Detection of Climate Change

Thomas Karl
National Climate Data Center

Let us start out this by looking at the past climate records. A number of issues arise if one looks at the past records in terms of detecting climate change.

Questions come up, how is the climate warm, has it become wetter, has the circulation changed, how about extreme events, has the climate become more extreme or variable, and how have long term climates varied in the past, beyond the instrumental record that we have? Questions such as these are very relevant to the broader issue of detecting climate change when you come down to the issue of, do the variations that we see in the climate record really reflect climate change, or are they due to natural variations.

In order to answer these questions, it requires some additional questions. Those are related to how well do we know what we think the causes of those changes would be, what are the potential forcings, and how well do we know the history of the changes in those forcings, because it is going to be pretty critical if we are able to piece together what is going on here, that we match changes in one part of our paradigm to another, where we are looking at the forcings.

Once we have asked these questions, then we say, how do we try and match things up? This involves estimates of the real natural climate variability, that is a major issue, and the way in which you go about analyzing the data.

We will start out by taking a look at what we see in the climate records, to give you an appreciation. There is no smoking gun, there is no single study that we will be able to say, ah-ha, we have detected it, ah-ha, we have proven there is no greenhouse effect, or there is no effect on aerosols. It is the culmination of the studies that one has to look at.

I will pose the question, has the climate warmed? I know there are still some people who may be skeptical. At least in the IPCC report, but I think a large number of scientists believe that the temperatures have warmed since the late 19th century. The estimates haven't changed much in the last couple of years, still three to six-tenths of a degree, because of our uncertainty of the past data. But warming is not globally uniform. It is greater over the continents, the mid-latitude continents, than it is over the oceans.

Over the last 40 years, the warming has been about two to three-tenths of a degree Celsius, and the warming has been greater at night in those areas where we have data. It has been greater at night than it has been during the daytime.

Since 1979, when we have had both weather balloons, balloons that are launched every day for operational weather forecasting, and we have also had satellites, both of these different observing systems depict a slight cooling since '79, while the surface temperature has warmed very slightly. There has been a lot of discussion, looking at the satellite records from '79 and saying, there is no global warming, but the satellites show it, so do the weather balloons. You must be very careful looking at a very short period of record, that you don't walk away with a long term broad-scale conclusion, because the climate has to be viewed on the longer time scale.

If you go back to the 1950s, look at the temperature measurements in the troposphere where the weather balloons are measuring, you find a warming very similar to what we see at the surface (Figure 1).

If you look back and say, what about other records besides just simply the measurement of temperatures from instruments, one has a number of proxies that one can look at. These are bore hole measurements of ground temperatures, changes in snow cover, glacier retreat,and sea level rise. All these other measures are broadly consistent with the warming signal that we have seen since the late 19th century.

Doctor Bolin has shown you this Figure One. I include it to remind you that the warming is not contiguous, it is not a simple trend over time, in the Northern Hemisphere, the Southern Hemisphere, and the global average, but there are bumps and wiggles. That is important to consider, because we are going to look at some of these records in terms of this issue of detection.

Now, if we look at the spatial change, some areas are increasing in temperature,some are decreasing. Since 1950, we find some areas of increasing temperature in the northeastern part of North America, over Eastern Europe, and there is are small decreases in China and South America.

The point here is that there is a spatial pattern of change that we can use to try and take into account in our detection efforts not only looking at the global mean temperatures, but looking at those patterns of change, as Bert has pointed out earlier this morning.

I mentioned earlier that temperatures are warming more at night from data that we do have available. Figure Two shows those changes over time. You can see here the maximum temperature, 1950 to '90. It increased, but not as strong as it increased at night, and the difference between the two, the diurnal temperature range, decreasing to some extent, another possible clue as to why the climate may be varying due to a host of causes.

Figure Three is a busy diagram, but I want to include it because I think it illustrates some important points we want to bring up later on. This is directly out of the current IPCC report. The top part of this diagram reflects changes of temperature in the troposphere as well as at the surface. The surface temperatures are given by the dashed lines. You can see, they mirror fairly well the more solid lines represented by the weather balloons. This is what I talked about earlier, a warming since the late '50s in the climate record from both weather balloons as well as at the surface.

Now, if we look at the temperatures since 1979, where we have reliable satellite data or more reliable satellite data, I should say, they are given by the solid dark line, and you can see, there is a reasonably good match between the weather balloons and the satellite data. If you look at the difference between the two, the differences over time are not systematic. That is, there are individual differences in any given year, but you don't see any long term trend, one particular observing system warming relative to another cooling.

