<<

Summer Regional Diurnal Temperature Range Variability With Soil Moisture Conditions

THESIS

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The State University

By

Robert Wayne Brewer, B.S.

Graduate Program in Atmospheric Science

The Ohio State University

2015

Master's Examination Committee:

Jeffery Rogers Advisor

Jay Stanley Hobgood

Jialin Lin

Copyrighted by

Robert Wayne Brewer

2015

Abstract

Long-term (1895-2012) soil moisture proxy data are collected and analyzed for its spatial and temporal variability across the United States in conjunction with air temperature and diurnal temperature range (DTR) variations over the same period.

Palmer Drought Severity Index (PDSI) summer data were subjected to a Rotated

Principle Component Analysis (RPCA) that identified 10 (components) having unique patterns of PDSI spatial and temporal variability. Four of those regions (RPC1:

Ohio River Valley; RPC2: and eastern Northern Plains; RPC3: southeastern United States; RPC5: Southern Plains) are analyzed further with regard to

DTR variations. In conjunction to the summer PDSI time series scores produced by the

RPCA, mean DTR, T-max, and T-min (maximum and minimum temperatures) were obtained using GHCNM station data within each of the regions of interest and analyzed for trends. The twelve wettest and driest summers were also identified for each of the 4 regions based on the rank of their PDSI time series scores. The average temperature/DTR for each of these cases (wet or dry) were then compared.

Soil moisture in the Valley (RPC1) has an increasing trend throughout the 20th-21st centuries. T-max shows a downtrend of 0.5°C while T-min has increased ~

0.7°C producing a downward trend in DTR throughout the period of record. The upper

Midwest and eastern Northern Plains (RPC2) produced similar behavior as the Ohio ii

River Valley with more moist soil conditions at the end of the 20th and early 21st century.

DTR trends downward in this due to a very clear upward trend in T-min coupled with a negligible downtrend in T-max. PDSI in the southeastern United States (RPC3) does not have a strong trend but does show a slight increase. T-max produces a trivial, but slight increasing trend while T-min shows a stronger increase in temperatures. This outcome produces a decreasing trend in DTR. Soil moisture in the Southern Plains

(RPC5) shows an overall decline in PDSI. T-max produced a long-term increase of ~

0.6°C. T-min produces an increasing trend slightly larger than that of T-max causing a very small decreasing DTR trend. The long-term DTR trends in each region seemed to be mostly influenced by the larger long-term increasing trends of T-min as compared to the smaller trends in T-max. However, DTR during the most extreme soil moisture summers

(wet or dry) seemed to be influenced more by the variability in T-max, as T-min did not fluctuate as much.

The 2012 summer drought was used as a case study to evaluate month-to-month

DTR variations in the context of variations in precipitation and drought conditions. On a statewide and month-to-month basis, 2012 DTR variations almost always declined

(increased) in response to increases (decreases) in rainfall. This variability agrees with that shown in the DTR soil-moisture portion of the analyses.

iii

Acknowledgements

I would like to recognize my family, friends, Dr. Jialin Lin, Dr. Jay Stanley

Hobgood, and especially Dr. Jeffery Rogers. Without your continued support and guidance, I would not be where I am today in my academic career. It has truly been an honor and privilege to surround myself with such knowledge, leadership, and humility during my time at The Ohio State University. Thank you.

iv

Vita

May 2007…………………………………....New Lebanon High School

2011…………………………………………B.S. Atmospheric Science, The Ohio State

University

2013 to present……………………………...Graduate Teaching Assistant, Department of

Geography, The Ohio State University

Fields of Study

Major Field: Atmospheric Sciences

v

Table of Contents

Abstract……………………………………………………………………………………ii

Acknowledgements……………………………………………………………………….iv

Vita………………………………………………………………………………………...v

List of Tables…………………………………………………………………………….vii

List of Figures…………………………………………………………………………...viii

Chapter 1: Introduction……………………………………………………………………1

Chapter 2: Literature Review……………………………………………………………...4

Chapter 3: Data and Methodology……………………………………………………….15

Chapter 4: Results………………………………………………………………………..19

Chapter 5: Case Study: The Drought of 2012 ……………………………………………63

Chapter 6: Conclusions…………………………………………………………………..94

References………………………………………………………………………………..99

vi

List of Tables

Table 5.1. U.S. Drought Conditions: End of August 2012 70

Table 5.2. States Analyzed for the U.S. Drought of 2012 78

Table 5.3. Statewide Averages for May 2012 80

Table 5.4. Statewide Averages for June 2012 82

Table 5.5. Statewide Temperature Changes from May-June 2012 85

Table 5.6. Statewide Averages for July 2012 87

Table 5.7. Statewide Temperature Changes from June-July 2012 89

Table 5.8. Statewide Averages for August 2012 91

Table 5.9. Statewide Temperature Changes from July-August 2012 93

vii

List of Figures

Figure 4.1. Spatial Centers of PDSI Components 1-5 20

Figure 4.2. PDSI Scores for the Ohio River Valley (RPC1) 22

Figure 4.3. PDSI Scores for the Upper Midwest/eastern Northern Plains (RPC2) 23

Figure 4.4. PDSI Scores for the southeastern United States (RPC3) 24

Figure 4.5. PDSI Scores for the Southern Plains (RPC5) 25

Figure 4.6. RPC1 Regional Averages of T-max 28

Figure 4.7. RPC2 Regional Averages of T-min 30

Figure 4.8. Ohio River Valley Regional Averages of DTR 31

Figure 4.9. RPC2 Regional Averages of T-max 33

Figure 4.10. RPC2 Regional Averages of T-min 35

Figure 4.11. Upper Midwest/eastern Northern Plains Regional Averages of DTR 36

Figure 4.12. RPC3 Regional Averages of T-max 38

Figure 4.13. RPC3 Regional Averages of T-min 40

Figure 4.14. Southeastern United States Regional Averages of DTR 41

Figure 4.15. RPC5 Regional Averages of T-max 43

Figure 4.16. RPC5 Regional Averages of T-min 44

Figure 4.17. Southern Plains Regional Averages of DTR 46

viii

List of Figures

Figure 4.18. PDSI Wettest vs Driest Summers: RPC1 48

Figure 4.19. T-max, T-min, and DTR for Wettest vs Driest Summers: RPC1 51

Figure 4.20. PDSI Wettest vs Driest Summers: RPC2 52

Figure 4.21. T-max, T-min, and DTR for Wettest vs Driest Summers: RPC2 54

Figure 4.22. PDSI Wettest vs Driest Summers: RPC3 55

Figure 4.23. T-max, T-min, and DTR for Wettest vs Driest Summers: RPC3 57

Figure 4.24. PDSI Wettest vs Driest Summers: RPC5 59

Figure 4.25. T-max, T-min, and DTR for Wettest vs Driest Summers: RPC5 61

Figure 5.1. Climate Division PDSI Values: May-August 2012 67

Figure 5.2. U.S. Drought Monitor: May-August 2012 Drought Expansion 68

Figure 5.3. End of May-End of August U.S. Drought Monitor Drought Severity Ranks 69

Figure 5.4. Climate Division PDSI Values: May-August 1956 72

Figure 5.5. Climate Division PDSI Values: 1956 Drought Peak vs 2012 Drought Peak 73

Figure 5.6. Climate Division PDSI Values: May-August 1988 75

Figure 5.7. Climate Division PDSI Values: May-August 1934 76

Figure 5.8. United States Climate Regions 78

ix

Chapter 1: Introduction

With increasing global temperatures and the melting of both glacial and polar sea ice, climate change has become reality. The ramifications of climate change have been highly scrutinized both politically and scientifically as extensive research from the past and present try to forecast the potential atmospheric and ecological changes that may come as consequence. One of the oldest climate theories is that of the late Milutin

Milankovitch who suggested that ice ages are related to planetary gravitational influences on the Earth’s orbit around the sun. Based on these principles, Milankovitch Cycles are

10,000-100,000 year variations occurring amidst ice ages as high latitude solar insolation waxes and wanes (Smith, 1990; House, 1995; Ruddiman, 2006; Huybers and Curry,

2006; Berger, 2012). Bringing this into today’s perspective, according to these cycles, the

Earth should be in the midst of an extensive period of high insolation and high temperatures. The distant future however holds a return of declining insolation and cooling. Since the late 20th and early 21st century however, global anthropogenic carbon emissions have increased leading to a more noticeable increase in temperatures than the

Milankovitch Cycles suggests, leading to the underlying principles of anthropogenic global warming and climate change (Shackleton, 2009; United States Environmental

1

Protection Agency, 2012; Hansen and Sato, 2012; National Assessment Synthesis Team,

2014).

In the current era of technology and data collection, the effects of climate change can be readily monitored and evaluated. With simple datasets containing long-term time series comprised of daily temperatures and precipitation, short and long-term trends become apparent as deviations from normal recur in one direction. Soil moisture is one parameter that plays a role in temperature variability, although the role it plays is not entirely understood.

Droughts and extreme precipitation events are expected to occur more frequently with a changing climate (Seager et al., 2009; Gutzler and Robbins, 2010; Mishra et al.,

2010; Mallya et al., 2013). These instances of precipitation variation also influence soil moisture and are a common occurrence of nature that can affect millions of people worldwide every year. It can alter water supplies, crop yields, and livestock which are very important cogs to the well-being of life. Research in problems associated with drought and flooding has expanded, especially in the United States. More recently, ongoing drought has impacted the United States from coast to coast with a protracted drought in the this century and a widespread Midwestern drought in 2012 (Mallya et al., 2013; Grigg, 2014). Crop loss, wildfires, and dwindling water reservoirs have become very common in these areas. In the history of the United States, this is not a rare development. The United States has experienced its fair share of droughts dating all the way back in its documented history to present day (Woodhouse and Overpeck, 1998; Cook et al., 1999). Understanding drought and its causes and

2 consequences is an ongoing topic in climate research. The causes of drought involve numerous meteorological variables including precipitation, air temperature, evaporation and soil moisture (Diaz, 1983; Dai et al., 1999; Hu and Wilson, 2000) and the interactions among these that work together to produce the unfavorable conditions.

In a warmer world, atmospheric water vapor is positively correlated with temperature and will also increase. Because of this, extreme precipitation events have become more common in the United States. This can be linked to long-term positive trends in precipitation as a whole and also an increase in localized extreme precipitation events leading to severe flooding (Kunkel et al., 1999; Groisman et al., 1999; Groisman et al. 2011). Soil moisture indicates that the late 20th century and early 21st century have shown characteristics of more moisture and severe precipitation events (Karl and Knight,

1998; Rahmstorf and Coumou, 2011; Peterson et al., 2013).

Associated with the effects of climate change driven by rising temperatures (e.g. droughts and precipitation events), changes in diurnal temperatures are occurring (Karl et al., 1993; Dai et al., 1999; Lauritsen and Rogers, 2012) along with component daily maximum (T-max) and minimum (T-min) values. Diurnal temperature range (DTR) is one important change potentially related to soil moisture and precipitation variability associated with climate change in the United States. This thesis will investigate historical

DTR to see how it varies through the 20th-21st centuries in conjunction with climate change and also how it varies in conjunction with the recent drought of 2012.

3

Chapter 2: Literature Review

A changing climate alters the typical patterns of regional and seasonal temperatures and precipitation as well as their frequency and severity of events on a daily scale (Shackleton, 2009). This has become evident throughout the end of the 20th century and early 21st century as global temperatures have warmed due to large increases of heat- trapping gases released from human activity (United States Environmental Protection

Agency, 2012; Hansen and Sato, 2012; National Assessment Synthesis Team, 2014).

Because of this warming, moisture levels in the atmosphere have risen because higher temperatures permit more mixing of water vapor into the air creating higher humidities.

Karl and Knight (1998) have shown that 20th century and early 21st century trends in precipitation have increased by about 10% in the United States. The major cause of this rise has been the increased frequency in heavy and extreme daily precipitation events

(Karl and Knight, 1998). Kunkel et al. (1999) made note that recent years with heavy precipitation have resulted in several damaging floods across the United States. One of the worst floods in recent history occurred in 1993 in the upper River Basin resulting in billions of dollars in damage (U.S. Dept. of Commerce, 1994; Eash, 1997).

Instances of dry soil conditions come with extreme precipitation events. Because of increasing temperatures and variations in weather patterns due to climate change, another outcome is severe drought. Drought has become a regular occurring phenomenon 4 in the United States since the 1950s as global temperatures started to rise (Easterling et al., 2007). Drought events have not necessarily trended toward higher or lower frequency, but Easterling et al. (2007) suggest that temperature increases lead to increases in spatial drought coverage while the precipitation increases over the last century have limited the more severe and extreme drought conditions.

By the end of 2012, the United States had found itself in a recovery from one of the costliest and most widespread dry spells in its history (Mallaya et al., 2013). The weak winter systems of 2011 coupled with natural climate fluctuations and low sea surface temperatures stemming from the La Niña of that year, brought very dry conditions to the United States especially in the Midwest and Northern Plains. During this time, monthly mean temperatures in the Midwest were warmer than the long-term averages for the summer months (Mallaya et al., 2013). This drought caused heavy economic, social, and environmental impacts and ranked as one of the worst in the country since the 1930s Dust Bowl (Grigg, 2014).

As temperature has changed over the last century, soil moisture has varied as well. The trends in daily site-based T-max and T-min temperatures can be collectively analyzed in a combination with diurnal temperature range (DTR) (T-max minus T-min).

With the available data, DTR has decreased worldwide during the last 4-5 decades (Dai et al., 1999). As technology has advanced, coupled with new atmospheric measurement technologies and new datasets, it has become easier to investigate the decreasing DTR trend. More importantly, scientists have been able to identify possible causes of this trend. Linking the long-term decrease in DTR to drought might give some insight to

5 possible effects of global climate change. This chapter reviews some of the key research on DTR and soil moisture variability. With the use of this scientific knowledge, this thesis then delves into the possible relationship between DTR and soil moisture in the recent global climate change. The research review will mainly assess findings for the

United States although DTR and soil moisture variabilities are global occurrences.

Since DTR is solely related to fluctuations of T-min and T-max, soil moisture and global climate change can be linked to changes in daily temperature maxima and nighttime minima. Karl et al. (1993) introduced the notion of decreasing DTR linked to increased cloudiness and low cloud cover across the United States resulting in increased daily T-min. Karl blamed a possible combination of observed global warming, increasing aerosols, and fluctuations in natural climate variability as indirect causes of the increase in cloudiness (Karl et al., 1993).

Dai et al. (1999) evaluated the effect of wind direction, clouds, water vapor, soil moisture, and surface wind speed on DTR during summer and autumn seasons. Each variable was treated separately in the analysis. They used site-averaged data created by

Betts and Ball (1998) based on higher-resolution data collected during the First

International Land Surface Climatology Project (FIFE) in the late 1980s in Kansas. The

FIFE dataset contained 30-min averaged surface air temperature, humidity, winds, precipitation, total cloud cover, latent and sensible heat fluxes, solar and longwave radiative fluxes, and daily soil moisture content (Dai et al., 1999). The FIFE dataset lacked a comprehensive look at cloud cover fluctuations so cloud data was obtained from

6 a nearby weather station from the National Center for Atmospheric Research data archives.

