President’s Corner

Weather, Climate, and the Evolving U.S. Climate Change Science Program

"Climate is what you expect; weather is what you get."
—Robert Heinlein, Time Enough for Love

Although it sums up how many think of weather and climate, this adage sets the two apart in a way that is detrimental to the study and prediction of both. Some people study “weather,” others study “climate.” Priorities are often set for one or the other, rather than looking at both. In reality, the two are intimately connected; climate on any time scale is the integral, or summation, of all weather over that time scale. The weather is the devil in the details of climate.

The connection between weather and climate was one of several recurring themes in the U.S. Climate Change Science Program workshop held in Washington, D.C., 3–5 December 2002. The CCSP (see “On the Web”) incorporates the U.S. Global Change Research Program, which was established in 1989 and authorized by Congress in 1990. It also includes the new Climate Change Research Initiative, sponsored by the White House, which seeks to accelerate progress in understanding climate change and to produce decision relevant information in two to five years.

The CCSP Planning Workshop for Scientists and Stakeholders was organized and led by James Mahoney, assistant secretary of commerce for oceans and atmosphere and director of the CCSP. Many senior members of the administration and international leaders presented keynote speeches. Twenty-four breakout groups met to discuss various aspects of the draft of the strategic plan for U.S. climate change and global change research. Altogether, approximately 1,200 people participated—more than twice the number originally expected.

The weather-climate connection was apparent in the discussions and conclusions coming from the workshop. Many users of climate information, from California to New Jersey, emphasized their need for detailed scenarios of potential future climate change on the regional scale. They were interested in possible changes in precipitation, drought, severe storms, wildfires, and other high-impact phenomena, with the changes ideally expressed as probability distribution functions (PDFs).

The interests and requirements of users of weather and climate information point to a number of high-priority items for the CCSP. For example, to produce what these customers are asking for will require enormous increases in computer power, in order to

  • increase the horizontal and vertical resolution of climate models,
  • assimilate the enormous volumes of in situ and remotely sensed observations,
  • increase the number of physical, biological, and chemical processes in the models, and
  • run ensembles of climate projections to generate PDFs of key variables and to define the uncertainties in the projections.

How much more computer power will we need to accomplish the above? One estimate is a million or more times beyond that now provided by the world’s most powerful computer, Japan’s Earth Simulator, which delivers about 50 teraflops of useful computer power. (This is approximately 10 times the power of NCAR’s present suite of computers, which places 10th in the world according to rankings at

Weather data for a changing climate

"Thou shalt not worship the radiosonde."
—Verner Suomi, pioneering satellite meteorologist

Another important priority emerging from the workshop was the urgent need for a robust climate observing system. The global system that has evolved over the previous century for operational weather prediction has provided an extremely important picture of global climate. However, observations that meet the needs of weather prediction often have serious flaws when it comes to defining the climate record.

Radiosondes, one of the mainstays of the global observational network since the 1930s, are a case in point. Radiosondes provide measurements with high vertical resolution of temperature and wind throughout the troposphere and of water vapor in the lower half of the troposphere. The accuracy of these data is good enough to meet the needs of numerical weather prediction. However, there are significant shortcomings when applied to climate. Radiosondes provide uneven coverage globally, since few are launched over the oceans; they do not resolve the diurnal cycle, because most stations launch sondes only once or twice daily; and they do not provide much information from the stratosphere.

Over the years, various temperature and humidity sensors have been used on radiosondes by different countries and by the same countries at different times. Each type of radiosonde has unique error characteristics for temperature and water vapor. The differences, which may be on the order of a degree Celsius and 10–50% in relative humidity, may be small enough not to affect weather prediction significantly, but they are on the same order as the magnitude of decadal trends in climate. Over the past couple of decades, satellites have complemented the radiosonde observations and have addressed some of the shortcomings above. However, satellite estimates of temperature have deficiencies of their own (for example, they generally have poor vertical resolution and suffer from instrumental drift, changes in calibration and orbit, and similar problems). The discrepancies between temperature trends derived from radiosondes versus satellites has been a major source of controversy—a topic discussed by one of the breakout groups in the CCSP workshop.

Recent results from the International H2O Project (IHOP2002) experiment (see figure) showed major limitations of operational radiosondes in measuring the relative humidity in the upper troposphere. Although this region is so cold that there is not much water vapor of significance for weather prediction, even small amounts of water vapor and cirrus clouds in the upper troposphere are extremely important for climate, as they strongly affect Earth’s radiation budget.

Over a dozen radiosonde launches during IHOP2002 bore not one, but two instrument packages. Along with the standard set of sensors, these launches also carried a “reference radiosonde” package designed at NCAR’s Atmospheric Technology Division. The reference package included a state-of-the-art humidity sensor nicknamed Snow White. The sensor used a chilled-mirror technique to measure humidity far more accurately than the two sensor types used on the most popular operational radiosondes.

At heights between about 10 and 14 kilometers (6–9 miles), Snow White often found a thin layer of high humidity, indicating the possible presence of cirrus clouds. In some cases, satellite imagery and/or data from a NASA airborne lidar have confirmed that cirrus clouds were present. Neither of the two standard humidity sensors showed the high moisture content present at these levels, however. Indeed, one of the two sensor types regularly indicated relative humidity below 30% throughout the upper atmosphere, even in cases where cirrus clouds were known to be present.

If this finding turns out to be robust across other regions and circumstances, it will have profound implications for climate monitoring. By and large, cirrus clouds act to warm the lower atmosphere. It is possible that decades of climate records have underestimated the amount of cirrus clouds in the global atmosphere. Future satellite systems may help, but the layers of moisture and clouds at these levels are often too thin for satellites to resolve. The challenge for the weather and climate communities is to work together to recognize and correct such problems. If a radiosonde network is designed for use in forecasting, how can the climate community ensure that its needs are met as well?

At right, a composite of six soundings launched from Dodge City, Kansas, during IHOP2002 shows major differences between the high-precision Snow White humidity sensor (red) and the hygristors used in ATD reference radiosondes (blue) and in the National Weather Service's operational Sippican (VIZ) radiosondes (green). At left, the RS80-H humicap device in the Vaisala sondes (green) shows a weak reflection of the high upper-level humidity observed in six soundings launched from the Homestead site in the eastern Oklahoma Panhandle. (Illustration courtesy Junhong Wang, NCAR.)

This question is a timely one. The National Weather Service (NWS) is replacing its present radiosondes with new GPS sondes over the next four years. While the GPS capability will improve wind measurements, some of the new sondes will continue to use carbon hygristors rather than the Vaisala RS80-H humidity sensors that are now carried on about two-thirds of all U.S. sondes. The IHOP2002 results indicate that carbon hygristors show no response at all to humidity at temperatures below –30C (–22°F), whereas the RS80-H sensors show at least some limited response. Although the new radiosondes could in fact be the best fit for the stated NWS requirements, they fall short of what the climate research community requires to accurately assess global change. -Rick Anthes


Also in this issue...

Putting travel risk on ice

More light, less heat

NSF center boosts space weather modeling

Science Bit: Acid-catalyzed process helps multiply particulates

India steps up its medium-range prediction

Observationalists meet to map out HIAPER sensors

Training the next generation of emergency managers

Unidata’s new director: Mohan Ramamurthy

Web Watch

UCAR Community Calendar

Governance Update