Tractor at Peanut Belt Research Station

Climate Change in North Carolina


There are many online resources to learn about global warming and global climate change, including Climate Change Collection by CIRES, Dept of Energy Global Warming page and the very comprehensive reports produced by the Intergovernmental Panel on Climate Change (IPCC). Our goal here is to provide an overview of climate changes in North Carolina, which are not completely disconnected with global warming but are influenced by a number of factors other than just greenhouse gases. Much of this material is adapted from a report provided to the University of North Carolina on request from members of the NC General Assembly in 2008.

There is generally high confidence that global greenhouse gas concentrations and global average temperatures are much higher than would otherwise be observed in the natural cycle, and therefore that humans are influencing our global climate. The best estimates of global average temperature are derived from satellite data, which provide information of temperature from both the land and ocean surfaces. However, these data are generally limited to the past 30 years or so, and so are insufficient for analysis of longer-term climate patterns and global warming trends. The satellite-based estimates we have show warming over the past 30 years, primarily at higher latitudes (Arctic, northern Canada, northern Europe, northern Asia, and Antarctica).

We have a longer (though slightly less accurate) temperature record derived from sensors on land since the mid-1800s. Temperature data from these sensors (thermometers) provide evidence that the northern hemisphere has generally been warming since the early 1900s. Similar estimates are available for global averages, but with higher error assumed since there is less land mass and fewer sensor records from the southern hemisphere.

Even longer-term estimates of global average temperature can be derived from proxies such as ice cores, soil cores, and tree rings. In particular, estimates from ice cores provide fairly accurate estimates of temperature and carbon dioxide concentrations in the atmosphere back over 400,000 years (see Figure 1). These proxy estimates provide strong evidence that carbon dioxide levels are today much higher than in the past 400,000 years, and well beyond any natural cycle of the Earth. A more recent study suggests greenhouse gas levels are higher now than in the past 600,000 years. Similarly, temperature estimates from Antarctica ice cores suggest it is likely warmer than it has been in nearly 100,000 years, although not yet at temperatures higher than those estimated at the peak of the interglacial periods in the past. While these estimates have higher error than more recent sensor and satellite estimates of global average temperature, they provide powerful evidence of global warming.

Global CO2 and Temperature
Figure 1. Estimates of temperature and carbon dioxide concentrations derived from ice cores.

But what does this mean for North Carolina?

Unfortunately, we don’t have any ice core samples in North Carolina that can provide similar estimates at a local scale. Observations of temperature and precipitation patterns are available for a few locations across the state since the early 1800s, with more complete statewide observations available only since the late 1800s. These observational data, most of which have been collected by volunteer observers for generations, provide the underlying basis for our understanding of North Carolina's climate — easily the most complex climate in the eastern United States. Our more recent advances in technology have allowed for improved automated local observations of weather and climate, however data is limited for only the last few decades.

Climate changes, including those associated with global warming, occur over decades and centuries; and so we rely on the longer-term records since the 1800s for analysis of changes to climate on statewide and local levels. For more extended records of climate, we must rely on estimates from tree rings and other proxies of climate. These estimates provide a longer record for analysis, but increase the uncertainty and error.


The data we do have for North Carolina from thermometers generally begin in the mid-1800s, with statewide averages available since 1895. The annual statewide average temperature for NC from 1895-2007 is given in Figure 2, along with the linear trend of the data. If one focused only on the period since the mid-1970s, a clear warming signal is seen. This corresponds well with warming observed in global average temperatures from the best satellite data. However, a review of the entire period of record suggests that the warming since the mid-1970s may not be unprecedented, especially when compared with the warming observed from 1910-1950. Overall, the trend over the 113-year period is flat, with no long-term trend over the period.

NC Statewide Average Temperatures Since 1895
Figure 2. Statewide average temperatures for North Carolina

Figure 3 combines the annual statewide average for NC since 1895 with the global average for the same period, again with linear trends fitted. This provides a nice comparison of NC temperature patterns compared with global averages. Of particular note is that while one can observe a pattern that is somewhat similar between the global and NC averages, the NC temperature are far more variable from year to year. This highlights one of the primary challenges with climate change analysis at local and even statewide scales: local climate variability is so high in NC that significant trends are difficult to deduce. Indeed, this is a classic problem of signal to noise; the temperature noise dominates, so a meaningful signal is hard to find. The noise is averaged and reduced relative to signal when combined to create global averages, but is generally much higher in analysis of local climate observations. A variety of statistical techniques have been used filter a meaningful trend in the data, but none have provided significant trends. Similarly, I’ve analyzed average temperature data by month, season, by region, and at local (individual station) scales. There is lots of variability, but most show the same basic pattern observed in Figure 3.