Dai et al. looked at the possible effects of wind direction on DTR. They narrowed the FIFE dataset to days of low cloud cover (cc<25%). They monitored the impact warm and cold air masses had on surface air temperatures and DTR by using clear days that had predominately southerly and northerly component winds. It was determined that wind direction alone was not a significant factor in DTR (Dai et al., 1999) although they expressed concern that other geographic and topographic locations in the United States might not necessarily exhibit the same result. Although wind direction can significantly affect daily mean surface air temperature, DTR is more dependent on everyday unbalanced local forcings such as soil moisture and sensible and latent heat fluxes (Dai et al., 1999). This will be discussed below.

The effect of varying cloud cover on DTR was investigated using data from non- precipitating and relatively dry days. These data were separated into clear and cloudy days (cc<12.5% and cc>62.5%). Cloud base height was also obtained from station reports to see if cloud type plays a role in affecting DTR. Cloud coverage was found to dampen

DTR by reducing the daytime T-max at the surface (Dai et al., 1999), presumably through a reduction in surface solar radiation. The width of the cloud base, thickness, and cloud height are all attributes that can alter DTR. Clouds with low bases are found to be the most efficient at reducing daily T-max and in turn, DTR (Dai et al., 1999). Cloud cover can reflect incoming radiation which dampens T-max and can also transmit downward longwave radiation which can increase daily T-min (Dai et al., 1999).

7

Atmospheric water vapor was evaluated on clear days that were separated into dry/low humidity and dry/high humidity days. They used the large differences in downward longwave radiation between the low and high humidity days as the measure of water vapor on DTR. Their correlation analysis showed that nighttime T-min was strongly related with surface humidity (Dai et al. 1999). According to Fung et al. (1984) and Zhang et al. (1995), surface downward longwave radiation is most sensitive to water vapor and temperature in the lower atmosphere. This is because there is ample moisture below clouds to help decouple temperature at the surface and the downward longwave radiation from clouds. Composite analysis determined that enhanced downward longwave radiation was linked to high humidity days (Dai et al., 1999). The enhancement of the downward longwave led to increases in T-max and T-min. Their study showed that high humidity days were associated with a small decrease in DTR.

Soil moisture can play a role in temperature near earth’s surface due to evaporation and sensible heat loss, as well as through fluxes of longwave radiation. Dai’s data were split into two categories; clear days with low soil moisture versus clear days with high soil moisture. Correlation analysis showed that soil moisture was negatively associated with DTR once other variables were removed (Dai et al., 1999). The higher the soil moisture the smaller the DTR since daily T-max is limited due to evaporative cooling occurring at the surface. Water vapor plays a role in T-min as a greenhouse gas absorbing infrared radiation emitted by the earth’s surface. Increased soil moisture is associated with precipitation and cloud cover, which also have DTR impacts.

8

To evaluate how surface wind speeds may play a role in modulating DTR, Dai et al. examined clear days with calm winds versus windy days (> 4.6 m/s). They tried to determine if wind speed plays a role in sensible and latent heat flux. After analyzing their correlations, wind speed showed no significant relationship to DTR at the FIFE site.

Sensible and latent heat fluxes showed no significant relationship to DTR between calm and windy days. Turbulent mixing at the surface on calm days is not a limiting factor for sensible and latent heat flux (Dai et al., 1999).

In summary, Dai et al. (1999) provided insight into the decline of DTR over the last four to five decades in the United States. The diurnal cycle of temperature at the surface is mostly driven by radiative fluxes and moisture (Dai et al., 1999). They investigated several variables that could modulate radiation at the surface and found that cloud cover, precipitation, and soil moisture were the biggest contributors to decreasing

DTR. Surface solar heating affects daily T-max while nighttime T-min is controlled by the greenhouse effect of atmospheric water vapor near the surface. Cloud cover, precipitation, and soil moisture were accountable for approximately 50% in the reduction of diurnal temperatures when compared to clear sky days (Dai et al., 1999). Clouds and soil moisture are able to decrease DTR by reducing T-max due to reflection of solar radiation and evaporative cooling. Precipitation from clouds affect temperature maximum by slowing the rate of diurnal warming. Greenhouse gases may be liable for increasing both the minimum and maximum temperatures, but decreasing DTR is mostly linked to asymmetric daily forcings (Dai et al., 1999). The authors suggest that analyzing land use

9 changes to see if DTR has changed due to surface evapotranspiration may be useful and that urbanization may play a role in changing DTR.

Lauritsen and Rogers (2012) sought to confirm the findings of Dai et al. (1999) using a dataset covering the entire United States and showed regional details of the long- term downward trend of DTR. They investigated the DTR variability over several regions in relation to moisture variables and atmospheric teleconnections. They found that the annual increase in T-min has been exceeding the annual increase of T-max across most of the United States, especially since 1950.

Lauritsen and Rogers (2012) used high resolution gridded data from the Climate

Research Unit (CRU) 2.1 dataset which shows land surface climate parameters dating from 1901 to 2002 covering the United States and parts of Canada and Mexico. Monthly

T-max and T-min, DTR, and precipitation were used and converted to annual values.

Version 3 of the Global Historical Climatology Network was used in conjunction with the CRU 2.1 dataset for comparison purposes. Cloud cover data were assembled from various weather stations across the United States dating from 1891 to 1987. Cloud cover data from 1987 were obtained from the NCDC which produced hourly data from 1988 to

1996. To ensure a complete set of could cover data to 2002, sunshine reports from the

CRU 2.1 dataset were used to serve as another measure of cloud cover.

To obtain estimates of soil moisture, Lauritsen and Rogers used the Palmer

Drought Severity Index (PDSI). The PDSI ranks from extremely dry (negative values) to extremely moist (positive) on its scale of measurement. Nine different teleconnection indices represented the atmospheric circulation including the North Atlantic Oscillation,

10

Arctic Oscillation, North Pacific Index, Southern Oscillation Index, Atlantic Multi- decadal Oscillation, Pacific Decadal Oscillation, Niño-3.4 SST Index, the Cold Tongue

SST Index, and finally the Tropical Pacific SST Index.

Lauritsen and Rogers (2012) regionalized DTR variability and identified pinpoint predictor variables within each region that explain DTR fluctuations. Regionalization was accomplished using principal component analysis to maximize the variance explained and a varimax rotation strategy to regionalize all of the patterns of DTR variance across the United States. Five rotated principal component patterns were identified across the

United States with centers on a narrow winding area of the Midwest and the northeastern states (RPC1), the southwestern United States (RPC2), the south-

(RPC3), extending from eastern Texas through and north to Tennessee, the

Great Plains (RPC4) and the western United States. (RPC5).

Within each of the five RPC regions, mean annual values of T-max, T-min, and

DTR were found along with average values of cloud cover, soil moisture, and precipitation. A stepwise multiple linear regression was then applied on the variables to identify the best combinations of predictors that explain the most variance of the temperature variables (T-min, T-max, and DTR). All five regions found by the RPCA, excluding the southwestern United States, showed highly significant downward DTR trends between -1.2°C and -1.9°C over the last century (Lauritsen and Rogers, 2012). The northeastern United States exhibited a statistically significant upward trend in Tmin and the DTR fell below 12°C after 1965. Increased cloud cover, precipitation, and soil moisture all influence this DTR decline. Cloud cover alone explained most of the DTR

11 variance in this region due to the upward 20th century trend (Lauritsen and Rogers, 2012).

The stepwise multiple linear regression indicated that cloud cover explained 63.2% of the

DTR variance while T-min variability was 21% due to cloud cover. Less than 5% of the variability of T-max was explained by cloud cover. Soil moisture and precipitation explained very little of the DTR variance in this region.

The southwestern United States did not show a DTR decline over the 20th century because upward trends occur in both Tmin and T-max leading to an unchanged DTR.

Two specific periods had large DTR variance. From 1946 to 1956, low soil moisture and precipitation led to very dry conditions in the southwestern United States. T-max was elevated which actually led to an increase in DTR (Lauritsen and Rogers, 2012). In contrast, moist conditions occurred from 1982 through 1986 producing low T-max values and decreased DTR. Throughout the entire century, 39.1% of the DTR variance is due to soil moisture and precipitation in the Southwest. The Atlantic Multi-decadal Oscillation produced 55% of the T-max variance during the century. Lauritsen and Rogers summarized that precipitation is accountable for reducing the daily T-max while simultaneously increasing T-min, resulting in a large impact on DTR. Soil moisture inhibited both T-max and Tmin and resulted in a reduced DTR impact.

The south-central United States had higher values of T-max until 1957 that produced higher DTR (Lauritsen and Rogers, 2012). Increased T-max was especially associated with a 1950s drought along with low cloud cover, precipitation, and soil moisture during this dry period. Following 1957, T-max and DTR took a dramatic

12 downturn, occurring along with an increase in cloud cover along with precipitation and soil moisture. These variables accounted for 66.5% of the DTR variance.

The Northern Plains had a significant decrease in annual DTR through the 20th century (Lauritsen and Rogers, 2012). This is in large part due to the sizeable upward trend in T-min coupled with only a small T-max increase. The 1930s drought reduced soil moisture increasing the T-max and also DTR. 60.2% of the variance of DTR in this region was explained by cloud cover, precipitation, and soil moisture.

An area in the western United States had a DTR peak in the 1930s followed by a steady decline. Low soil moisture and precipitation led to the peak in DTR producing high values of T-max. However, daily T-min has steadily increased, explaining the DTR decrease following the 1930s. Soil moisture and precipitation explained most of the T- max in this region while cloud cover explained 22% of the variance in T-min (Lauritsen and Rogers 2012). A combined effect of all of the moisture variables (cloud cover, soil moisture, and precipitation) explained 57% of the DTR variance in the western United

States.

Lauritsen and Rogers (2012) produced important United States regional information on DTR occurring throughout the 20th century. All 5 of the DTR regions exhibited an increase in T-min throughout the century while T-max varied regionally.

Prolonged sub-periods of wet and dry conditions played a large role in T-max and DTR variance. Variations in the moisture parameters cloud cover, soil moisture, and precipitation were all additional causes of DTR changes. The DTR decline was most noticeable after 1950 when cloud cover started an upward trend. By regionalizing the

13

United States, it was shown that different areas were succumbing to different combinations of moisture parameters and teleconnections to produce the falling DTR trend over the last half of the 20th century. Their research helped to further pin point localized causes of changing DTR.

14

Chapter 3: Data and Methodology

3.1. Data

Air temperature data are obtained from the Global Historical Climatology

Network Monthly (GHCNM) dataset. These data contain monthly average T-max and T- min values for many United States weather stations within each state. This expansive data set covers the years from 2012 to years in the 1890s and early 1900s, depending on the station. It was intended to start the analysis using station data in 1895. Most stations have some missing values because either the station started reporting after 1895 or the station missed reports for months or years for other reasons. If only one year of data are missing, the adjacent summer values are used to form an average for the missing year. In all other cases, a station can be defaulted to a missing value and will not be further incorporated in the study. This use of station data records differs from the gridded CRU2 data, used by

Lauritsen and Rogers (2012), which only extended to 2002.

A unique dataset of summer air temperatures and precipitation quantities has been gathered from the NCDC and organized for a case study of the summer of 2012 along with climatological averages. Temperatures for this dataset are comprised of either first order weather stations reporting on individual office websites,

Automated Surface Observing System (ASOS) stations which are part of the United

15

States Historical Climatology Network (USHCN) version 2.5, or GHCN data

(www.ncdc.gov/cag/). These data are available for the , statewide, Climate Divisions, Climate Regions, National Weather Service Regions, and

Agricultural Belts that come from the United States Climate Divisional Database, which extend from 1895 to the present. They are used in an analysis of historical drought- summer comparisons to the drought of 2012. Data for May, June, July, and August are averaged to make summer mean T-max and T-min values. These data will also be used in determining DTR by taking the T-max value at a location minus the T-min value at that same location for each respective month.

Palmer Drought Severity Index, or PDSI, data are used and are available for

1895-2012. These data are collected from the National Climatic Data Center of the

National Oceanic and Atmospheric Administration (NOAA)

(http://www.ncdc.noaa.gov/sotc/drought). The PDSI is a numerical index value that measures water content in a soil layer based on precipitation and air temperature data for the current and past months. A mean summer PDSI value will be obtained for each summer for each of the 344 United States climate divisions, following the procedures described for temperature. The climate division PDSI data are used in the rotated principal component analysis to obtain unique regional PDSI variability patterns.

Precipitation data are also from the climate division dataset and these data are averaged over the PDSI-regions to help further identify moist and dry summers.

16

3.2. Methods

Spatial and temporal patterns of United States soil moisture data are created using a Principal Component Analysis (PCA) on the PDSI dataset correlation matrix. PCA reduces the dimensions in data, leaving only its basic spatial component modes of variability. The principal components are the key modes in explaining the majority of the dataset variance. The eigenvectors initially obtained by the PCA maximize the variance explained in each component of the analysis such that all eigenvectors following the first are orthogonal. Due to this orthogonality constraint, second and higher eigenvectors do not show any unique patterns of spatial variability that may be present in the dataset. To resolve this issue, the eigenvectors are subjected to a rotation procedure that transforms them into a nonorthogonal linear structure. This leads to compact patterns that regionalize the variance in the dataset (Lauritsen and Rogers, 2012) although the patterns may overlap spatially and are not orthogonal.

The Rotated Principal Component Analysis (RPCA) identifies 10 United States regions (components) having unique patterns of PDSI spatial and temporal variability.

The regional centers defined by the RPCA are identified by using the component numerical loadings, which are similar to correlation components with values of L between -1 and +1. Each component will have a unique clustered set of grid points over the United States where the loadings reach critical large values. Loadings are similar to correlation coefficients and in this case L > 0.71 is chosen since L² ≈ 0.50 (50%) indicating that no other (among 9 remaining) RPCs would explain as much of the climate division’s variance as did the one having L > 0.71. Of the 10 significant regions

17 identified by the analysis, 4 of them were considered key regions of interest for further analysis in this thesis.

The wettest and driest summers were then identified from the time series scores of the rotated principal components (RPC) for the 4 selected regions. The time series scores for each RPC generally have values between -3 and +3, much like standard deviations, and remain orthogonal to each other as part of the rotation procedure. The scores for each region are highly correlated to the original PDSI values occurring at the grid points in those regions. The mean DTR, T-max and T-min values (along with standard deviations) are obtained for these wettest and driest summers using the GHCNM station data lying within each of those regions. The temperature/DTR means for the two sets of cases, wet versus dry, are compared and tested for statistical significance using a two-tailed t-test.

Concluding is work is a case study of 2012 temperature and DTR conditions relative to the composite means of historic past.

18

Chapter 4: Results

4.1. Rotated Principal Components of summer PDSI

Summer soil moisture indices (PDSI) from all United States climate divisions for

1895-2012 were subjected to a rotated principal component analysis (RPCA) that produced 10 regions (components) exhibiting relatively unique soil moisture variability.