Now, if we use the same satellite measurements and the upper air measurements, looking at what is going on in the lower stratosphere, we find that there is indeed some interesting characteristics here, some rather dramatic cooling that has taken place since the mid-1980s. This is not unrelated to the ozone depletion in the stratosphere, another major global issue. In many senses, ozone depletion is not a completely distinct problem from global climate change.

In fact, you can see very clearly in this record the Mt. Pinatubo eruption in the stratosphere, a very strong warming, reflected both by the weather balloons and the satellite. You can see this also reflected in the surface. The surface temperature is cooled, as well as the temperatures in the troposphere. I might point out that the tropospheric temperatures above the surface cooled more with the Mt. Pinatubo eruption than did the surface temperatures. We find that the tropospheric temperatures tend to have more variability associated with them than at the surface.

If we were completely mistaken about our ability to measure global warming, it would be hard to find matches equal to these in Figure Four. On this diagram is reflected Northern Hemisphere snow cover variations. They begin when satellite records began, in the early 1970s. Each year is reflected by the bars. The curves represent snow cover and temperature over those same regions for different seasons of the year, October through November, winter, December to March, spring, summer, and then annual (Figure 5). You can see a very nice match between the temperatures and the change in snow cover, which gives us added confidence that we indeed are measuring a warming signal that is fairly robust.

Another question that has been raised, switching our focus, has the climate become wetter? During the 20th century there has been a small positive increase in worldwide land precipitation on the order of about 1 percent from the best analyses we can produce to date. But perhaps more importantly, the precipitation increase has not been uniform. In fact, precipitation has increased over land in high latitudes, and in the last few decades decreased significantly in the tropics, especially since the 1960s. We believe this is not unrelated to the ENSO effects.

As related to the hydrological cycle, where we have cloud measurements from surface weather observers, we believe that indeed, the cloud cover has increased in many areas over land as well as over the ocean. Travelling ships who record the weather, simply because they are interested in getting accurate weather forecasts, have been doing this for many, many decades.

Over land, the evaporation appears to have decreased since 1951 over the U.S. and in the former Soviet Union. There is a very nice match between this evaporation decrease and the increase in cloud cover. It gives us more confidence about the decrease in evaporation. It gives us more confidence about the cloud cover, and there is a very nice match between those two changes and the decrease in the diurnal temperature range. They all go together: more clouds, less sunshine, less evaporation from the surface.

Over the ocean, we think that evaporation has increased in all areas over the tropics. It is tempting to say as a result, water vapor has been observed to increase over North America and the tropics since about 1973, when we have reliable records.

This is very important, because increases in carbon dioxide initially warm the climate. As the climate warms, more water vapor is put into the atmosphere, which gives an enhancement to the greenhouse effect.

One of the problems in looking at atmospheric circulation is the dearth of good analyses, even where we have good consistent data. There are efforts going on now to re-analyze data, put it into models in a kind of a back-cast mode, so we can better understand how the atmospheric circulation has changed. That is ongoing, and it really needs to continue.

Figure Six is an example of the Southern Oscillation index, as defined by Trenberth and Horne in 1995. I just point out to you the unusual characteristics of the ENSO over the last decade, especially since 1990, the rather persistent behavior. Right now, the ENSO is back up to slightly above the line in this index. But certainly this is an unusual characteristic in this climate record. This all has to be considered in trying to understand what impacts are causing the changes that we observe. It even brings us to a broader question, how is the Southern Oscillation index expected to change as other forcings may impact the climate, such as increases in greenhouse gasses?

Another very important question has been, has the climate become more variable or extreme? I want to begin by saying, this is one case where we have an inadequate number of analyses and not very good databases to say anything very global. We do have some good databases in some regions of the world. Where people have taken a look at that, in some respects we have found some increases in extreme events, but in other areas we found little change.

In particular, there is a clear trend toward increasing extreme precipitation rates in the U.S., as well as Northeast Australia. It just doesn't rain much in Central Australia, as you might have guessed. Other areas -- for example, we have looked at China back to 1950, and we don't see much evidence for change in that area.

Increases in extreme extra-tropical cyclones in the North Atlantic since the late 1980s have been pretty robust. There has been a lot of discussion about hurricanes, especially in this country, because of the activity in the North Atlantic. If you take a broader perspective over the last 40 years, hurricanes have actually decreased, but, over the last decade or so, there has been a tendency for an increase. We have found that the day- to day variability of temperatures actually decreases in the Northern Hemisphere. That is, the difference in temperature between today and tomorrow has actually gone down to some extent, if we look at the mid-latitudes and high latitudes of the Northern Hemisphere over land. If we look at the difference in temperature between one year and the next year, we really don't find any long century scale trend associated with the fluctuations of inter-annual temperature variability.