NC Statewide Average Temperatures Since 1895
Figure 3. Statewide average temperatures for North Carolina, overlayed with global average temperatures.

However, when we separate average temperatures into daily maximum and minimums (highs and lows), we can start to see some meaningful trends. While maximum temperatures show no trend, we do see a significant trend in minimum temperature (morning lows) in urban areas. Minimum temperatures are increasing in many urban areas. These minimum temperature trends are significant, but are not linked to the broader global warming. Indeed, we do not see similar trends at rural locations, suggesting that the observed changes in minimum temperatures are associated with urbanization of our cities and surrounding areas. By changing the materials that cover our land surface, we're changing the thermal properties of the land from material that has lots of moisture and lower heat release (vegetation) to material that provides little moisture and has a high heat release (non-vegative surfaces like roads, roofs, etc). Several other observational and modeling studies in NC and elsewhere confirm this analysis – the changes we're making to our land use patterns are much more closely linked to local-scale changes in minimum temperatures than any broader global warming. Certainly, this is another example of how humans can and do impact climate at local scales, but it's not evidence of any impacts from greenhouse gases and global warming to North Carolina's climate.

The State Climate Office has also analyzed dozens of other temperature data (including seasonal patterns, days with extreme high and low temperatures, a variety of temperature thresholds, degree days, growing season data, frost data), and one common thread joins them all: the trends are generally flat, and even when there is a positive or negative trend, the variability is always too large to get a meaningful or significant signal from the noise.


If our estimates of global temperature are somewhat accurate, our estimates of long-term precipitation patterns are much less accurate. The nature of precipitation is such that we have not yet found accurate proxies (like ice cores) to show how precipitation patterns might have changed over the past 600,000 years. We rely more on climate models, which themselves do not accurately handle precipitation physics. Indeed, accurate representation of precipitation in a model is generally considered the gold standard; if a model can capture the correct timing, duration, and amount of precipitation, then we are confident that all the other model physics are very accurately simulated. But the best climate models we have still struggle with precipitation. Advances in the science and computing power will hopeful improve this limitation over the next 20 years.

And if the historical temperature patterns for North Carolina are highly variable and noisy, then the precipitation patterns are even more so. Figure 4 shows the statewide average precipitation from 1895-2007 with a linear trend similar to Figure 2. Note that the linear trend is slightly positive over this period, but that this trend is so slight as to not be meaningful (i.e. statistically significant) given that the year-to-year variability is so high. Again, the signal-to-noise ratio is too low to detect a significant trend.

NC Statewide Precipitation Since 1895
Figure 4. Statewide average precipitation for North Carolina.

Unlike temperature in NC, we do have longer-term estimates of seasonal precipitation during the growing season derived from tree rings. Tree rings have been analyzed across the US and provide a meaningful historical record of tree growth. And while there is higher error than data from precipitation gages, tree ring data can be calibrated to measures of wet and dry periods from modern sensors. Figure 5 shows estimated precipitation during the late spring and early summer for North Carolina for nearly the last 1000 years. Similar to statewide patterns since 1895 (Figure 4), this longer-term data suggest no trend and high variability. Indeed, it suggests drought periods in the past that may have been more severe than those witnessed in NC in modern times.

NC Statewide Average Temperatures Since 1895
Figure 5. April through June precipitation, estimated from tree ring data.

In summary, analysis of precipitation observations and estimates for North Carolina show no significant trend. And like the signal seen in temperature analysis, the variability in precipitation is high.

Severe Events

While analysis of general temperature and precipitation patterns in North Carolina allows us to investigate broader climate changes, our economy and communities are most sensitive to severe weather and climate events. Indeed, North Carolina experiences almost every kind of severe weather patterns in existence (sand storms don't have much impact here). So to provide a more comprehensive assessment of climate change and impacts to NC, one must include an analysis of severe events.