Four of these regions will be discussed in detail below, focusing mostly on the regions east of the that explained the greatest amounts of dataset variance. The spatial component of the RPCA is the loadings, which numerically express how much of the PDSI variability in each climate division is represented in a particular rotated principal component (RPC). The variability of each climate division was considered to be part of a particular individual RPC if it had a loading L > 0.71. The time series of the

RPC scores are shown below. The process of rotation of principal components orthogonalizes all of the RPC time series but they are spatially uncorrelated among themselves.

The RPCA produces 10 regions exhibiting unique moisture variability from 1895-

2012, each uncorrelated to the variability in the other components. These are shown in

Fig. 4.1. The cumulative explained variance for the 10 patterns is 67.2%. The component regions evaluated further here are RPCs 1, 2, 3, and 5 with component 4 for the Pacific

Northwest United States being left out of further analysis along with lesser components 19

6-10 that explain much less dataset variance. RPC1 centers on , ,

Kentucky, and Ohio (Fig. 4.1). RPC2, which isolated the upper Midwest, contains ,

Minnesota, Nebraska, , , and . Alabama, Georgia,

North Carolina, and South Carolina establish the southeastern region known as RPC3.

Finally, RPC5 includes Kansas, Oklahoma, and Texas and is generally called the

Southern Plains. The 4 RPCs explained 20.4%, 11.6%, 9.3%, and 4.7% of the total dataset variance, respectively. The number of loadings (L) >0.71 was relatively limited in each RPC and as such, Fig. 4.1 shows the climate divisions where L is as low as L= 0.6.

Fig. 4.1: United States Climate Divisions with the highest RPC loadings on each of the first 5 summer PDSI components. The map symbols represent the core of climate divisions with the highest loadings (> 0.6) for a particular RPC. Each RPC is represented by a different symbol or shading to tell them apart. Any climate division where loadings failed to reach 0.6 on any RPC are represented as small dots. Climate divisions represented with a cross (+) had loadings in excess of 0.6 but occurring on a lower ranked RPCs (6 through 10).

20

Fig. 4.1 illustrates that regional centers of unique spatial and temporal PDSI variability occur across many parts of the United States. This area contains valuable farm and graze land that are very important to the well-being of the economy and health of the

United States. While the areas would be prone to wet summers most are also subject to drought conditions. Fig. 4.1 shows the spatial locations of the central core of PDSI centers while other key features in each region’s long-term time series represent the moist and dry events occurring in history. The remainder of this section focuses on time variability associated with RPCs 1, 2, 3, and 5 representing respectively the Ohio River

Valley, the upper Midwest, the southeastern United States, and the Southern Plains.

Figs. 4.2 through 4.5 represent time series of the RPC scores based on the PDSI for the summer months of June, July, and August over the last century (1895-2012) at the

4 regional centers. The RPC scores for each region are highly correlated to the regional mean PDSI but their scores are orthogonal with every other region. PDSI ranges on a numeric scale from 4.0 (extremely moist) to -4.0 (extremely dry) but the RPC time series scores range from -3 (dry) to +3 (moist), effectively representing standard deviations.

Values greater than 0 are considered moist conditions while values less than 0 are considered dry.

Fig. 4.2 represents the time series of PDSI through RPC1 for the Ohio River

Valley. According to the time series, RPC1 experienced drier conditions prior to 1955.

Some of the drier summers included 1901, 1934, 1936, 1941, 1953, and 1954 while wet summers occurred in 1927 and 1950. Following 1955, higher soil moisture conditions were prevalent. The driest summers included 1991 and 2012 but they are apparently not

21 as dry as the just mentioned pre-1955 summers. This variability is represented around the least-squares linear trend line associated with the time series that indicates that PDSI on average is on the rise in this region.

Fig. 4.2: Time series of RPC1 scores representing the orthogonalized PDSI values for the Ohio River Valley.

Fig. 4.3, for RPC2 contains the upper Midwest and eastern Northern Plains states, and has a similar looking PDSI trend in its scores as RPC1. Dry conditions dominated the first half of the 20th century followed by moist conditions during the second half into the beginning of the 21st century. From 1910-1941, only 6 summers had moist scores above

22 zero with only one such summer during the dry summers of 1929-1941. The summers of

1931 and 1934 are among the driest, along with 1988 and 1989. The summer with greatest soil moisture is 1993, when persistent heavy rains in the upper Mississippi and

Missouri river basins contributed to widespread severe flooding farther down the river basins around eastern Missouri (Wahl et al., 1993; Changnon, 1996; Eash, 1997).

Fig. 4.3: Time series of RPC2 scores representing the orthogonalized PDSI values for the Upper Midwest.

The time series of RPC3 represents PDSI variations in the southeastern United

States (Fig. 4.4). RPC3 also shows a trend towards increasingly moist soil conditions

23 through the 20th century although the trend is not as strong as in RPCs 1 and 2. A notable feature is the extended series of moist summers from 1957 through 1976 as well as the recent series of dry summers after 2005. It was also persistently dry from 1950-1956. The region was not particularly affected by excessively dry conditions in the 1930s and most of the driest summers do not correspond to those of RPC1 or 2, including 1988 and 2012.

Overall, it would appear that dry summers have occurred more frequently in recent decades than occurred in RPC1 and 2. This reduces the size of the upward trend despite many moist summers with large positive scores.

Fig. 4.4: Time series of RPC3 scores representing the orthogonalized PDSI values for the southeastern United States.

24

Fig. 4.5 represents RPC5 and the southern Plains of the United States. The PDSI scores in this region were the only one of the four to show a trend towards drier soil conditions although the trend is negligible. The 1930s drought was severe from 1933-

1940 but the southern Plains were particularly dry in the 1950s drought from 1952-1956

(Fig. 4.5). These droughts are followed by some of the wettest years in the record. No major drought occurs again until 1997 and this region is particularly dry in 2011 and

2012. The wettest summers are 1987 and 2007.

Fig. 4.5: Time series of RPC5 scores representing the orthogonalized PDSI values for the Southern Plains.

25

4.2. Regional T-max, T-min, and DTR, 1895-2012

A major goal of this work is to analyze how DTR has changed throughout the last century in the United States by focusing exclusively on regions that have the greatest variability in soil moisture. Previous studies such as Dai et al. (1999) and Lauritsen and

Rogers (2012) have shown decreasing DTR with time across the United States since

1950. By only using summer temperature data, this phenomenon is investigated and compared for each of the RPC regions. This section will discuss how regional average T- max and T-min have fluctuated throughout the last century based on available GHCNM station data. Time series show regional temperature changes are linked to the decreasing

DTR trend. Summer averaged T-max and T-min time series are presented while standard deviations about the regional mean among the stations are also plotted. This indicates the degree of variation in the temperatures and DTR among the stations in each region in each summer.

Figs. 4.6-4.8 depict the times series of mean T-max, T-min, and DTR for stations of the Ohio River Valley (RPC1) from 1895 through 2012. Fig. 4.6 shows the regional average T-max over the last century. Regional T-max has large fluctuations but overall exhibits a downward trend over the span of the century. According to the T-max trend line, mean T-max decreases about 0.5°C over the period of record. There are a few years that stand out on this T-max time series. First, the year 1915 has a regional T-max of approximately 26°C, the lowest for a summer over the entire span of the regional dataset.

1936 also stands out on the time series as the warmest summer regional T-max for this region. This year was very dry summer mostly associated for a heat wave and drought

26 that took place in the Midwest. Following these two dates, regional average T-max became progressively lower based on the trend line in Fig. 4.6. A span of consistently unusually low T-max summers occurs from about 1960-1976 followed by more interannual variability after 1976 and higher T-max.

The standard deviations of the summer T-max values (Fig. 4.6) among the stations of the Ohio River Valley is relatively large in the early 20th century but becomes much lower, especially following the mid-1980s. Most stations chosen in this analysis had temperature data that started as close as possible to 1895, while a few stations started data collection following that date. Overall missing data tended to occur early in the 20th century as observation stations were developing. These factors may partly be affecting the temporal variation in the standard deviations throughout Fig. 4.6.

Figure 4.7 illustrates the progression of regional T-min averages in the Ohio River

Valley. The time series clearly shows that regional T-min for this area has increased over the last century. This positive trend line increases about 0.7 °C through the period of record. The summer of 1926 exhibited the lowest T-min than any other summer in the record, just as its corresponding T-max value (Fig. 4.7) had been very low. Temperatures decreased to an average of under 14°C. Comparing Fig. 4.2, 1926 was an average year based on precipitation in this region according to its PDSI score. The highest average T- min in this region occurred more recently in 2010 when average T-min were over 18°C.

T-min were consistently relatively high into the 1930s but starting around 1985, the T- min values begin to rival and even exceed those of the 1930s. T-min were consistently low in the 1960s. The summer of 2010 was dry based on PDSI (Fig. 4.2), but not very

27

88 28 28

Fig. 4.6: Regional Averaged T-max for RPC1.

28

extreme. The standard deviations around the regional mean summer T-min is much more evenly and randomly distributed in Fig. 4.7 than occurs for T-max in Fig. 4.6. The standard deviations about the station-based regional T-min values are not as consistently high in the early decades of the 20th century, as it had been for T-max. The T-min standard deviations show little trend over the period of record.

The DTR time series for region RPC1 (Fig. 4.8) shows that average summer DTR is on the decline in the Ohio River Valley over the last century. Despite the century-long downward trend (Fig. 4.8), DTR increased from 1895 onward into the peak DTR summers of the 1930s. The peak DTR in the 1930s corresponds to the very dry and warm

(high T-max) summers of this decade. Four summers, 1930, 1936, 1988, and 2012 appear as years where DTR was much larger than others (Fig. 4.8), with the latter two summers having somewhat lower DTR than 1930 and 1936, apparently in keeping with the decreasing long-term trend line. The very dry summer of 1936 (Fig. 4.2) exhibited the largest DTR overall. The regional average DTR was slightly over 15°C, with the average

DTR over the entire period of record being just under 12.5°C. Overall, Ohio River Valley

DTR is much lower than the rest of the period of record after 1965, with the exception of the drought summers of 1988 and 2012. The standard deviation of the DTR among the regional stations also steadily declines through time, seemingly mirroring the downtrend noted in the T-max time series (Fig. 4.6). It appears to become abruptly lower after 1952.

Figs. 4.9-4.11 show the time series of T-max, T-min, and DTR for the upper

Midwest and portions of the eastern Northern Plains (RPC2). Based on similarities between Figures 4.2 and 4.3, one would expect the variability in RPC2 to broadly

29

30

Fig. 4.7: Regional Averaged T-min for RPC1.

30

31

Fig. 4.8: Ohio River Valley DTR.

31 resemble that of RPC1.

The upper Midwest has T-max variability around a steady non-trending regional average of approximately 26.9°C for the period of record (Fig. 4.9) as opposed to the downtrend in Fig. 4.6. The period from 1930-1943 exhibits unusually high T-max values.

The summers of 1915 and 1992 exhibited lower temperatures than the average. The summer of 1915 was one of the wettest in Figure 4.3 while 1992 was the summer after the eruption of Mt. Pinatubo when the United States overall had its third coldest summer in 77 years (geography.about.com/od/globalproblemsandissues/a/pinatubo.htm). RPC2 also had warmer than average temperatures in 1936 and 1988. These years produced almost identical regional average high temperatures. Both summers were among some of the driest summers on record in this region (Fig. 4.3). The trend line established in Figure 4.9 depicts a very flat but slight down turn representing a decreasing regional T-max average.

The standard deviation about the regional mean T-max values among stations is higher early in the 20th century than in the second half of that century.

Regional average T-min for RPC2 follows a clear warming trend throughout the last century continuing into present day (Fig. 4.10). The trend starts around 13°C in 1895 and has increased to almost 14.25°C in 2012. The warming of regionally averaged T-min was also present in RPC1 (Fig. 4.7) but it was not as large a trend as in RPC2. Within the last 30 years, especially after 1982, summer T-min has on several occasions been among the highest on record. At the same time however, the summers of 1985, 1992, 2004, and

2009 are among the lowest T-min. This mimics the behavior of PDSI in Fig. 4.3 showing that summer soil moisture levels are on the increase and could help explain why T-min is

32

33

Fig. 4.9: Regional Averaged T-max for RPC2.

33 increasing with time. Cloud cover and moisture have been found to inhibit cooling.

Standard deviations about the mean T-min among the stations of the region are consistently higher early in the period of record until the early 1950s, after which many summers have lower standard deviations, especially after the early 1980s.

Fig. 4.11 depicts the extent of DTR in the upper Midwest and eastern Northern

Plains for the last century. When you combine the long-term steady regional T-max and notable warming of T-min in this region, the DTR exhibits a decreasing trend (Fig. 4.11) that, at about 1.1°C, is almost as large as the T-min upward trend. There is a clear decline in DTR from the beginning of the last century to present day albeit in a somewhat step- like fashion. For example, DTR appears highest from about 1910-1935, then it declines in value somewhat until about 1990, followed by some of the lowest DTR values following

1990. The downward step after 1935 is associated with a large decrease in T-max (Fig.

4.7) while coinciding T-min (Fig. 4.8) does not decrease. The stepdown in DTR following 1989 occurs as T-max remains relatively low (Fig. 4.7) but T-min has some of its highest values (Fig. 4.8).

Summer 2012 had a relatively large DTR amidst a preceding 20+ year tendency for low summer DTRs. Most of the extreme DTR summers depicted in Fig. 4.11 are also summers that are particularly wet or dry. The summers of 1910, 1933, 1936, and 1988 all rank within the driest summers on record also shown in Fig. 4.3. The dry summer of 1976 does not fall within that category. Summers with wet conditions such as 1951, 1993, and

2010 also were years where DTR also exhibited the smallest range within the last century. The standard deviations appear to mirror the same low frequency behavior with

34

35

Fig. 4.10: Regional Averaged T-min for RPC2.

35

36

Fig. 4.11: Upper Midwest and eastern Northern Plains DTR.

36 high deviations early in the record and some of the lowest after 1990 (Fig. 4.11).

The southeastern United States, RPC3, shows some similarities to temperature and DTR patterns in RPC1 and 2, but it has its uniquely different characteristics (Figs.

4.12-4.14). Regional averaged T-max exhibits an overall non-existent trend amidst oscillating summer temperatures (Fig. 4.12). The highest summer T-max values occur in

1952, 1954, 2010, and 2011. The summer of 2011 essentially represents the highest T- max, breaking records from the 1950s. Summer T-max in the Southeast is particularly below normal from 1960-1976. From that period onward, T-max is on a warming trend throughout the second half of the century. Standard deviations about the mean T-max values are also relatively low through the middle 20th century indicating the consistency of the low temperatures among the regional GHCNM stations. Fig. 4.4, shows a period in the 1960s that carried over into the 1970s where soil conditions were very moist. These summers ranked within the wettest on record according to averaged PDSI values. This can explain the cooled regional averaged T-max for RPC3 as cloud cover or frequent precipitation can suppress T-max values (Lauristen and Rogers, 2012). Rogers (2012) discusses how low summer temperatures in the Southeast from 1961-1976 are due to a combination of some cloudy summers and some sunny summers with dominant polar air masses. Following these years, drier conditions developed producing warmer T-max. The trend line slightly picks up this warming trend with very small positive slope (Fig. 4.12).

