Over climate time scales of thirty years or more, average temperature and related indicators in the Mobile Bay region are expected to reflect the multiple influences of global change, modified by local factors including topography, small-scale feedback processes, and land use. The magnitude and rate of global change depends on human emissions of heat-trapping gases, as well as on the sensitivity of the Earth's climate system to those emissions. As global change increases in magnitude, its influence on local-scale climate is likely to grow.
This section summarizes historical observed trends and the changes in temperature and temperature-related secondary indicators that are projected to occur in response to global change. Projected changes are consistent across all five stations; unless otherwise indicated, plotted values correspond to the average value across the five stations.
Annual and Seasonal Temperatures
Historical observed trends in average temperatures vary by month and station. In this analysis, trends were detected using a Mann-Kendall trend analysis. A "significant" trend is defined here as one with a p-value less than 0.1, and a "consistent" trend is one where multiple stations have a significant trend with the same sign of the Kendall1: either negative (decreasing) or positive (increasing).
For average, seasonal, and monthly temperatures there were no historical trends for the period 1960 through 2010 that were significant and consistent across all five stations. When the definition of consistency was relaxed to require only four out of five stations showing a trend in the same direction, negative trends (or cooling) were detected for average minimum temperature for three indicators: monthly values in April and September, and seasonal values for Fall (SON).
Future projections assume that local factors, including topographical influences, changes in land use, and small-scale feedback processes that determine the response of local climate to larger-scale influences, remain invariant. The only factors permitted to change in these future projections are the magnitude of global climate change, and its influence at the regional scale.
Figure 3.1 Change in Southeast annual average minimum and maximum temperature, in degrees F, projected to occur under global mean temperature increases of 1oC (top), 2oC (middle) and 3oC (bottom). Shown here is the mean value as simulated by the ten climate models used in this analysis. Projected changes for the Mobile Bay region are consistent with the broader changes projected to occur across the Gulf Coast.
Figure 3.2 Projected change in average monthly temperature for 2070-2099 relative to 1980-2099, in degrees F, for the higher (A1fi), mid-high (A2) and lower (B1) emission scenarios. Values averaged across all climate models and stations.
Under these assumptions, annual average temperature is expected to increase in the future. Over the next few decades, projected temperature changes are expected to be similar regardless of the emissions pathway followed over that time (the contribution of scenarios to future uncertainty is discussed further in Section 5). This uniformity arises due to the inherent lag time inherent in the climate system, as well as the lags built into our energy system (i.e., it is unrealistic to consider a scenario where all fossil fuel use could be nearly eliminated within a decade or two). The majority of the changes that will happen over the next few decades are the result of heat-trapping gas emissions that have already built up in the atmosphere or are already entailed by our existing infrastructure (Stott & Kettleborough, 2002).
By 2010-2039, annual temperature is projected to increase by an average of +1.5oF across all scenarios.2 By mid-century, increases range from 2.4 to 4.6oF by 2040-2069, depending on the future scenario. By the end of the century (2070-2099), projected increases under higher emissions (+7.7oF) are more than double those expected under lower emissions (+3.2oF).
Changes in the Mobile Bay region are consistent with those projected to occur across the larger southeastern U.S. and the Gulf Coast (Fig. 3.1). In general, slightly greater changes are projected for minimum as compared to maximum temperature, and for inland as compared to coastal regions. Geographic differences likely reflect the moderating influence of the Gulf of Mexico on coastal temperatures.
Both observed and projected future temperature changes vary by season. In the future, over climate timescales of 30 years or more, temperature in all months is projected to increase. Greater warming, on the order of 1oF under lower emissions and more than 2oF under higher emissions by end-of-century, is expected in later summer and fall as compared to other months (Fig. 3.2). The smallest amount of warming is projected in winter months. Under higher emission scenarios, the season of higher warming extends from May through October.