Unfortunately, our records of severe events are not as extensive as our records and estimates of general precipitation and temperature patterns, both globally and locally. For North Carolina, our best data on severe thunderstorms, high winds, hail, flooding, and tornadoes only go back to the 1950s. And since these severe events are highly localized, the data are highly sensitive to population density. Prior to modern weather radars, we relied on citizens to report these events. So we only have data on reported severe events, not the numbers of actual events. However, we do have good estimates of landfalling hurricanes in NC back nearly 150 years and estimates of drought back over 1000 years.

The available data on severe thunderstorms, high winds, hail, flooding, and tornadoes are limited, and generally show no trend. Since these kind of severe events are more rare, there are relatively few samples and therefore our techniques for analyzing trends and climate pattern shifts becomes more limited. Generally, there are insufficient data on these events to determine any climate pattern shift or trend of the period or record. While some theory and models suggest we may see and increase in severe events like strong thunderstorms and floods, the data we have do not suggest we’ve experienced any meaningful increase so far.

Since we have more extensive data on hurricane frequency and intensity, more detailed analysis is possible. Our most accurate global information on hurricane activity and intensity is limited to only the past 30+ years since this is the period when satellites have been able to accurately monitor the open oceans for such storms. And over the past 30+ years we've observed an increase in tropical storm activity, especially since 1995. But studies of global ocean circulation theory and observations suggest there is a natural oscillation in ocean heat content of approximately 25-30 years that is closely linked with tropical storm activity. Analysis of the more extensive record of landfalling hurricanes better identifies this oscillation. And in North Carolina, we certainly see the same pattern: the 1940s, 1950s, and early 1960s were generally a very active period for hurricanes to impact NC, while the 1970s, 1980s, and early 1990s were a very inactive period. Similarly, NC's experienced an increase in hurricane landfalls since 1996 that is likely to continue for the next 15 years or so. But trend analysis of landfalling hurricane records suggests that there is no meaningful increase over the entire 150+ year record.

The best global climate models suggest the tropical Atlantic Ocean will get warmer, and hurricane strength is directly linked to ocean temperatures suggesting that in the future the warm oscillations may produce more intense hurricanes. Unfortunately, the historical record doesn't provide the detail we need to validate this theory. Indeed, there are other factors in the atmosphere that also influence tropical storm development and intensity, and we certainly don’t yet have sufficient understanding of tropical storm dynamics to be confident if future scenarios of our global climate will produce more frequent or intense hurricanes. The impact of global warming on hurricane frequency and intensity is very much an area of active research.

We have a much longer record of drought in North Carolina. However, drought is a more complex event, with no clear beginning or end, and with impacts that very greatly by economic sectors and community. Drought vulnerability is also very closely linked with available water supply and water demand. But drought is also by far the most costly of the natural disasters, especially to agriculture production. Historical analysis of drought severity and frequency shows that in the last 10 years, NC has experienced 2 very severe events (2002, 2007). Figure 6 shows annual drought severity as measured by four different indices and a composite of all four since 1895. Since there is no single best measure of drought, several indicators are used to capture the broader historical pattern. While NC has certainly felt the impact of the 2 recent droughts in 2002 and 2007, the last 10 years may not be unusual when compared with the period 1924-1934. Similarly, analysis of tree ring estimates suggests that drought is not a modern issue, and that NC may have experienced more severe droughts in the distant past (see Figure 5 as an example). However, with increasing population and demand on our water supply, we are increasing our vulnerability to drought. It is also important to note that we have the ability to reduce our vulnerability through more efficient use of water and development of new water resources. Drought is not a new hazard for North Carolina, and we are sure to experience it again.

NC Statewide Average Drought Indicators Since 1895
Figure 6. Annual NC statewide drought severity as measured by four different indices and a composite of all four since 1895.

Climate Model Evaluation and Implications for the Future

With the advancement of numerical modeling we have a wealth of climate model resources, in additional to the sensor data discussed thus far, to analyze past climate patterns and provide guidance for the future. The global climate models that are used by the IPCC assessments generally do a good job in simulating global average temperatures over the past 50-100 years, and so we have confidence in their ability to predict future global average temperatures with some accuracy. Indeed, the models are another source of evidence for the impact of greenhouse gases to global warming. When models simulations are run that include greenhouse gas emissions and physics, the models fairly accurately simulate global temperatures. However, when models simulations are performed with greenhouse gas emissions and physics removed, the models simulate a global temperature pattern much cooler than that observed.