T-min time series for the southeastern United States (Fig. 4.13) summer T-min during the last century had some similar characteristics to the T-max variability in Fig.

4.12. RPC3 T-min leading into the 21st century also shows a warming trend. The T-min

37

38

Fig. 4.12: Regional Averaged T-max for RPC3.

38 time series mimics the wet summer characteristics in the 1960s and 1970s with very low minimum temperatures. Starting in 1967, there is a clear warming trend in T-min. The trend line after 1967 would likely be more pronounced than that shown for the entire period of record. The summer of 2010 in RPC3 really stands out on the time series as having the warmest regionally averaged T-min on record, breaking the record set earlier in 2005. Looking at the summer of 2010 of RPC3 in Fig. 4.4, 2010 was dry, but doesn’t register as one of the driest on record. The summer T-min standard deviation about the mean based on the stations representing the southeast shows a relatively unique gradual increase through the period of record.

Because the warming trend in T-min of RPC3 is larger than the near-zero trend in

T-max, DTR will still show a decreasing trend. This is shown in Fig. 4.14 based on the overall regional DTR trend, the mean southeastern DTR goes from 11.5°C in 1895 to about 11.24°C by the beginning of the 21st century. The DTR from 1960-1976 is not particularly notable, these summers had low values in both T-max and T-min.

Particularly low DTR summer values only occur between 1989 and 2005. Large DTR occurs in 1925-1932 and into the mid-1950s. Comparing Figure 4.14 with 4.4 reveals some answers to the behavior of DTR over the time span of data in RPC3. Drier summers in the beginning of the 20th century (Fig. 4.4) (e.g. 1911, 1914, and 1925) correspond to summers of larger DTR (Fig. 4.14) peaking in the summer of 1955 which was also a dry summer. The wet period in the 1960s and 1970s start the decline in DTR (Fig. 4.14). The smallest DTRs occurred in the more recent summers of 2003 and 2005 where both summers rank as the first and second wettest summers in RPC3 (Fig. 4.4). The latter half

39

40

Fig. 4.13: Regional Averaged T-min for RPC3.

40

41

Fig. 4.14: Southeastern United States DTR.

41 of the 20th century leading into the 21st century seem to be influencing the decline in DTR more than the dry summers that occurred in the beginning of the 20th century.

Finally, Figs. 4.15-4.17 depict the T-max, T-min, and DTR time series for the

Southern Plains (RPC5). Fig. 4.15 shows that summer T-max exhibits a warming trend over the 20th century into the 21st century. This is the only region of the 4 analyzed here that has a clear upward T-max trend. The trend magnitude is roughly 0.6°C over the period of record. The summer of 2011 had the highest T-max over the period of record, exceeding those of 1934 and 1936. The summers exhibiting the highest averaged T-max occurred in the years of 1934 and 2011 and rank as dry summers in Fig. 4.5. This would explain the very warm temperatures during these summers. During dry conditions, cloud cover is usually at a minimum. The driest summer in RPC5 occurred in 1956 but didn’t produce the warmest summer of the record. T-max during this summer was ~ 2°C lower than that of 1934 and 2011. The cooler summers of 1915 and 1992 can be found on Fig.

4.5 as some of the wettest summers on record. The seasonal standard deviations of T-max among the regional stations is highly variable through the period of record, especially in the first 3 decades.

The time series of regionally averaged T-min in the southern plains (Fig. 4.16) almost identically mimics the behavior of T-max in Fig. 4.15. T-min shows the same warming trend from the beginning to end of the dataset having a magnitude of about

0.7°C, slightly larger than that for T-max. The two warmest summer T-min occur in the same years as the T-max in 1934 and 2011. The summers of 1915, 1920, and 1992 show the lowest regionally average T-min. Figure 4.15 also shows that T-max during these

42

43

Fig. 4.15: Regional Averaged T-max of RPC5.

43

44

Fig. 4.16: Regional Averaged T-min for RPC5.

44 summers were abnormally cool as well. Similar to T-min exhibited in the other RPCs

(Figs. 4.7, 4.10, and 4.13), a long-term trend in T-min warming is shown (Fig. 4.16).

Large annual T-min station standard deviations occur early in the record.

The time series of DTR for RPC5 is relatively unique among the 4 regions in that is has a near zero trend (Fig. 4.17). DTR in the Southern Plains does not seem to be changing like the other regions. Even though the trend line does not show much visually, the slope is just slightly negative. This means that DTR is in fact decreasing over the entire 20th century into the beginning of the 21st century for all the 4 regions studied in this analysis supporting other works based on this topic. The largest DTR of RPC5 took place in the summer of 1936 and recent high DTR values occur in 2011 and 2012.

Although 1936 was not one of the driest summers on record for this region, PDSI values indicate that this summer experienced dry soil conditions. The smallest DTR occurred in

1992 which has been established as one of the wettest and coldest summers on record

(Fig. 4.5). The wettest and driest PDSI values for each region will be discussed in the next section.

45

46

Fig. 4.17: Southern Plains DTR.

46

4.3. Wettest vs Driest summers: Regional T-max, T-min, DTR, and PDSI

Previous studies have evaluated how diurnal temperature range (DTR) fluctuates for given dry or moist soil conditions (Karl et al., 1986; Dai et al., 1997; Dai et al., 1999).

Lower DTR is linked to years of higher precipitation compared to years with lower PDSI values when DTR is larger. In this section, T-max, T-min, and DTR are calculated and compared among the 12 wet and dry summer cases for each region. The 12 wettest and driest summers are determined based on the actual summer mean soil moisture (PDSI) index value for the group of climate divisions forming the region. The highest or lowest

PDSI summers generally correspond to the highest and lowest PDSI scores shown in

Section 4.1 (Figs. 4.2-4.5) although there are some differences. The selection of the 12 most moist and dry summers did not include any prior to 1910 as fewer stations were available in that era that reported temperature and were limit the reliability of the analysis.

Even numbered figures between Figs. 4.18-4.25 show the 12 wettest summers versus the 12 driest summers in each respective RPCA region. Wet versus dry was determined by summer soil moisture PDSI values. Odd numbered figures portraying

PDSI show wet summers with a blue bar while dry summers are indicated by an orange bar. In these figures, T-max (darker) and T-min (lighter) are shades of green for the 12 wettest summers while DTR is represented by a blue dot. T-max and T-min for the 12 driest summers are represented with red and orange bars while their corresponding DTR is shown by yellow dots.

The Ohio River Valley (RPC1) 12 highest and lowest PDSI summers are

47

48

Fig. 4.18: The 12 summers with the highest (blue) and lowest (red) PDSI values from 1910-2012 for the Ohio River Valley (RPC1).

48 represented by Fig. 4.18. Three quarters of the driest summers occurred prior to the peak dry summer of 1954 (PDSI of ≈ -2.3 (Fig. 4.18)) with a period in the 1930s containing the majority of the lowest soil moisture. The more recent year of 2012 was relatively dry ranking within the dozen driest PDSI scores with a value similar to the dry period that occurred in the 1930s. The majority of the wettest summers occurred in the latter half of the 20th century leading into the more recent year of 2009. The two wettest summers occurred in the first half of the 20th century in the years 1927 and 1950. The summer of

1950 was the wettest summer of record with an average PDSI score for the region topping at ≈ 2.4.

Fig. 4.19 shows the twelve wettest and driest summers within the Ohio River

Valley (RPC1) throughout the last century in terms of T-max, T-min, and DTR.

Averaged summer T-max is clearly higher during summers that were dry, with magnitudes all exceeding T-max occurring in any of the wet summers. The same is true for DTR with magnitudes in excess of 13°C (right-side Y-axis in Fig. 4.19) in every dry summer and values below that level restricted to wet summers. Based on these averages,

T-max was almost 3°C warmer during drier summers than wet, which is over 5°F.

The Ohio River Valley became moister over the period of record (Fig. 4.2) and T- max declined during that period (Fig. 4.6). Fig. 4.19 confirms that dry conditions are linked to hot summer T-max while moist soils are linked to cooler summers. T-min did not show as much variation as between wet and dry years as T-max, but there was still a slight difference. Average T-min was, on average, just over 1°C higher in dry summers compared to wet conditions. DTR displays an increase slightly above 2°C for drier

49 summers compared to wetter years. T-max and DTR values in 1988 and 2012 rival those in magnitude of the early 20th century droughts only exceeded by 1930 and 1936. The results in Fig. 4.19 indicate that both T-max and T-min are higher in dry conditions in the

Ohio River Valley, more so for the former. The higher DTR in dry summers agrees with findings in Dai et al. (1999) and Lauritsen and Rogers (2012), but the higher T-min in dry summers is not expected. Higher T-min may occur because of stronger daytime solar heating in dry summers that helps keep T-min values higher overnight.

The PDSI values of the wettest and driest summers of the upper Midwest are represented by the time-magnitude plot in Fig. 4.20. One similarity RPC2 has with RPC1 is the extremely dry summers that occurred in the 1930s. This means there was a vast area of the United States experiencing drought-like conditions for a long period of time.

As with the Ohio River Valley, most of the driest summers in terms of PDSI occurred in the first half of the 20th century. The second half is characterized by summers with extremely moist soil conditions. The summer of 1993 exhibited soil moisture conditions with an average PDSI value of ≈ 2.6 (Fig. 4.20). This ranks wetter than any summer that occurred in the Ohio River Valley. The driest summer occurred in 1934 with a PDSI value of ≈ 2.7. This is also drier than any PDSI value occurring in any of the driest summers of RPC1. The summers of 1931, 1934, and 1936 registered as extremely dry summers in both the upper Midwest and the Ohio River Valley. The dust bowl is very noticeable in both regions. The summers of the 1930s rank among the driest for each region respectively. The summer of 2012 was particularly dry for both regions. This summer will individually be discussed in Chapter 5.

50

RPC1 Wettest vs Driest Summers: T-max, T-min, DTR 35 15.5

15

30 14.5

14 25 13.5

13 20

51 12.5

Temperature (Celcius) Temperature 12 15

11.5 Diurnal Temperature Range (Celcius) Range Temperature Diurnal

10 11

1914 1925 1927 1930 1931 1934 1936 1941 1950 1953 1954 1958 1964 1973 1974 1981 1988 1990 1993 1996 2004 2008 2009 2012 Year

Tmin-Dry Tmax-Dry Tmin-Wet Tmax-Wet DTR-Dry DTR-Wet

Fig. 4.19: T-max, T-min, and DTR (all °C) during the 12 wettest and 12 driest summers since 1895 in the Ohio River Valley (RPC1). T-max is represented by green (wet) and red (dry) vertical bars; T-min by light green (wet) and orange (dry) and DTR is represented by blue dots (wet) and yellow dots (dry) respectively. 51

52

Fig. 4.20: Same as Fig. 4.18 but for the upper Midwest (RPC2).

52

The upper Midwest (RPC2; Fig. 4.21) has similar characteristics to those discussed above about the Ohio River Valley. As with RPC1, T-max is higher on average during summers that were abnormally dry. In RPC2, T-max is over 2°C warmer in dry summers than in abnormally wet summers. The largest T-max (dark green bars) in some wet summers exceed T-max of some of the dry summers (red bars) in the upper Midwest

(e.g., 1983). T-min does not seem to show much variation between dry and wet summers.

T-min is less than a half degree Celsius warmer during the driest summers relative to the wet cases. This would suggest that the DTR difference between the driest and wettest summers in RPC2 can mostly be linked to the increase in T-max during the driest summers. The T-max and DTR of summer 1988 rivals in magnitude the values for the

1930s and earlier drought summers, but the same does not hold for summer 2012 and some of the other dry summers. Still, DTR in the driest summers exceeds that of any of the 12 wettest summers (Fig. 4.21).

The most extreme summers in the southeastern United States (RPC3) are largely different than those of any other region. Fig. 4.22 shows the driest summer PDSI values are confined to both the beginning and end of the 20th century into the beginning of the

21st century. The driest soil moisture conditions in RPC3 occurred in the summer of 1914 with a regionally averaged PDSI value of ≈-2.1 (Fig. 4.22). This PDSI value is not quite as negative as in the driest summers of RPC1 or RPC2. The majority of the wettest soil conditions occurred in the second half of the 20th century leading into the 21st century much like RPC1 and RPC2. The summer of 2003 ranked as the wettest summer of the dataset registering a regional average PDSI value of ≈2.6 (Fig. 4.22). This is very

53

RPCA2 Wettest vs Driest Summers: T-max, T-min, DTR 35 16

15.5 30 15

25 14.5 14 20 13.5

13 15 12.5

54

10

Temperature (Celcius) Temperature 12

11.5 5

11 (Celcius) Range Temperature Diurnal

0 10.5

1911 1915 1931 1933 1934 1936 1937 1940 1944 1945 1951 1956 1977 1979 1983 1984 1986 1988 1989 1993 1994 2008 2010 2012 Year

Tmin-Dry Tmax-Dry Tmin-Wet Tmax-Wet DTR-Dry DTR-Wet

Fig. 4.21: Same as Fig. 4.19 but for the upper Midwest (RPC2).

54

55

Fig. 4.22: Same as Fig. 4.18 but for the southeastern United States (RPC3). 55 similar to soil moisture indices that occurred ten years prior in the summer of 1993 in

RPC2 (Fig. 4.20). Hurricane Isabel brought torrential rainfall in September to this region that had already been experiencing high soil moistures from severe weather that occurred during the summer. The driest summers in RPCs 1-3 are all very similar in terms of magnitude when an average “abnormal” dry summer PDSI is found. All three regions produce an abnormally dry summer PDSI value of approximately -1.6. However, RPC3 exhibited the largest magnitudes of the wettest summer PDSI scores.

Fig. 4.23 shows the wettest and driest southeastern United States summers throughout the last century in terms of T-max, T-min, and DTR. The driest summers in this this region generally bookend a time period of relatively wet summers during the

1960s and 70s. T-max in this region does not show as much of a difference from dry to wet summers. The average T-max for the driest summers came in at about 1.5°C higher than the wettest. The T-max for all dry summers climbs well over 30°C, while all wet summers have mean T-max closer to 30°C. T-min displays an even smaller amount of variation between wet and dry summers most of which have values near 20°C. During the driest summers, T-min is barely a quarter of a degree Celsius on average warmer than T- min during a wet summer. DTR values in RPC3 do not resemble the same pattern as in

RPCs 1-2, where the DTR in a drier summer was always larger than that during a wet summer. DTR is a little erratic in that some summers with wet conditions had a higher

DTR than in some summers having dry conditions (e.g. 1929, 1975, and 1976). Overall,

DTR exhibits a larger range during dry summers compared to wet with the difference

56

RPC3 Wettest vs Driest Summers: T-max, T-min, DTR 35 13.5

30 13

12.5 25

12 20 11.5 15 11

57

10 Temperature (Celcius) Temperature 10.5

5 10 Diurnal Temperature Range (Celcius) Range Temperature Diurnal

0 9.5

1911 1914 1925 1927 1928 1929 1931 1949 1954 1955 1961 1964 1971 1973 1975 1976 1981 1986 1989 2000 2003 2005 2007 2011 Year

Tmin-Dry Tmax-Dry Tmin-Wet Tmax-Wet DTR-Dry DTR-Wet

Fig. 4.23: Same as Fig. 4.19 but for the southeastern United States (RPC3).