Hurricanes have gotten quite a bit of discussion in this country this year. Figure Seven compares 1995 to some selected years. You can see other years rivalling 1995 in terms of the number of both tropical storms and hurricanes, a fair number of them.

If we take a look at the number of hurricanes reaching landfall and look at it by decade (Figure 8), one can see a dip during the 1970s and if the rate of hurricane production continues as we see in the first five years of this decade, the projection out to the end of the decade will bring us back up to more typical levels.

This is a very uncertain area in terms of how hurricanes would be expected to change with global warming. The theoretical basis would suggest they become more intense, more extreme. How frequent they are is also subject to changes in circulation. Right now, models aren't capable of giving us good information. Nonetheless, storms are very important, but this is not the type of thing we would use to detect whether or not the climate is changing, and whether or not it is changing due to an increase in greenhouse effect. We simply don't know enough about it.

I mentioned the increase in extreme precipitation events. Figure Nine shows a very interesting signal in the U.S. What you see is a rather remarkable trend, the proportion of precipitation in the United States that is falling from precipitation events of two inches or more on any given day. That percentage has increased from the order of 8 or 9 percent in the early part of the century to somewhere in the neighborhood of about 11 percent at the present time.

This is a very interesting signal. It has been discussed in IPCC as one of the anticipated changes we might expect with an enhanced greenhouse effect, that is warmer climate leading to more extreme precipitation events, a greater precipitation rate.

How have climates in the past varied? We can take a very broad look at what has gone on in the last thousand years to try and put what we have seen in the last century in a better perspective. We can rely on tree rings, historical records, corals, and shallow ice cores. There are a number of proxies that are used.

They all have their limitations, because they are not globally distributed, so they are somewhat difficult to interpret on a global basis. But nonetheless, with that caveat, based on the best available evidence we have today, the 20th century appears to be warmer than any century we see, going back to 1400, based on a collection of these paleo climatic records (Figure 10).

It is also important to point out that the warming began during one of the coolest periods during the last 600 years. In some areas of the globe, from the paleo climatic evidence, we are now clearly warmer than we have ever been, at least in the last thousand years or so.

Going back much further, 100,000 years, there is other paleo climatic evidence that doesn't give you the good resolution, but gives you a broader look at how climate has varied in the past. Based on these records, it seems very unlikely that at any time over the last 100,000 years there has been a change in temperature of more than a degree Celsius in a century. As we mentioned before, it looks like this past century has warmed already half a degree. Projections are anywhere between one and a half and three and a half by the year 2100, clearly larger than anything we have seen in many years.

Let me go on here, in the interests of time, to pose this next question. Do these climate variations reflect real changes? We do have some notion, but our confidence in just how much of an effect this would have on the climate system does vary (Figure 11). The greatest confidence we have is in the direct effect of the greenhouse gasses.

We have other issues to be concerned with, however, if we want to find out what is causing those changes. Other forcings are going on. Depletion of stratospheric ozone, tropospheric ozone,sulfates, fossil fuel soot, biomass burning, the indirect effect of tropospheric aerosols,and solar variability, all are factors. As you may be aware, we only have direct measurements of solar variations since about the late 1970s. Every time we keep putting up a new satellite, we don't have a good overlap in the record and thus, we still have a difficult time in putting together a good long term time series of solar variability, even when we have satellites up there. So many of these time histories have to be done through proxy measurements of solar variations.

Well, how well do we know the histories? I have already given you a clue, at least my assessment. Well-mixed, long lifetime anthropogenic greenhouse gasses -- we have fairly good data since the Industrial Revolution. Earlier than that we have to rely on ice cores. Perhaps a D+ is our ability to discern whether or not changes in carbon dioxide are occurring before increases in temperature or after.

With other greenhouse gasses such as tropospheric ozone we are not doing a very good job. Soot, sulfate aerosols? We can make some guesses from looking at the history of changes in sulfur emissions. Biogenic aerosols? Not doing a terribly good job. This is our current knowledge, which suggest they are minor forcings, but nonetheless, they are important to try and understand if one wants to narrow uncertainties around the observed climate variations and changes.

What are some of these other issues? We have to estimate natural variability, in order to assess whether or not the changes we see are due to any specific forcing. The question arises, what measure do we use? Analyzing the data, what particular statistical technique are you going to use? Can we uniquely assign observed climate changes to specific forcings or combinations of forcings? This is the issue that is going to dog us, because we know we are going to always leave out some of these factors in any of our analyses of climate. We cannot relate them back to changes, because we don't have models that have information about how the histories have changed. Whether or not this could be modelled is another question.

How can we discern the impacts on climate record? With the models, future projections, the more detailed, the less chance we have of being misled by a false signal. If we can look at all the information, it is probably better than looking at the grand global averages.