For extreme temperature indicators, there were no historical trends for the period 1960 through 2010 that were significant and consistent across all five stations. With the definition of consistency relaxed to require only four out of five stations showing a trend in the same direction, increases were detected in the number of consecutive days per year over 95oF and 100oF. Decreases were observed in the 5th and 25th percentile of warmest consecutive 4 days of the year, as well as in the 50th percentile and mean of the warmest consecutive 7 days.
As mean temperatures increase, extreme heat is also expected to become more frequent and more severe. Calculations for a large number of heat indices are summarized in the Excel files developed as part of this work. (Note: these are provided as an appendix in the Task 2 report.) Here, representative results from these extreme heat calculations are highlighted, including projected changes in: the hottest day of the year, hottest consecutive 7 days of the year, and hottest day in 30 years or three decades (Figure 3.3), as well as projected changes in the number of days per year over 95 and 100oF (Figure 3.4).
Calculations of the hottest 7 days of the year sample from the 98th percentile of the distribution; the hottest day of the year, the 99.7th percentile; and the hottest day in 30 years, the 99.99th percentile. The degree to which climate simulations are accurate at these percentiles is evaluated in Section 5 using a cross-validation technique to reproduce the historical observations and quantify the biases in simulated vs. observed. In general, however, projected changes beyond the 99.9th percentile of the distribution, such as projections for the hottest day in 30 years, should be taken as qualitative rather than quantitative in nature. The statistics used here to relate regional climate to global-scale change are not intended to be accurate to that scale.
The hottest 7 days of the year historically averages just under 95oF (Figure 3.3a). Within the next few decades, the average temperature of the hottest 7 days of the year is projected to increase by approximately 1.5oF. As indicated by the fact that the average values for each scenario fall within the error bars of the others on the plot, any difference between scenarios over this time scale is purely the result of differences in natural variability between the model simulations. By mid-century, the hottest 7 days of the year are projected to range from 97 to 99oF, with some differences beginning to emerge between scenarios. By the end of the century, average temperature on the hottest 7 days is projected to average 98oF under lower emissions (+3oF relative to the historical period) and almost 102oF under higher emissions (+7oF relative to the historical period), with a statistically significant difference between the values projected under higher vs. lower emissions, as indicated by the fact that the values projected for the mid-high A2 and higher A1FI scenarios lie outside the error bars for the lower B1 scenario. These increases are very similar to those projected for the mean, suggesting that the mean of the distribution could increase at the same rate as the 98th percentile of the distribution.
Sampling from the 99.7th percentile of the distribution, the hottest day of the year for the five weather stations currently averages between 96 and 97oF (Figure 3.3b). This is expected to increase to 98oF by 2010-2039. By 2040-2069, the hottest day of the year is expected to average between 99 and 101oF, depending on emission scenario. By the end of the century, the hottest day of the year could average more than 103oF under higher emissions (+7oF relative to the historical period), or 99oF under lower (+3oF relative to the historical period). Again, the magnitude of this increase is very similar to that projected for the hottest 7 days of the year, and the mean value of the distribution (average annual temperature).
Daily maximum temperature for the hottest day in 30 years currently averages around 101oF. This index is also projected to increase in the future, with some indication of greater changes under higher emission scenarios as compared to lower (Figure 3.3c). Increases are approximately the same magnitude as projected for the hottest day and week of the year; however, these type of climate projections are not intended to be accurate to the one-in-10,950th day or the 99.99th percentile of the distribution. Hence, these results should be interpreted cautiously, with greater emphasis on qualitative direction of trends rather than quantitative numbers.
Figure 3.4 Projected change in individual (a, b) and consecutive (c,d) days per year over (a,c) 95oF and (b,d) 100oF for the average of the 5 weather stations as simulated for the B1 lower, A2 mid-high, and A1FI higher emission scenarios averaged over 10 (B1, A2) and 4 (A1FI) independent climate model simulations. Error bars show range of projected values for A1Fi and 2s range for A2 and B1 (i.e., one s above and below the mean).