57 being a little over 1°C, a smaller difference than in the previous 2 regions. A lot of this is to blame on T-min. It appears that the southeastern United States does not cool off as much as the upper Midwest or Ohio River Valley. This might be attributable to a greater humidity in the southeastern United States relative to the other regions.

Fig. 4.24 depicts the rankings of the wettest and driest summer PDSI values in the

Southern Plains states defined as RPC5. Almost all of the driest summers occurred within the first half of the 20th century from 1911 to 1956. There are also dry summers in the early 21st century including more recent summers since 2006 (Fig. 4.24). As in the upper

Midwest and Ohio River Valley, the summer of 2012 registers as one of the driest summers on record in this region. The driest summer of the record in the Southern Plains occurred in 1956 with an average score of approximately -2.5 (Fig. 4.24). This ranks as the second driest regionally averaged PDSI value from all 4 regions behind the summer of 1934 in the upper Midwest (Fig. 4.20). The end of the second half of the 20th century experienced the wettest summers in terms of soil moisture (Fig. 4.24). The 1990s alone accounted for a third of the wettest dozen summers recorded in the Southern Plains. The summer with the highest soil moisture occurred in 2007 with a PDSI score of ≈ 2.8. This average is wetter than any other summer PDSI value from any other region.

The Southern Plains states of Kansas, Oklahoma, and Texas (RPC5) experience very hot summers in spite of soil moisture conditions. Fig. 4.25 illustrates that T-max during the driest and wettest years maintain an average over 33°C or 91°F. The majority of the driest summers took place within the first half of the century, while most of the latter half has consisted with more years featuring wetter conditions. T-max in the driest

58

59

Fig. 4.24: Same as Fig. 4.18 but for the Southern Plains (RPC5). 59 summers have higher temperatures than that of T-max during wet summers all cases but one (1955 is not warmer than 1957). What sets the Southern Plains apart from the other regions is that the average difference of over 3°C in T-max between dry and wet summers. Some of the dry summers had very large average T-max values. The summers of 1934 and 2011 produced average T-max in the mid-90s averaged over all weather stations of the three states (Fig. 4.25). Average T-min for these wettest and driest years is about 20°C. Like the other RPC regions, T-min exhibits warmer temperatures during the driest conditions compared to the wettest. T-min in drier summers is a little over 1°C warmer than T-min during the wettest summers. DTR is affected in the same way. DTR is about 1.8°C larger in the driest summers compared to the wettest. There is one wet summer (1973) in which DTR was slightly higher than the DTR during the dry summers.

In this section, averaged T-max and T-min were obtained over the 4 regions for each of the 12 wettest and driest summers. DTR was then calculated by taking the simple difference in T-max and T-min for every summer per region. Based on Figs. 4.19, 4.21,

4.23, and 4.25, T-max seemed to vary the most between wet and dry summers for each region. In some regions such as the Southeast, T-max varied widely amongst individual dry, and even wet summers (Fig. 4.23). It was however generally larger in dry summers compared to wet summers.

T-max exhibits small trends (Figs. 4.6 and 4.15) or no trend at all (Figs. 4.9 and

4.12) over the period of 1895-2012. The results in this section indicate that T-max is much higher in dry summers, most of which occur in the early 20th century. Increases in soil moisture since 1950 in every region but RPC5 have likely been important in

60

RPC5 Wettest vs Driest Summers: T-max, T-min, DTR 40 15.5

35 15

30 14.5

25 14 20 13.5 15

61

13 Temperature (Celcius) Temperature 10

5 12.5 Diurnal Temperature Range (Celcius) Range Temperature Diurnal

0 12

1911 1915 1917 1918 1925 1934 1941 1942 1945 1953 1954 1955 1956 1957 1973 1987 1992 1993 1995 1997 2006 2007 2011 2012 Year

Tmin-Dry Tmax-Dry Tmin-Wet Tmax-Wet DTR-Dry DTR-Wet

Fig. 4.25: Same as Figure 4.19 but for the Southern Plains states (RPC5)

61

suppressing any tendency for increased T-max. T-min for all regions did not seem to be affected as much by soil conditions. By looking at Figs. 4.19, 4.21, 4.23, and 4.25, T-min does not fluctuate very much during a summer when extreme soil moisture conditions

(wet or dry) was classified. T-min however was generally higher in dry summers than in wet soil condition summers. Using annually averaged data, Lauritsen and Rogers (2012; their Fig. 3) showed that while soil moisture has an important regional impact on T-max, its relation to T-min is much weaker. The T-min • PDSI relation is negative in their study, indicating dry conditions are linked to higher T-min.

In terms of DTR for the wet and dry summers of each region, fluctuations in T- max play an overall larger role in determining the extent of DTR. This agrees with results from Lauritsen and Rogers (2012) which, while PDSI explains some of the regional variance in DTR, cloud cover is more important in all regions while in some areas the importance of precipitation variability also exceeds that of soil moisture in influencing

DTR. The overall increase in DTR seems to be more heavily related to the change in T- max when it comes to comparing the wettest versus the driest years for each region.

62

Chapter 5: Case Study: The Drought of 2012

5.1. Overview of the summer drought of 2012

The recent summer of 2012 brought the most widespread severe drought conditions across the central United States in over a half century. Figures 4.19, 4.21 and

4.25 pointed out that the summer of 2012 ranked as one of the top twelve driest years in

RPCs 1 (Ohio River Valley), 2 (Upper Midwest), and 5 (Southern Plains) respectively.

This chapter will specifically focus on how this summer unfolded in these regions as well as other nearby states and how it compares to dry soil conditions that occurred in the past.

This chapter will also take a look at PDSI values obtained over the summer of 2012 and analyze how they changed as the months progressed. The 4 maps in Fig. 5.1 show the month to month progression of the drought of 2012 in terms of soil moisture conditions as measured by the PDSI. As Fig. 5.1 shows, most of the extreme drought conditions were confined to the southwestern and the southeastern United States leading into the summer in May. By the end of June however, the drought intensified in the western states as well as along a swath extending from eastern Texas to southern . The environment in the Midwest United States progressed to drought in a month between

May and June. By the end of June, most of the continental United States was experiencing some extent of drought which continued for the remainder of the summer.

The drought intensified in July (Fig. 5.1) although the expansion of its areal coverage

63 may have been minimal. The drought expanded in the Central Plains and Midwest but was reduced in the southeastern states. In August, the areal extent of drought remained relatively fixed, with some areas improving in the Southeast, and the overall severity of the drought changed little.

Fig. 5.2 illustrates the progression of the summer 2012 drought as presented by the U.S. Drought Monitor (http://www.drought.gov/drought). The U.S. Drought Monitor is jointly produced by the National Drought Mitigation Center at the University of

Nebraska-Lincoln, the United States Department of Agriculture, and NOAA. Produced every Thursday, the map is based on climatic, hydrologic, soil conditions, and also reports generated from observations from individual and agency contributors all over the

United States. Fig. 5.3 better shows the ranking system used by the U.S. Drought Monitor where areas within states can be ranked from having no drought to abnormal dryness

(D0) through D4 which is exceptional drought.

The August map (Fig. 5.2), places the greatest drought severity from the Rockies across the Central Plains into the Midwest and Ohio River Valley. The map sequence also gives more indication of drought intensification from July to August than appears in

Fig. 5.1. Fig. 5.2 also downplays the severity of the drought along the Rocky Mountain

States, relative to the PDSI.

Fig. 5.3 shows the percentage of land mass in the United States associated with each of the U.S. Drought Monitor’s drought categories both at the end of May and the final week of August 2012. By the end of May, 64% (100% minus 35.98%) of the United

States was experiencing some extent of drought whether it be abnormally dry conditions

64

(D0) all the way to exceptional drought (D4; Fig. 5.3). Just over 70% of the D0-D4 area experiencing drought ranked in the two lowest categories D0 or D1 (45.08% divided by

64%). This means roughly 30% of the areas with dry conditions were suffering from severe through exceptional drought (D2-D4). A substantial change in drought coverage takes place from May 29th to August 28th in 2012.

By the end of August, almost 78% of the United States was considered to be experiencing drought (D0-D4). This was up from 64% at the end of May. Of all areas experiencing dryness, only 45% were characterized by abnormally dry or moderate drought conditions (D0-D1). This meant the majority of areas experiencing drought

(55%) had severe conditions or worse (D2-D4) at the end of August. The biggest change from May 29th to August 28th was the expansion of areas covered by extreme (D3) and exceptional (D4) drought. Of the total area of the United States experiencing drought at the end of May, ~8% were categorized as D3-D4 drought severity. By the end of August, areas of drought with severity D3-D4 conditions had risen to ~30% coverage, an increase of 22% (Fig. 5.3).

Table 5.1 further shows the statistics from the U.S. Drought Monitor on August

28, 2012 (http://droughtmonitor.unl.edu/MapsAndData/MapArchive.aspx). Table 5.1 makes drought comparisons between the last full week of August and earlier periods. At the beginning of the year of 2012, 49.59% of the United States was experiencing dry conditions. The majority of these areas were classified as moderately or abnormally dry.

2011 was also a dry year over almost half of the United States. On August 30, 2011,

11.21% of the United States had exceptional drought compared to 6.04% on August 28,

65

2012 indicating exceptionally dry conditions also existed in summer 2011. However the

2011 drought was not as expansive as that of 2012, as much more surface area was experiencing D0-D3 level drought in 2012 relative to 2011.

66

67

Fig. 5.1: Climate Division average PDSI values from May through August 2012. Data are from the NCDC. (http://www.ncdc.noaa.gov/temp-and-precip/drought/historical-palmers/maps) 67

68

Fig. 5.2: 2012 drought expansion as shown by the U.S. Drought Monitor 68

69

Fig. 5.3: Drought ranking system used by the U.S. Drought Monitor; percentage of each drought category at the end of May to the end of August 2012

69

Table 5.1: Data directly from the U.S. Drought Monitor website for the end of August 2012. Previous year (2011) and beginning of year drought statistics are shown and can be compared to Aug. 21, 2012 United States drought conditions.

70

5.2. Comparing the drought of 2012 to droughts of historic past

How bad was the drought of 2012? Looking back in history to find dry conditions as extensive as those of 2012, one would have to go back to the summer of 1956, (Fig.

5.4). It was discussed in Chapter 4 how the summer of 2012 ranked as one of the driest summers in the timeframe of the data collected in RPC1 (Ohio River Valley), RPC2

(upper Midwest and eastern Northern Plains), and RPC5 (Southern Plains) (see Fig. 5.1).

Similarly to the 2012 summer, the summer of 1956 ranks as one of the driest in RPC2 and RPC5. Unlike the summer of 2012 where drought expansion and severity peaked in

August, conditions following the summer of 1956 worsen as drought conditions peaked in December (Fig. 5.5) leading to one of the longest droughts in United States history. By the middle of the summer of 2012, the drought had become so expansive across the

United States that it ranked right behind 1956 as one of the largest droughts in history

(http://www.ncdc.noaa.gov/sotc/drought/2012/8). Fig. 5.4 shows the expansiveness of the drought of 1956 in terms of PDSI through the summer of that year. Comparing this figure with Fig. 5.1, one can see that the drought in 2012 covered a larger land mass than the drought in 1956 by the end of August. The drought of 1956 was located mainly in the central and Southern Plains with large areas of area experiencing PDSI values of -4 or worse ranking them in the extreme drought category. Fig. 5.5 shows the peak of the drought of 1956, occurring in December of that year, compared to the peak of dryness in

2012, which occurred in August. The two droughts covered a lot of the same land area in terms of extreme drought with 2012 having a slightly larger overall land coverage.

71

72

Fig. 5.4: Climate Division average PDSI values from May through August 1956. Data are from the NCDC. (http://www.ncdc.noaa.gov/temp-and-precip/drought/historical-palmers/maps). 72

73

Fig. 5.5: Peak of the 1956 drought (December) compared to the peak of the 2012 drought (August).

73

The extremely warm temperatures and dry soil conditions experienced in 2012 across the United States were also very similar to droughts that occurred as recently as

1988 and as far back as 1934. The drought of 1988 (Fig. 5.6) was a wide spread event covering much of the and states with moderate to extreme drought. The summer of 1988 registered in RPCs 1 and 2 was one of the driest summers on record. The drought of 1988 was well known for its costly forest fires including devastation to Yellowstone National Park. The drought of 2012 brought very similar temperatures to that of 1988 as well as precipitation values.

One of the worst and most expansive droughts in history occurred in 1934 which spanned much of the central and western United States (Fig. 5.7). The summer of 1934 appeared as one of the driest in RPCs 1, 2, and 5 (Southern Plains). This summer ranked as the overall driest in the upper Midwest region. The drought of 1934 represents the largest expansion of extreme drought in United States history. The mid-1930s era was known as the “dust bowl” both for the extreme dryness on the Southern Plains as well as expansive dust storms that were prevalent during this time. Poor soil management strategies during this time allowed soil to dry out, turn to dust, and become susceptible to wind erosion (Worster, 1979).

74

75

Fig. 5.6: Climate Division average PDSI values from May through August 1988. Data are from the NCDC. (http://www.ncdc.noaa.gov/temp-and-precip/drought/historical-palmers/maps).

75

76

Fig. 5.7: Climate Division average PDSI values from May through August 1934. Data are from the NCDC. (http://www.ncdc.noaa.gov/temp-and-precip/drought/historical-palmers/map).

76

5.3. Temperature change and precipitation in summer 2012

This section will analyze month to month changes from May to August 2012 in T- max, T-min, DTR, and precipitation. Temperature and precipitation data from the summer of 2012 were collected and analyzed from 21 states comprising parts, or all of five of the nine climate regions identified (Fig. 5.8) by the National Climatic Data Center

(NCDC) and used by the National Oceanic and Atmospheric Administration (NOAA)

(http://www.ncdc.noaa.gov/monitoring-references/maps/us-climate-regions.php). These nine regions were created due to their consistent climate characteristics which make climate anomalies occurring in current times easier to identify against historical averages

(Karl and Koss, 1984). Table 5.2 lists the specific states from which NWS weather data were collected for the 2012 summer.

Four tables (Tables 5.3, 5.4, 5.6, and 5.8) depict the monthly mean T-max, T-min,

DTR, and precipitation totals of the 21 states (Table 5.2) from May to August 2012.

Three others (Tables 5.5, 5.7, and 5.9) were created to depict the monthly fluctuations in

T-max, T-min, DTR, and precipitation compared to seasonal changes. DTR is calculated from the air temperature data as was average monthly temperature along with departures from normal. Precipitation departures from normal are also available for the corresponding states. Highlighted values in Tables 5.3, 5.4, 5.6, and 5.8 note the state with the highest T-max, highest T-min, largest DTR, highest average temperature, largest departure from normal temperature, lowest and highest precipitation, and finally the greatest departure from normal. Highlighted values in Tables 5.5, 5.7, and 5.9 represent

T-max, T-min, and DTR changes larger than seasonally expected values. Average

77

Fig. 5.8: United States Climate Regions (http://www.ncdc.noaa.gov/monitoring- references/maps/us-climate-regions.php#references) used by NOAA. The 5 regions indicated by a star are those from which NWS station data are gathered during May- August 2012. The states of these regions are listed in Table 5.2.