We do need a measurement of background variability. A number of studies have done this. They have looked at paleo climate information to give an estimate of background natural climate variability. They have looked at GCM model simulations themselves, without any forcings, and trying to use that as a measure of natural variability, or they have used the observed record that we see of the last 100 to 150 years.

The exact answer you get depends on which one of these measures you use, so you could have someone doing the exact same study, coming out with the probability that the change we see is 95 percent certain, it is real greenhouse and not natural variability. If they are using another measure as opposed to this measure, someone else may do that same study and find there is only an 85 percent chance, and it is not statistically significant. This is important to keep in mind, because it is a way in which some apparent conflicts arise in the literature.

How can we discern their impacts? Qualitative analyses, I believe, are still very important, because from a qualitative analysis, you can assess whether there seems to be a general agreement between the predictions and observations.

Quantitative analyses are the bread and butter. These can take a number of forms. They can be multivariate and uni- dimensional, or multi-dimensional, they can look at a number of aspects of the change in a particular variate, or they can include a number of variables, and they can look at all the spatial scales associated with those variables. This we haven't approached yet. We would expect that work would be ongoing in the next few years.

Recent studies have added a lot of information to the climate record, because of the role sulfate aerosols have played.

I want to give you one example of why we think this has been very important. If you take a look at sulfur dioxide emissions since 1980 (Figure 12), the larger circles represent larger emissions. You can see the greater emissions in North America compared to some of the lesser developed parts of the world. But the emissions have not been constant. If we look closer (Figure 13), we find that in the U.S. since 1966, the emissions have actually decreased. Same thing in Europe. Other areas have seen significant increases, particularly in China.

One can use that information and say, what if we were to look at the changes in zonal temperatures and see if there is a spatial pattern across that zone that reflects over this period, 1966 to '80, which would reflect the expected changes.

We would expect sulfate aerosols to have their strongest impact in the high sun season, summer, where there is a lot of solar radiation being reflected back to space. Most of the warming actually occurs in the wintertime, which is perhaps not unexpected, in the sense that there isn't a lot of solar radiation at mid and high latitudes in the wintertime, and there is often snow cover. The effect of sulfate aerosols is not nearly as great in winter as in spring and summer.

Let us examine the SOX emissions in the United States (Figure 14). This shows that they seem to match quite nicely the changes in maximum temperature, without much of an effect on the minimum temperature.

If you only look at carbon dioxide, you find significant warming throughout much of the troposphere, peaking in the high troposphere. If you add in the sulfate aerosols, you find some symmetry, not as much warming in the Northern Hemisphere.

In trying to put together all the forcings and trying to come up with a clear picture, we did an analysis in the U.S., looking at a number of indicators that have been discussed in 1990 and the most recent IPCC report,to determine the types of changes that you would expect with the greenhouse effect in mid- latitude locations.

One of these indicators is the percent coverage with much above normal temperatures. Another is the percent of the country with much above normal cold season precipitation, the precipitation that will enhance greenhouse effect which is expected to increase more during the cold season than at other times. A third is the percent increase in extreme severe droughts during the warm season because this increase in temperature causes more evaporation. As I already indicated, this perhaps has not been the case. The last two are the percent of the U.S. with these extreme precipitation events, and the reduction in the day-to-day temperature variability, another expectation.

If you put all those indicators together in an index, you can come up with a curve that looks like Figure Fifteen. You can see some very high values in recent decades, do statistical analyses, and come away with the notion that it is very unlikely this would have happened simply by chance, maybe a one chance in 20, if there was not some real effect going on.

Figures Sixteen and Seventeen are out of the IPCC report. It is good to stand back and look at the types of projections that have been expected to occur, comparing those to what has been observed. Stratospheric temperatures are expected to cool, we have seen that. The troposphere is expected to warm. Snow cover is expected to decrease. These are all temperature indicators.

Little change or slight decrease of the El Niño Southern Oscillation is expected. That perhaps one might argue has not occurred, there is still some work to be done in that area. Near surface air temperature increases, greater in the Northern Hemisphere mid-latitude areas, especially in the wintertime. We expect reductions in sea ice. Land nighttime temperatures are expected to warm more than daytime. Ground temperatures are expected to warm. There are many changes in the model projections that are consistent with what we see, consistent in a qualitative sense.

So in conclusion, I want to say that there is no doubt the climate has warmed over the last century. It has warmed despite the Mt. Pinatubo eruption, and 1995 numbers would suggest that we probably are going to be very close to one of the warmest years on record. The emerging evidence does point toward detectable human influence on global climate, far more than we were able to show a couple of years ago.

Go To Discussion - Session I