Temperature thresholds also show increases in the number of days per year exceeding a given value. Projections were requested for number of days exceeding 95, 100, 105 and 110oF, as well as for the maximum number of consecutive days exceeding those thresholds. For all five weather station locations, there are currently no days per year over 105oF and no significant changes in this number are projected for the future; hence, only projections for days over 95 and 100oF are shown here.
On average, the Mobile Bay region currently experiences between 8 to 9 days per year above 95oF, with 4 to 5 or just over half of those days occurring during one single consecutive period (i.e., during a heat wave) and very few days per year over 100oF (Fig 3.4).3 The average number of consecutive days per year over both 95 and 100oF already show significant trends from 1960 through 2010. As average temperatures increase, the number of days per year over 95oF is projected to increase, as is the number of consecutive days. Projections of nearly 40 consecutive days over 95oF under higher emissions and 13 consecutive days under lower emissions suggest that heat waves, if defined as occurring when maximum daily temperatures exceed 95oF, could increase in length by as much as a factor of 10 under a higher emissions scenario, and 3 under a lower scenario. The number of individual days per year is projected to be approximately twice the number of consecutive days.
Days with maximum temperature exceeding 100oF are currently rare (the hottest day on record in Mobile, AL is 105oF in 2000). Within the next few decades, several such days are expected each year. By mid-century, between 1.5 and 6 days per year could be over 100oF, depending on the emission scenario. By the end of the century, an average of 3 individual days and 2 consecutive days per year over 100oF are projected under lower emissions and up to 20 individual days or 8 consecutive days per year under higher emissions. The analysis for days over 100oF samples from the far tail of the distribution of daily temperature so projections are less robust than for less extreme temperature thresholds.
For extreme cold temperature indicators, there were no historical trends in the indicators requested for the period 1960 through 2010 that were significant and consistent across all five stations.
Figure 3.5 Projected change in the temperature of the coldest day of the year (or the 99.7th percentile of the distribution of minimum temperature) for Mobile, AL as simulated for the B1 lower, A2 mid-high, and A1FI higher emission scenarios averaged over 10 (B1, A2) and 4 (A1FI) independent climate model simulations. Error bars show range of projected values for A1Fi and 2s range for A2 & B1 (i.e., one s above and below the mean).
As average temperatures increase, however, cold temperatures are also expected to become less frequent and less severe. For example, the temperature of the coldest day of the year (currently 18oF, which is the 99.7th percentile of the distribution) is expected to increase by +2oF, to 20oF, within the next few decades, and by an average of +3-4oF, to 21-22oF, by mid-century (Fig. 3.5). By the end of the century, there is some difference between the expected temperatures under lower emissions (21oF) as compared to higher (24.5oF, or an increase of 6.5oF compared to historical).
Average winter temperatures are projected to increase across the distribution, from the 5 to 95th percentile, by an average of 5-6oF under the higher A1fi and mid-high A2 scenarios, and 3oF under the lower B1 scenario by the end of the century. The contribution of scenarios to overall uncertainty is discussed further in Section 5.
1 The p-value measures the probability of obtaining a given value in error. A p-value of 0.1, therefore, indicates 90% confidence that the trend detected is real. The Kendall tau shows how a given variable is correlated with time in order to demonstrate whether a trend is present. The sign of tau indicates whether the trend is positive or negative (i.e. increasing or decreasing with time). Tau values can range from -1 to +1, with larger absolute values indicating stronger trends.
2 In Sections 3 and 4, for the purpose of consistency comparisons across time periods contrast historic simulated values with future simulated values, and not observed or monitored data, unless otherwise noted.
3 Unless otherwise indicated, all figures compare model-based historical with future simulations. After downscaling, the average statistics of simulated climate for 1980 - 2009 are nearly identical to observed data but may not match precisely because global climate models represent slightly different samples or subsets of all possible combinations of the natural variability that could have occurred during that period.