2012 Drought in the United States Arkansas Colorado Illinois Indiana Iowa Kansas Kentucky

Louisiana Michigan Mississippi Missouri Nebraska North Dakota Ohio Oklahoma South Tennessee Texas Wisconsin Wyoming Dakota

Table 5.2: States for which NWS first-order station data were collected for the months of May-August 2012

78 temperature change between months is also indicated by a red color if it is warmer than normal or blue if cooler.

Table 5.3 shows data for the month of May leading into the summer of 2012.

Every state shown experienced warmer than normal temperatures for that time of year.

Ohio led the way with mean temperatures 3.5°C above normal. The state that was closest to having normal temperatures for May was North Dakota (+0.81°C). Precipitation was below normal in all of the states in the study excluding Wisconsin and Minnesota, which received 7.14 cm above normal. Arkansas had just 3.73 cm of precipitation, 9.40 cm below normal, which was the most below normal for any state in the study. Colorado received the lowest amount of precipitation of any other state in May and was coincidentally also the state with the largest DTR. Colorado, Nebraska, and Wyoming for the most part displayed the largest DTR of any of the 21 states in the study for all of the investigated months (May-August 2012). Based on previous studies (e.g., Dai et al.,

1999), the high DTR values are likely due to climatologically low precipitation compared to the other states, very low surface humidities (little Gulf of Mexico moisture), and relatively high topographic elevations. A combination of higher than normal temperatures and lack of precipitation were the initial precursors for the drought that occurred in these

21 states in the summer of 2012.

As May turned to June, drought conditions worsened (Figs. 5.1 and 5.2). As states warmed up in June, the Plains states continued to have average temperature departures well above normal especially in Wyoming, Colorado, South Dakota, Nebraska, and

Kansas (Table 5.4). The Midwestern states continued to experience warmer than normal

79

MAY 2012 STATE AVG AVG AVG DTR Monthly DPTR FM Total DPTR FM Temperature Temperature (°C) AVG Normal PRECIP Normal MAX (°C) MIN (°C) TEMP TEMP (cm) PRECIP(cm) (°C) (°C) Wyoming 18.06 2.00 16.06 10.03 +0.89 3.89 -1.75 Colorado 21.67 4.33 17.33 13.00 +1.92 2.31 -2.69 N. Dakota 20.00 5.61 14.39 12.81 +0.81 5.18 -0.89 S. Dakota 22.11 7.17 14.94 14.64 +1.31 7.09 -0.33 Nebraska 25.00 8.61 16.39 16.81 +2.11 5.18 -3.68 Kansas 28.56 12.61 15.94 20.58 +3.08 3.73 -6.02 Oklahoma 29.50 15.61 13.89 22.56 +2.44 5.46 -6.68 Texas 30.94 17.39 13.56 24.17 +1.58 7.32 -2.13 Arkansas 29.56 16.22 13.33 22.89 +2.44 3.73 -9.40 Louisiana 30.56 18.67 11.89 24.61 +1.50 5.51 -6.99 Tennessee 27.94 14.78 13.17 21.36 +2.36 8.26 -3.10 Mississippi 30.17 16.83 13.33 23.50 +1.58 7.95 -3.84 Minnesota 21.28 8.00 13.28 14.64 +2.28 15.14 +7.14 Iowa 25.33 11.39 13.94 18.36 +2.89 8.92 -1.57 Missouri 27.67 14.06 13.61 20.86 +2.89 5.49 -6.65 Wisconsin 22.33 8.28 14.06 15.31 +2.69 12.22 +3.18 Illinois 26.94 13.17 13.78 20.06 +3.25 6.27 -4.32 Indiana 26.89 12.78 14.11 19.83 +3.47 6.99 -3.73 Michigan 22.17 7.78 14.39 14.97 +2.92 7.62 -0.36 Ohio 26.06 12.22 13.83 19.14 +3.50 9.35 -0.48 Kentucky 27.39 14.11 13.28 20.75 +2.69 10.80 -0.48 Table 5.3: May 2012 state averaged T-max (°C) and T-min (°C) temperatures with departures from normal. Monthly precipitation (cm) is also included along with departures from normal. The average DTR (°C) is the difference T-max – T-min.

80 temperatures although departures from normal were lower than in May (Table 5.3).

Conditions in Colorado quickly deteriorated with temperatures over 3.5°C warmer than normal and precipitation 2.82 cm below normal. Average temperature during the day in

Colorado reached 29.33°C with the low temperatures falling to 10.56°C. This led to an average DTR of 18.78°C, the largest of any of the 21 states for the month of June.

Outside of neighboring Wyoming, DTR for this area was greater than or equal to 2.5°C larger than any other state. Like Colorado, Wyoming also received below average precipitation albeit a larger deficit of 3.43 cm. These temperatures and precipitation values suggest net clear sky conditions, especially at night, allowing the air temperatures to drop considerably as shown by the large DTR values (Table 5.4). Precipitation values were below normal for every state in the study for June except for Minnesota, which received a surplus of 0.56 cm (Table 5.4). In some states, absolute values of the departures from normal were larger than the actual rainfall totals thus indicating precipitation was less than half of normal. Conditions in the Midwest were the worst in terms of dryness as many states were over 4.5 cm short of normal precipitation. Indiana saw the largest deficit of precipitation suffering a loss of 7.16 cm of rainfall, receiving roughly 30% of normal.

Table 5.5 shows the temperature changes from May to June that occurred within the 21 states analyzed. Air temperatures should seasonally increase in all states during the progression of May to June. The normal month to month changes were calculated based on monthly average temperature changes dating from 1895-2011 (Column 8). This allows for comparison to the month to month temperature changes that occurred during the

81

JUNE 2012 STATE AVG AVG AVG Monthly DPTR FM Total DPTR FM Temperature Temperature DTR AVG Normal PRECIP Normal MAX (°C) MIN (°C) (°C) TEMP TEMP (°C) (cm) PRECIP(cm) (°C) Wyoming 26.39 7.78 18.61 17.08 +2.81 1.14 -3.43 Colorado 29.33 10.56 18.78 19.94 +3.58 1.07 -2.82 N. Dakota 25.72 11.56 14.17 18.64 +1.47 6.65 -2.11 S. Dakota 29.11 13.61 15.50 21.36 +2.61 5.54 -3.18 Nebraska 31.06 14.94 16.11 23.00 +2.81 4.19 -5.23 Kansas 33.33 17.22 16.11 25.28 +2.25 5.49 -4.83 Oklahoma 33.17 19.11 14.06 26.14 +1.14 6.22 -4.01 Texas 35.06 20.89 14.17 27.97 +1.44 4.72 -2.64 Arkansas 32.72 18.39 14.33 25.56 +0.75 5.03 -5.16 Louisiana 32.94 21.17 11.78 27.06 +0.47 10.80 -1.22 Tennessee 30.44 15.83 14.61 23.14 -0.17 5.46 -5.28 Mississippi 32.22 19.06 13.17 25.64 -0.14 8.79 -1.83 Minnesota 26.06 12.61 13.44 19.33 +1.67 11.07 +0.56 Iowa 28.39 15.44 12.94 21.92 +1.25 7.14 -4.70 Missouri 31.00 16.61 14.39 23.81 +0.97 4.83 -6.96 Wisconsin 26.50 12.56 13.94 19.53 +1.64 8.48 -2.13 Illinois 29.67 15.44 14.22 22.56 +0.61 4.39 -6.02 Indiana 29.44 14.56 14.89 22.00 +0.53 3.28 -7.16 Michigan 26.11 12.17 13.94 19.14 +1.69 7.62 -0.69 Ohio 28.39 14.11 14.28 21.25 +0.67 5.38 -4.65 Kentucky 30.11 14.89 15.22 22.50 -0.11 3.81 -6.96 Table 5.4: June 2012 state averaged T-max (°C) and T-min (°C) temperatures with departures from normal. Monthly precipitation (cm) is also included along with departures from normal. The average DTR (°C) is the difference T-max – T-min.

82 summer of 2012. In 2012, average May-June temperature changes range from a warming of about 2°C to ~7°C (Column 7). The normal May-June temperature change in these states is a much smaller range of warming from ~3.5-5.5°C.

Wyoming, Colorado, and the Northern Plains states of North and South Dakota warmed more than any other state in the transition from May to June. These states also warmed more than what occurs seasonally (Table 5.5 Columns 2, 4, and 8). Wyoming saw the highest T-max change at a little over an 8 °C increase in statewide averaged T- max from May to June. Most states examined had below the normal seasonal warming, primarily because May (Table 5.3) departures were very large while those of June (Table

5.4) were smaller. Monthly average temperatures were above normal for both May and

June for most of these states.

All of the states excluding North Dakota, Nebraska, Louisiana, Mississippi, Iowa, and Michigan had a larger increase in T-max than in T-min leading to an increase in DTR between the 2 months. Drought conditions became noticeably worse in Wyoming and

Colorado (Figs. 5.1) as PDSI values took a turn to extreme drought conditions. These two states would continue to experience extreme drought for the rest of the 2012 summer. T- max and T-min both increased well above seasonal normal changes in these two states as well. In the Dakotas and Nebraska, T-max and T-min also increased above the average, but there was little change in DTR unlike Wyoming and Colorado. Nebraska had a substantial change in drought in June (Fig. 5.1), but the Dakotas changed only slightly.

Drought generally intensified from May to June in the region between Texas and

Ohio (Figs. 5.1 and 5.2). As this area became drier, a net increase in DTR occurred in

83 most of the states with Iowa and Wisconsin excluded (Table 5.5); despite the smaller than usual increases in both T-max and T-min. Iowa became drier from May to June (Fig.

5.1), but had a small net decrease in DTR (Table 5.5). Mississippi became somewhat moister (Fig. 5.1) between May and June and its DTR went down slightly. Overall (Table

5.5), DTR increased in many states as the drought worsened from May to June.

July (Table 5.6) had the warmest temperatures over the summer of 2012 with state wide average T-max between 29°C-38°C. Oklahoma had the highest average T-max of

37.50, which was over 3°C warmer than normal. The average daily temperature for

Oklahoma was also the warmest for any state in July at 29.94°C. Along with the scorching heat, Oklahoma had one of the smallest amounts of precipitation for any state

(Table 5.6) with only 2.95 cm, which was almost 5 cm below normal precipitation.

Nebraska received the smallest amount of precipitation at 2.62 cm, 5.05 cm below normal. As in June, the Midwest was still experiencing extreme heat with temperatures ranging from 2 to 3.5°C warmer than normal. Illinois had the largest anomaly of the region with temperatures 3.47°C warmer than normal for July. Nightly average Wyoming temperatures were much lower than any other state in Table 5.6 resulting in the largest

DTR for the month. Even though temperatures were much higher in the month of July, average precipitation values were not as far below normal in June. States such as

Colorado, Texas, Louisiana, Tennessee, Mississippi, Michigan, and Kentucky received above normal rainfall. DTR values on average were lower in these wet states compared to the drier states. Louisiana received the most precipitation out of any state in July at 19.66 cm, which was almost 5.5 cm above normal. Louisiana had the highest T-min (Table 5.6)

84

TEMPERATURE CHANGES FROM MAY TO JUNE 2012 1. 2. 3. 4. 5. 6. 7. 8. Δ Normal Δ Normal Δ Normal Δ AVG Normal STATE T-MAX Δ T-MIN Δ DTR Δ TEMP Δ AVG 2012 T-MAX 2012 T-MIN 2012 DTR 2012 TEMP (°C) (°C) (°C) (°C) (°C) (°C) (°C) (°C)

Wyoming +8.33 +5.89 +5.78 +4.39 +2.56 +1.50 +7.06 +5.14 Colorado +7.67 +5.83 +6.22 +4.72 +1.44 +1.11 +6.94 +5.28 N. Dakota +5.72 +4.72 +5.94 +5.61 -0.22 -0.89 +5.83 +5.17 S. Dakota +7.00 +5.22 +6.44 +5.61 +0.56 -0.39 +6.72 +5.42 Nebraska +6.06 +5.44 +6.33 +5.56 -0.28 -0.11 +6.19 +5.50 Kansas +4.78 +5.56 +4.61 +5.50 +0.17 +0.06 +4.69 +5.53 Oklahoma +3.67 +4.83 +3.50 +4.94 +0.17 -0.11 +3.58 +4.89 Texas +4.11 +3.83 +3.50 +4.06 +0.61 -0.22 +3.81 +3.94 Arkansas +3.17 +4.28 +2.17 +4.44 +1.00 -0.17 +2.67 +4.36 Louisiana +2.39 +3.28 +2.50 +3.67 -0.11 -0.39 +2.44 +3.47 Tennessee +2.50 +4.06 +1.06 +4.56 +1.44 -0.50 +1.78 +4.31 Mississippi +2.06 +3.67 +2.22 +4.06 -0.17 -0.39 +2.14 +3.86 Minnesota +4.78 +4.94 +4.61 +5.67 +0.17 -0.72 +4.69 +5.31 Iowa +3.06 +4.94 +4.06 +5.44 -1.00 -0.50 +3.56 +5.19 Missouri +3.33 +4.67 +2.56 +5.06 +0.78 -0.39 +2.94 +4.86 Wisconsin +4.17 +5.06 +4.28 +5.50 -0.11 -0.44 +4.22 +5.28 Illinois +2.72 +5.11 +2.28 +5.17 +0.44 -0.06 +2.50 +5.14 Indiana +2.56 +5.00 +1.78 +5.22 +0.78 -0.22 +2.17 +5.11 Michigan +3.94 +5.39 +4.39 +5.39 -0.44 0.00 +4.17 +5.39 Ohio +2.33 +4.78 +1.89 +5.11 +0.44 -0.33 +2.11 +4.94 Kentucky +2.72 +4.28 +0.78 +4.83 +1.94 -0.56 +1.75 +4.56 Table 5.5: Temperature changes that occurred from May to June 2012. The May to June 2012 progressions of T-max and T-min are represented in Columns 1 and 3. Diurnal temperature range is shown in Col. 5. Comparative long-term normal values are in Columns 2, 4, and 6. Columns 7 and 8 represent average temperature change in 2012 and seasonal normal average temperature changes respectively.

85 resulting in the smallest DTR of any state in July. Amidst all of the precipitation that

Louisiana received, monthly average temperatures were still warmer than normal (Table

5.6). Amongst the wetter states, Arkansas experienced a similar DTR but remained dry, receiving 2.41 cm of precipitation below normal.

During an average summer, the transition from June to July should bring warmer temperatures across the United States (Column 8; Table 5.7). Compared to seasonal warming, the change of monthly temperatures in 2012 was very extreme. Only four states

(Wyoming, Colorado, Texas, and Louisiana) had June-July temperature increases that were below seasonal average changes (Column 7; Table 5.7). Some states in the Midwest

(Illinois, Indiana, and Kentucky) experienced average temperature change twice as large as occurs normally. Texas had no change in T-max from June to July but did experience a warming in T-min that was below seasonal fluctuations. This resulted in an average DTR change 1°C smaller than normal (Column 5; Table 5.7). Not only were average temperatures in July warmer than normal (Table 5.6), the change in both T-max and T- min between June and July for most of the states were larger than normal.