Unfortunately, the best global climate models we have do not do a good job of simulating the temperature and precipitation patterns over the southeastern US, and NC in particular. An ongoing evaluation of the models used in the latest IPCC assessments shows that all of the models predict warming over the past 50 years that has not been observed. Similarly, the models generally don't accurately simulate the location and amount of precipitation over the southeastern US and North Carolina. Some of this is likely due to limitations in computing power – the models solve the physics of the atmospheric at regular geographic intervals, but the intervals are too large to properly handle local weather and climate patterns. As computer power increases, future simulations will be able to better simulate local weather and climate dynamics. Another factor that may be producing the errors is in how the climate models handle aerosols. Aerosol physics is a research area with large error, and sulfate aerosols in particular may have prevented warming over the past 30-40 years. As researchers improve our understanding of aerosol dynamics, the models may improve their ability to simulate past conditions over NC.

So the best climate models we have don't currently do a very good job simulating NC's climate. It doesn't necessarily mean the model predictions for the future are wrong, but it does mean that we have very low confidence in their future predictions. With such high errors in simulating the past over NC, and low confidence in their future outlook for NC, these models simply do not currently produce useful guidance for what global warming impacts might occur in North Carolina.

As the State Climatologist, this is very frustrating for me. Most of the agencies and businesses that come to us are concerned about the future and what might happen at the state and local level. Certainly our society is more aware of global warming, and more concerned about it impacts. But while there is substantial climate awareness, there is little climate education. And the best scientific models we have are not useful for NC, which leads to increased uncertainty for our clients and partners. Moreover, we often see climate model simulations improperly used without calibration or even an evaluation of their accuracy. Research to calibrate the global climate models to NC and downscale their forecasts is very much needed to provide more confident guidance on the possible impacts to NC. In addition, resources for climate education and climate applications are required to meet the needs of statewide and local agencies and businesses.

So what is likely to happen in the future? What is North Carolina's climate going to look like in 50 to 100 years?

These are the most common questions asked, and very challenging ones to answer. The observations we have in North Carolina are limited, and generally show no meaningful trend that is associated with global warming. And the best climate models we currently use have been shown to inadequately simulate past conditions in NC. Until the models improve or new trends are discovered in the historical climate record, we are likely to continue to answer the above questions with a reluctant "We don't know".

There is one aspect of North Carolina's environment that is closely linked to global warming: our rising sea level. While this is mostly outside our area of expertise (there are many others with training and experience with coastal marine and geological dynamics), we do mention it here for completeness. Observations show that the sea level on NC's coast has risen approximately 1 foot since observations began in the 1930s, and this is closely linked to broader global warming. As the ocean warms, it expands and sea level rises. As continental ice melts, that water flows into the ocean and the sea level rises. Ice that is over the oceans doesn’t contribute to sea level rise (the example that I often hear to explain this phenomenon is that of ice floating in a glass of water - as it melts, the water level doesn’t change). And certainly as a coastal state with a dynamic coastline and large estuarine system, North Carolina is sensitive to a rising sea level. Future climate forecasts suggest the sea level may increase by another 2-3 feet over the next 100 years. Unless this increase happens very quickly, there will not be a dramatic inundation of the barrier islands and estuary grasslands and bogs. Since our coastline is made of dynamic shifting sands and bogs that slowly grow with the rising sea level, out coast line may look different in 100 years than it does today, but it will likely not be a simple increase of 2-3 feet of water over the current land structure.

It should also be noted that since NC already experiences almost every kind of severe weather and climate, and that our best science suggests these events will not become less severe (and may become worse), our State can do a lot to manage future risk by taking steps to mitigate our vulnerability to current hazards. Indeed, there is much that we understand about current climate patterns and our State’s vulnerabilities. Our known vulnerabilities to winter storms, hurricanes, flooding, and drought serve as strong reasons for action even in the absence of global warming. By simply addressing known risks to sensitive geographic area and communities, much of the uncertainty associated with future hazards could be reduced. This is applicable to a wide range of sensitivities, not just emergency response and mitigation. Successful examples of climate science applications to transportation, agriculture, and natural resource management have shown the benefits of current climate information and extension services to provide valuable decision support tools and save our agencies and businesses substantial money.