peaks in the stratocumulus, deep convection, and cirrus categories.

Discussion

This preliminary analysis indicates that the DARLAM model produces more realistic cloud properties at the mature stage of the storm where the atmospheric forcing is strong, than at the beginning stage of the storm where the atmospheric forcing is weak. At the beginning stage the model produces clouds that are generally too high and too optically thick compared to the observed ones. It is comparisons like the one presented here, preformed over a multitude of storms and with a multitude of models, that we hope will enable us to identify and correct the problems in the models' representations of midlatitude storm clouds.

Contact

Please address inquiries regarding this research to Dr. George Tselioudis.