The Midwest and Plains drought worsened in July according to PDSI values (Fig.

5.1) and the U.S. Drought Monitor (Fig. 5.2). Combining the decrease in soil moisture and extreme seasonal warming, the drought of 2012 was only getting worse. Parts of

Illinois, Indiana, Kentucky, and Missouri had joined the ranks of extreme drought with

PDSI values of -4.00 and below by the end of July. Extreme drought conditions in the

Rocky along with areas in the Plains saw no improvement in soil moisture conditions throughout July. Portions of southeastern Texas received drought

86

JULY 2012 STATE AVG AVG AVG Monthly DPTR FM Total DPTR FM Temperature Temperature DTR AVG Normal PRECIP Normal MAX (°C) MIN (°C) (°C) TEMP TEMP (°C) (cm) PRECIP(cm) (°C) Wyoming 29.83 12.44 17.39 21.14 +2.58 2.84 -0.36 Colorado 29.78 13.11 16.67 21.44 +1.89 5.77 +0.43 N. Dakota 30.72 16.06 14.67 23.39 +2.81 5.61 -0.97 S. Dakota 33.94 18.00 15.94 25.97 +3.39 3.91 -2.51 Nebraska 35.11 18.22 16.89 26.67 +3.03 2.62 -5.05 Kansas 37.28 20.94 16.33 29.11 +3.08 3.25 -5.33 Oklahoma 37.50 22.39 15.11 29.94 +2.33 2.95 -4.42 Texas 35.06 22.00 13.06 28.53 +0.58 6.86 +0.74 Arkansas 35.44 22.50 12.94 28.97 +2.14 7.16 -2.41 Louisiana 33.00 22.89 10.11 27.94 +0.22 19.66 +5.49 Tennessee 33.06 20.83 12.22 26.94 +1.86 14.61 +3.28 Mississippi 33.39 22.11 11.28 27.75 +0.61 17.83 +5.18 Minnesota 30.00 17.00 13.00 23.50 +3.03 8.33 -0.81 Iowa 33.33 19.50 13.83 26.42 +3.19 2.97 -6.99 Missouri 36.11 21.28 14.83 28.69 +3.36 4.32 -5.36 Wisconsin 30.67 16.94 13.72 23.81 +3.22 8.10 -1.52 Illinois 34.67 20.67 14.00 27.67 +3.47 3.56 -5.89 Indiana 33.78 19.78 14.00 26.78 +3.19 6.43 -3.45 Michigan 30.06 16.11 13.94 23.08 +3.00 9.12 +1.35 Ohio 31.78 18.78 13.00 25.28 +2.47 8.76 -1.50 Kentucky 33.39 20.50 12.89 26.94 +2.36 14.07 +2.84 Table 5.6: July 2012 state averaged T-max (°C) and T-min (°C) temperatures with departures from normal. Monthly precipitation (cm) is also included along with departures from normal. The average DTR (°C) is the difference T-max – T-min.

87 relief, albeit very small-scale, as most of the state continued to show moderate drought or worse (Fig. 5.1). Louisiana also improved in soil moisture as only two northern climate regions within the state were experiencing drought conditions at the end of July.

DTR variability in the Midwest between June and July responded to changes in precipitation amount. For example, DTR decreased in Colorado, Texas, Louisiana,

Tennessee, Mississippi, and Kentucky between June and July (Table 5.7) while the rainfall went up between the 2 months (Table 5.6). This was especially true in Tennessee and Kentucky where rainfall was well over 2.5 cm above normal (Table 5.6) and DTR fell over 2°C between June and July. Louisiana saw 5.49 cm of rainfall above normal in

July (Table 5.6), the most of any state in the record, but only had a DTR decrease of

1.67°C (Table 5.7). However, in Kansas, Missouri, and Illinois, little change in DTR occurs (Table 5.7) and soil conditions remained relatively dry and rainfall over 5 cm below normal. Rainfall was 6.99 cm below normal in Iowa between June and July and

DTR increased by 0.89°C. Minnesota was not as adversely affected by drought conditions in June or July, with slightly below normal rain in both months, while DTR was relatively low compared to all the other Midwestern states. The conditions between

June and July reveal the changes that can occur on a state level as opposed to a regional level. They also reveal that the basic physical links between precipitation PDSI and DTR are confirmed.

In August (Table 5.8), temperature departures were within a degree of normal in most states. At this point in the year, the drought had already taken its toll on a lot of areas. Precipitation was much more variable across this month with high rainfall in

88

TEMPERATURE CHANGES FROM JUNE TO JULY 2012 1. 2. 3. 4. 5. 6. 7. 8. Δ Normal Δ Normal Δ DTR Normal Δ AVG Normal STATE T-MAX Δ T-MIN Δ 2012 Δ DTR TEMP Δ AVG 2012 T-MAX 2012 T-MIN (°C) (°C) 2012 TEMP (°C) (°C) (°C) (°C) (°C) (°C)

Wyoming +3.44 +4.94 +4.67 +3.61 -1.22 +1.33 +4.06 +4.28 Colorado +0.44 +3.11 +2.56 +3.28 -2.11 -0.17 +1.50 +3.19 N. Dakota +5.00 +4.00 +4.50 +2.83 +0.50 +1.17 +4.75 +3.42 S. Dakota +4.83 +4.44 +4.39 +3.22 +0.44 +1.22 +4.61 +3.83 Nebraska +4.06 +3.78 +3.28 +3.11 +0.78 +0.67 +3.67 +3.44 Kansas +3.94 +3.22 +3.72 +2.78 +0.22 +0.44 +3.83 +3.00 Oklahoma +4.33 +2.94 +3.28 +2.28 +1.06 +0.67 +3.81 +2.61 Texas 0.00 +1.39 +1.11 +1.44 -1.11 -0.06 +0.56 +1.42 Arkansas +2.72 +2.17 +4.11 +1.89 -1.39 +0.28 +3.42 +2.03 Louisiana +0.06 +1.00 +1.72 +1.28 -1.67 -0.28 +0.89 +1.14 Tennessee +2.61 +1.56 +5.00 +2.00 -2.39 -0.44 +3.81 +1.78 Mississippi +1.17 +1.11 +3.06 +1.61 -1.89 -0.50 +2.11 +1.36 Minnesota +3.94 +2.89 +4.39 +2.72 -0.44 +0.17 +4.17 +2.81 Iowa +4.94 +2.67 +4.06 +2.44 +0.89 +0.22 +4.50 +2.56 Missouri +5.11 +2.72 +4.67 +2.28 +0.44 +0.44 +4.89 +2.50 Wisconsin +4.17 +2.67 +4.39 +2.72 -0.22 -0.06 +4.28 +2.69 Illinois +5.00 +2.22 +5.22 +2.28 -0.22 -0.06 +5.11 +2.25 Indiana +4.33 +2.11 +5.22 +2.11 -0.89 0.00 +4.78 +2.11 Michigan +3.94 +2.61 +3.94 +2.67 0.00 -0.06 +3.94 +2.64 Ohio +3.39 +2.17 +4.67 +2.28 -1.28 -0.11 +4.03 +2.22 Kentucky +3.28 +1.78 +5.61 +2.17 -2.33 -0.39 +4.44 +1.97 Table 5.7: Temperature changes that occurred from June to July 2012. The June to July 2012 progressions of T-max and T-min are represented in Columns 1 and 3. Diurnal temperature range is shown in Col. 5. Comparative long-term normal values are in Columns 2, 4, and 6. Columns 7 and 8 represent average temperature change in 2012 and seasonal normal average temperature changes respectively.

89

Louisiana and Mississippi while several drier states were less than 2.5 cm below normal.

Arkansas DTR continued to increase between July and August despite the 2.62 cm of rainfall above normal. DTR values of Louisiana and Mississippi were the smallest of any state but remained steady during this very wet month (Table 5.8), compared to June when conditions were dry. Colorado and Wyoming remain warmer than normal in August

(Table 5.8) accompanied by very small amounts of rainfall relative to their averages for

August. Wyoming received the least amount of precipitation out of any state at 0.66 cm.

Similar to previous months in 2012, DTR values remained the largest within Wyoming with Colorado and the Northern Plains states (ND, SD, NE, KS) experienced relatively high DTR. This was due partly to a substantial drop in T-min in some states but also due to a continuing lack of rainfall in August.

As July turns to August, mean temperatures, not unexpectedly, begin to decline in every state except Texas (Table 5.9). Although most states in August were still unseasonably warm, they were not far from normal (Table 5.8). Outside of Texas the average temperature change declined more than is seasonally expected in every state

(Columns 7 and 8; Table 5.9). DTR on the other hand increased in most states as T-min declined more than T-max. Only Mississippi and Michigan had a decrease in DTR between July and August.

As rainfall increased in Louisiana and Mississippi, conditions in the upper

Midwest and Plains continued to deteriorate (Fig. 5.1), as shown by the rain departures in

Table 5.8. Many climate divisions along the Gulf Coast were very moist by the end of

August as the Plains and Midwest trended towards extremely dry. The cooling shown in

90

AUGUST 2012 STATE AVG AVG AVG Monthly DPTR FM Total DPTR FM Temperature Temperature DTR AVG Normal PRECIP Normal MAX (°C) MIN (°C) (°C) TEMP TEMP (°C) (cm) PRECIP(cm) (°C) Wyoming 28.39 10.00 18.39 19.19 +1.72 0.66 -2.01 Colorado 28.44 11.28 17.17 19.86 +1.33 2.72 -2.39 N. Dakota 27.94 11.44 16.50 19.69 +0.31 3.86 -1.50 S. Dakota 30.33 12.94 17.39 21.64 +0.25 2.84 -2.59 Nebraska 31.39 14.11 17.28 22.75 +0.33 2.69 -4.14 Kansas 32.94 16.50 16.44 24.72 -0.47 6.63 -1.30 Oklahoma 35.28 19.72 15.56 27.50 +0.28 6.15 -1.14 Texas 35.78 21.67 14.11 28.72 +1.00 4.67 -1.22 Arkansas 33.56 20.00 13.56 26.78 +0.39 11.10 +2.62 Louisiana 32.83 22.67 10.17 27.75 +0.14 22.50 +10.72 Tennessee 30.56 17.89 12.67 24.22 -0.33 9.09 -0.51 Mississippi 32.00 20.78 11.22 26.39 -0.47 22.12 +11.94 Minnesota 26.39 12.17 14.22 19.28 +0.22 5.18 -3.58 Iowa 29.06 14.22 14.83 21.64 -0.28 7.32 -2.54 Missouri 32.22 16.89 15.33 24.56 +0.06 6.20 -3.20 Wisconsin 26.50 12.56 13.94 19.53 +0.33 5.94 -3.61 Illinois 30.50 15.72 14.78 23.11 +0.03 8.89 +0.05 Indiana 29.33 15.06 14.28 22.19 -0.31 10.29 +1.47 Michigan 26.22 13.17 13.06 19.69 +0.75 6.93 -0.97 Ohio 28.72 15.06 13.67 21.89 +0.17 7.54 -1.27 Kentucky 30.39 16.94 13.44 23.67 -0.25 8.05 -1.30 Table 5.8: August 2012 state averaged T-max (°C) and T-min (°C) temperatures with departures from normal. Monthly precipitation (cm) is also included along with departures from normal. The average DTR (°C) is the difference T-max – T-min.

91

Table 5.9 did not bring relief for the drought of 2012 as low soil moisture continued to be extensive. The drought of 2012 had peaked by the end of August in terms of land coverage, but would continue to be prevalent throughout the end of the year in northern parts of the south and the Northern Plains.

92

TEMPERATURE CHANGES FROM JULY TO AUGUST 2012 1. 2. 3. 4. 5. 6. 7. 8. Δ Normal Δ Normal Δ DTR Normal Δ AVG Normal STATE T-MAX Δ T-MIN Δ 2012 Δ DTR TEMP Δ AVG 2012 T-MAX 2012 T-MIN (°C) (°C) 2012 TEMP (°C) (°C) (°C) (°C) (°C) (°C)

Wyoming -1.44 -1.11 -2.44 -1.06 +1.00 -0.06 -1.94 -1.08 Colorado -1.33 -1.22 -1.83 -0.83 +0.50 -0.39 -1.58 -1.03 N. Dakota -2.78 -0.83 -4.61 -1.56 +1.83 +0.72 -3.69 -1.19 S. Dakota -3.61 -1.06 -5.06 -1.33 +1.44 +0.28 -4.33 -1.19 Nebraska -3.72 -1.28 -4.11 -1.17 +0.39 -0.11 -3.92 -1.22 Kansas -4.33 -0.83 -4.44 -0.83 +0.11 0.00 -4.39 -0.83 Oklahoma -2.22 -0.22 -2.67 -0.56 +0.44 +0.33 -2.44 -0.39 Texas +0.72 -0.06 -0.33 -0.39 +1.06 +0.33 +0.19 -0.22 Arkansas -1.89 -0.28 -2.50 -0.61 +0.61 +0.33 -2.19 -0.44 Louisiana -0.17 +0.06 -0.22 -0.28 +0.06 +0.33 -0.19 -0.11 Tennessee -2.50 -0.39 -2.94 -0.67 +0.44 +0.28 -2.72 -0.53 Mississippi -1.39 -0.11 -1.33 -0.44 -0.06 +0.33 -1.36 -0.28 Minnesota -3.61 -1.39 -4.83 -1.44 +1.22 +0.06 -4.22 -1.42 Iowa -4.28 -1.33 -5.28 -1.28 +1.00 -0.06 -4.78 -1.31 Missouri -3.89 -0.72 -4.39 -0.94 +0.50 +0.22 -4.14 -0.83 Wisconsin -4.17 -1.50 -4.39 -1.28 +0.22 -0.22 -4.28 -1.39 Illinois -4.17 -1.06 -4.94 -1.17 +0.78 +0.11 -4.56 -1.11 Indiana -4.44 -1.00 -4.72 -1.17 +0.28 +0.17 -4.58 -1.08 Michigan -3.83 -1.39 -2.94 -0.89 -0.89 -0.50 -3.39 -1.14 Ohio -3.06 -1.11 -3.72 -1.06 +0.67 -0.06 -3.39 -1.08 Kentucky -3.00 -0.61 -3.56 -0.72 +0.56 +0.11 -3.28 -0.67 Table 5.9: Temperature changes that occurred from July to August 2012. The July to August 2012 progressions of T-max and T-min are represented in Columns 1 and 3. Diurnal temperature range is shown in Col. 5. Comparative long-term normal values are in Columns 2, 4, and 6. Columns 7 and 8 represent average temperature change in 2012 and seasonal normal average temperature changes respectively.

93

Chapter 6: Conclusions

The purpose of this work was to investigate the behavior of spatial and temporal

DTR through the 20th and 21st centuries in relation to variability in soil moisture that could range from extreme dryness to moist conditions. Principal component analysis was applied to a historical Palmer Drought Severity Index dataset producing 10 significant regions of the United States exhibiting the most unique soil moisture variability. The analysis was subjected to a rotation procedure leaving 4 regions east of the Rocky

Mountains that account for the most soil moisture variability of the dataset. The 4 analyzed regions were the Ohio River Valley (RPC1), the upper Midwest and eastern

Northern Plains (RPC2), the southeastern United States (RPC3), and the Southern Plains

(RPC5). T-max, T-min, and DTR were acquired for all available locations within the 4 regions and then analyzed for long-term trends in comparison to variability in long-term soil moisture.

Soil moisture indices in the Ohio River Valley exhibited an overall increase in

PDSI throughout the 20th century into the early 21st century. Prior to 1955, drier conditions were prevalent. Long-term T-max in this region shows a downward trend of

~0.5°C to date. T-min has shown an overall increase of 0.7°C. Due to the overall trends in daily temperatures, DTR in the Ohio River Valley has decreased over the period of

94 record. The beginning of the 20th century actually saw an increase in DTR as drier conditions were persistent but the latter half of the 20th century and early 21st centuries have seen more moisture causing an overall decrease in DTR.

The upper Midwest and eastern Northern Plains (RPC2) showed similar results to

RPC1 with drier conditions early in the record with moist conditions in the last half of the

20th century into the 21st. PDSI has trended towards more moisture in this region. Long- term T-max in RPC2 has shown a steady non-trending regional average throughout the period of record with an overall very small decline in T-max. Two summers of very high

T-max were experienced in 1936 and 1988 that can be attributed to very dry conditions that influence the slight negative trend in overall T-max. T-min however, exhibits a very clear warming trend. Several summers following 1983 have been the highest T-min of the record producing a larger warming profile than is shown in RPC1. Because of the clear warming in T-min, overall DTR has decreased in RPC2 with the summers of 1951, 1993, and 2010 exhibiting the lowest DTR of the record. These summers correspond to very high PDSI summers indicting very moist conditions.

The southeastern United States hasn’t shown a strong trend in PDSI over the course of record but has shown a slight overall increase in PDSI much lower than the trends in moisture in RPCs 1 and 2. More recent dry conditions have made the PDSI trend not as positive as the other regions. T-max in RPC3 does not produce an overall trend due to oscillating summer temperatures throughout the period of record. Cool and wet conditions in the 1960s-70s gave way to drier and warmer temperatures closing the

20th century producing a very negligible but slight increase in overall T-max. T-min in

95

RPC3 shows a similar but more positive long-term trend than T-max, as lower temperatures were experienced in the 1960s and 1970s followed by warming with the highest T-min in the more recent summer of 2010. This summer was dry according to

PDSI but not considered as extreme as other dry summers that occurred in this region.

DTR in the southeastern United States shows an overall decrease due in large part to the overall larger increase in T-min than T-max. The wet period through the 1960s and 1970s starts the decline in DTR with the lowest DTR summers occurring in the late 20th and early 21st century, influencing the DTR trend more than the early dry summers in the record.

Soil moisture in the Southern Plains (RPC5) produced an overall decline in in

PDSI although it is very negligible. RPC5 shows wetter conditions at the beginning of the record giving way to dry conditions early in the 20th century with constant soil moisture until the late 20th into the 21st century. T-max in this region produced on overall long- term increase of ~0.6°C with the highest T-max in 2011. RPC5 was the only region to show an overall increase in T-max. The summers of 1934 and 2011 exhibited the warmest T-max of the record which were also linked to dry soil moisture conditions. The driest summer in RPC5 took place in 1956 but produced T-max ~2°C cooler than the summers of 1934 and 2011. T-min shows a warming trend slightly larger than that of T- max. The summers of 1934 and 2011 also produced the warmest T-min. Long-term DTR shows a near zero trend but does in fact decrease slightly over the course of record.

However, when analyzing the most extreme soil moistures either wet or dry, T- max seems to vary the most in all regions. T-min on the other hand did not seem to

96 fluctuate based on whether or not the soil moistures were wet or dry. In general, both T- max and T-min tend to be warmer during drier summers compared to cooler during wet summers. Variability in T-max seemed to play the largest role in DTR during extreme soil moisture cases as T-min did not fluctuate as much. As the case study of the drought of 2012 proved, DTR is larger for summers exhibiting dry soil moistures. As the climate continues to change, DTR should continue to shrink as heavy precipitation events become more common. DTR has been decreasing in the areas of the United States that have increases in long-term soil moisture.

As this thesis focuses on one specific variable that impacts the variability of

DTR (soil moisture), future work could focus on the influence of other parameters. This work briefly discussed the effect that precipitation has on temperatures. This could be expanded to analyze variability in precipitation in available data. Investigation of potential case studies of precipitation events across the United States can be conducted while probable temperature variations can be analyzed. Comparing the effects of cloud cover during precipitation events on temperatures compared to the effects of non- precipitating cloud cover would also be interesting. It seems cloud cover plays a role in limiting T-max while providing insulation for the Earth keeping T-min elevated at night

(Frich, 1992; Karl et al., 1993; Dai et al., 1999). Humidity is an interesting variable as well due to the fact that atmospheric water vapor is on the rise from increasing air and ocean water temperatures. The effects of varying humidites could also affect temperature.

With changing climates, it would be helpful to determine what types of these occurrences are becoming most common, as these events will be critical in influencing the long-term

97 trends in temperature into the future. Educating the public about the societal and economic impacts of a changing climate brings upon us important underlying issues.

What do these changes in temperature mean for life moving forward? What are the consequences of a changing climate? Can it be stopped or mitigated? These types of questions will continue to surface and be debated until there is a universal understanding backed by scientific evidence. Continued production of high resolution computer models with increased effectiveness and accuracies will be critical in forecasting the weather and climates on Earth in the future.

98

References

Berger, A. "A Brief History of the Astronomical Theories of Paleoclimates." Climate Change Inferences from Paleoclimate and Regional Aspects. Wien: Springer, (2012). 107-130.

Betts, A. K., and J. H. Ball, 1998: “FIFE surface climate and site-average dataset: 1987– 1989”. Journal of the Atmospheric Sciences 55.7 (1998): 1091.

Changnon, Stanley A., 1996: The Great Flood of 1993: Causes, Impacts, and Responses. Boulder, CO: Westview. 321pp.

Climate Change Indicators in the United States, 2012. 2nd ed. United States Environmental Protection Agency.

Cook, Edward R., David M. Meko, David W. Stahle, and Malcolm K. Cleaveland. "Drought Reconstructions for the Continental United States*." Journal of Climate 12 (1999): 1145-162.

Dai, Aiguo, Anthony D. Del Genio, and Inez Y. Fung. "Clouds, Precipitation and Temperature Range." Nature (1997): 665-66.

Dai, Aiguo, Kevin E. Trenberth, and Thomas R. Karl. "Effects of Clouds, Soil Moisture, Precipitation, and Water Vapor on Diurnal Temperature Range." Journal of Climate 12 (1999): 2451-473.

Dai, Aiguo, Kevin E. Trenberth, and Taotao Qian. "A Global Dataset of Palmer Drought Severity Index for 1870–2002: Relationship with Soil Moisture and Effects of Surface Warming." Journal of Hydrometeorology 5 (2004): 1117-130.

Diaz, Henry F. "Drought in the United States: Some Aspects of Dry and Wet Periods in the Contiguous United States, 1895-1981." Journal of Climate and Applied Meteorology 22.1 (1983): 3-16.

99

Eash, David A. "Effects of the 1993 Flood on the Determination of Flood Magnitude and Frequency in Iowa." Floods in the Upper Basin, 1993. Washington, DC: United States Government Printing Office, 1997.

Easterling, David R., Trevor W. R. Wallis, Jay H. Lawrimore, and Richard R. Heim. "Effects of Temperature and Precipitation Trends on U.S. Drought." Geophysical Research Letters 34.L20709 (2007): 1-4.

Fung, Inez Y., D. E. Harrison, and Andrew A. Lacis. "On the Variability of the Net Longwave Radiation at the Ocean Surface." Reviews of Geophysics 22.2 (1984): 177-93.

Frich, Povl. "Cloudiness and Diurnal Temperature Range." 5th International Meeting on Statistical Climatology 22-26 June 1992, Toronto, Canada. Canada: Environment Canada, 1992. 91-94.

Grigg, Neil S. "The 2011–2012 Drought in the United States: New Lessons from a Record Event." International Journal of Water Resources Development (2014): 183-99.

Groisman, Pavel Ya., Thomas R. Karl, David R. Easterling, Richard R. Knight, Paul F. Jamason, Kevin J. Hennessy, Ramasamy Suppiah, Cher M. Page, Joanna Wibig, Krzysztof Fortuniak, Vyacheslav N. Razuvaev, Arthur Douglas, Eirik Forland, and Pan-Mao Zhai. "Changes in the Probability of Heavy Precipitation: Important Indicators of Climatic Change." Climate Change 42 (1999): 243-83.

Groisman, Pavel Ya., Richard W. Knight, and Thomas R. Karl. "Changes in Intense Precipitation over the Central United States." Journal of Hydrometeorology 13 (2012): 47-66.

Gutzler, David S., and Tessia O. Robbins. "Climate Variability and Projected Change in the Western United States: Regional Downscaling and Drought Statistics." Climate Dynamics 37.5-6 (2011): 835-49.

Hansen, James H., and Makiko Sano. "Paleoclimate Implications for Human-Made Climate Change." Climate Change Inferences from Paleoclimate and Regional Aspects. Wien: Springer, 2012. 21-48.

House, M. R. "Orbital Forcing Timescales: An Introduction." Geological Society, London, Special Publications 85 (1995): 1-18.

Hu, Qi, and Gary D. Wilson. "Effects of temperature anomalies on the Palmer Drought Severity Index in the central United States." International Journal of Climatology 20.15 (2000): 1899-1911. 100

Huybers, Peter, and William Curry. "Links between Annual, Milankovitch and Continuum Temperature Variability." Nature 441 (2006): 329-32.

Karl, Thomas, and Walter James Koss. Regional and National Monthly, Seasonal, and Annual Temperature Weighted by Area, 1895-1983. Asheville, N.C.: National Climatic Data Center, 1984.

Karl, Thomas R., George Kukla, and Joyce Gavin. "Relationship between Decreased Temperature Range and Precipitation Trends in the United States and Canada, 1941–80."Journal of Climate and Applied Meteorology 25.12 (1986): 1878-886.

Karl, Thomas R., George Kukla, Vyacheslav N. Razuvayev, Michael J. Changery, Robert G. Quayle, Richard R. Heim, David R. Easterling, and Cong Bin Fu. "Global Warming: Evidence for Asymmetric Diurnal Temperature Change." Geophysical Research Letters 18.12 (1991) 2253-2256.

Karl, Thomas R., Richard W. Knight, Kevin P. Gallo, Thomas C. Peterson, Philip D. Jones, George Kukla, Neil Plummer, Vyacheslav Razuvayev, Janette Lindseay, and Robert J. Charlson. "A New Perspective on Recent Global Warming: Asymmetric Trends of Daily Maximum and Minimum Temperature." Bulletin of the American Meteorological Society 74.6 (1993): 1007-023.

Karl, Thomas R., and Richard W. Knight. "Secular Trends of Precipitation Amount, Frequency, and Intensity in the United States." Bulletin of the American Meteorological Society 79.2 (1998): 231-41.

Kunkel, Kenneth E., Karen Andsager, and David R. Easterling. "Long-Term Trends in Extreme Precipitation Events over the Conterminous United States and Canada." Journal of Climate 12.8 (1999): 2515-527.

Lauritsen, Ryan G., and Jeffrey C. Rogers. "U.S. Diurnal Temperature Range Variability and Regional Causal Mechanisms, 1901–2002." Journal of Climate 25.20 (2012): 7216-231.

Mallya, G., L. Zhao, X. C. Song, D. Niyogi, and R. S. Govindaraju. "2012 Midwest Drought in the United States." Journal of Hydrologic Engineering 18.7 (2013): 737-45.

Mishra, Vimal, Keith A. Cherkauer, and Shraddhanand Shukla. "Assessment of Drought Due to Historic Climate Variability and Projected Future Climate Change in the ." Journal of Hydrometeorology 11.1 (2010): 46-68.

101

National Assessment Synthesis Team, Climate Change Impacts on the United States: The Potential Consequences of Climate Variability and Change, US Global Change Research Program, 400 Virginia Avenue, SW, Suite 750, Washington DC, 20024

Peterson, Thomas C., Richard R. Heim, Robert Hirsch, Dale P. Kaiser, Harold Brooks, Noah S. Diffenbaugh, Randall M. Dole, Jason P. Giovannettone, Kristen Guirguis, Thomas R. Karl, Richard W. Katz, Kenneth Kunkel, Dennis Lettenmaier, Gregory J. Mccabe, Christopher J. Paciorek, Karen R. Ryberg, Siegfried Schubert, Viviane B. S. Silva, Brooke C. Stewart, Aldo V. Vecchia, Gabriele Villarini, Russell S. Vose, John Walsh, Michael Wehner, David Wolock, Klaus Wolter, Connie A. Woodhouse, and Donald Wuebbles. “Monitoring and Understanding Changes in Heat Waves, Cold Waves, Floods, and Droughts in the United States: State of Knowledge.” Bulletin of the American Meteorological Society (2013): 821-34.

Rahmstorf, S., and D. Coumou. "Increase of Extreme Events in a Warming World."Proceedings of the National Academy of Sciences 108.44 (2011): 17905- 7909.

Rogers, Jeffrey C. "The 20th Century Cooling Trend over the Southeastern United States." Climate Dynamics 40.1-2 (2013) 341-52.

Ruddiman, William F. "Orbital Changes and Climate." Quaternary Science Reviews 25 (2006): 3092-112.

Seager, Richard, Alexandrina Tzanova, and Jennifer Nakamura. "Drought In The Southeastern United States: Causes, Variability Over The Last Millennium, And The Potential For Future Hydroclimate Change*." Journal of Climate 22.19 (2009): 5021-045.

Shackleton, Robert. Potential Impacts of Climate Change in the United States. Washington, D.C.: Congress of the U.S., Congressional Budget Office :, 2009.

Smith, David G. "Milankovitch Cyclicity and the Stratigraphic Record-a Review." Terra Nova 1.5 (1989): 402-04.

The Great Flood of 1993. Washington, D.C.?: U.S. Dept. of Commerce, National Oceanic and Atmospheric Administration, National Weather Service, 1994.

Wahl, Kenneth A., Kevin C. Vining, and Gregg J. Wiche. "Precipitation in the Upper Mississippi River Basin, January 1 through July 31, 1993." Floods in the Upper Mississippi River Basin, 1993. Washington, DC: United States Government Printing Office, 1993. 102

Woodhouse, Connie A., and Jonathan T. Overpeck. "2000 Years of Drought Variability in the Central United States." Bulletin of the American Meteorological Society 79.12 (1998): 2693-714.

Worster, Donald, 1979: Dust Bowl: The Southern Plains in the 1930s. : Oxford University Press. 283pp.

Zhang, Y-C., W. B. Rossow, and A. A. Lacis. "Calculation of Surface and Top of Atmosphere Radiative Fluxes from Physical Quantities Based on ISCCP Data Sets 1. Method and Sensitivity to Input Data Uncertainties." Journal of Geophysical Research 100.D1 (1995): 1149-165.

103