A Cross-Sectional Analysis of Environmental Radiation Levels
Across Four U.S. Locations and Relevance to Radiation Exposure
During Long-Duration Spaceflight
An exploratory epidemiological-style analysis examining geographic
variation in environmental radiation measurements and what these
patterns can — and cannot — tell us about radiation-related cancer
risk during future space missions.
Elizabeth Khimey
NYC Lab High School for Collaborative Studies
From a Broad Question to an Exploratory Epidemiological Analysis
This project began with a broader question about the relationship
between radiation exposure and cancer risk during long-duration
spaceflight.
My original goal was to explore this question using evidence from
astronauts and other exposed populations. However, I encountered an
important limitation: the number of astronauts with long-duration
deep-space radiation exposure is very small.
This creates a challenge for epidemiological research. When a sample
is very small, it becomes difficult to identify reliable patterns,
estimate risk with confidence, or determine whether an observed
relationship is meaningful.
Research pivot:
Because the astronaut population was too small for the type of
epidemiological analysis I wanted to explore, I changed the scope
of the project. I conducted an exploratory epidemiological-style
analysis using environmental radiation-monitoring data from four
U.S. locations.
What the Analysis Allowed Me to Investigate
The environmental dataset provided thousands of radiation measurements,
giving me a larger dataset in which to examine geographic differences,
distributions, temporal patterns, and statistical relationships.
This approach does not directly measure cancer risk in astronauts.
Instead, it allowed me to investigate radiation measurements using
statistical and epidemiological concepts while recognizing the
limitations of the available evidence.
Broad Research Question
→
Small Astronaut Sample
→
Research Limitation
→
Environmental Dataset
→
Exploratory Analysis
01 · Overview
The Research Question
As scientists develop plans for longer missions to the Moon and Mars,
radiation exposure remains one of the major health concerns of deep-space
exploration.
Central question:
How does radiation exposure relate to cancer risk, and what can
environmental radiation measurements tell us about radiation
exposure relevant to long-duration spaceflight?
Ionizing radiation can damage DNA, and accumulated or incorrectly repaired
DNA damage can contribute to cancer. However, directly studying cancer risk
in astronauts is difficult because relatively few people have experienced
long-duration deep-space missions.
This project therefore combines background research on radiation and cancer
with an exploratory analysis of 2025 environmental radiation-monitoring data
from four U.S. locations. The environmental analysis was chosen as an
alternative approach after recognizing that the small number of astronauts
with long-duration deep-space exposure makes direct epidemiological analysis
of astronaut cancer outcomes difficult.
02 · Radiation Biology
Why Does Radiation Matter?
Ionizing radiation has enough energy to remove electrons from atoms and can
damage biological molecules, including DNA. Cells have repair systems, but
repair is not always perfect.
Radiation Exposure
→
DNA Damage
→
Cellular Repair
→
Mutation
→
Potential Cancer
DNA damage does not automatically cause cancer. Cells can repair damage,
and heavily damaged cells may die. However, mutations affecting genes that
control cell growth and division can sometimes contribute to cancer.
03 · Space Radiation
Where Does Space Radiation Come From?
Galactic Cosmic Rays
Extremely energetic particles originating outside our solar system.
They are continuously present and can be difficult to shield against.
Solar Particle Events
Bursts of energetic particles released by the Sun that can temporarily
increase radiation levels around a spacecraft.
Trapped Radiation
Charged particles trapped by Earth's magnetic field in regions known
as the Van Allen radiation belts.
Earth's Protection
Earth's atmosphere and magnetic field provide important protection
from parts of the space radiation environment.
04 · Cancer Risk
Does More Radiation Mean More Cancer?
Scientists have strong evidence that sufficiently high doses of ionizing
radiation can increase cancer risk. Evidence comes from populations including
atomic-bomb survivors, radiation-exposed workers, patients receiving certain
radiation treatments, and other exposed groups.
Understanding the risks associated with relatively low doses is more difficult.
Cancer is influenced by many factors, including age, genetics, smoking,
diet, ultraviolet radiation, occupation, lifestyle, and other environmental
exposures.
The Linear No-Threshold Model
The Linear No-Threshold model is one approach used to estimate radiation-related
cancer risk. Under this model, risk is assumed to increase approximately
linearly with radiation dose, with no dose considered completely risk-free.
However, the LNT model is a risk-estimation model rather than a direct
prediction of what will happen to every individual.
05 · Previous Evidence
Astronauts and Radiation Workers
Astronaut Studies
Researchers have studied early NASA astronauts to investigate whether estimated
space-radiation exposure was associated with cancer mortality. These studies
are valuable but limited by the small number of astronauts and relatively few
cancer deaths.
INWORKS
The International Nuclear Workers Study included approximately 309,932 workers
from France, the United Kingdom, and the United States. Researchers found that
mortality from solid cancer increased as cumulative radiation dose increased,
although uncertainty was greater at relatively low doses.
Important distinction:
Occupational radiation exposure is not identical to deep-space radiation.
Space radiation includes galactic cosmic rays and solar particle events
with different physical characteristics.
06 · Environmental Study
2025 Environmental Radiation Analysis
Why This Is an Epidemiological-Style Analysis
Epidemiology investigates patterns of health-related exposures and
outcomes across populations. In this project, environmental radiation
measurements represent the exposure-related variable being investigated,
while location serves as the population-level comparison.
Because this dataset does not contain individual cancer diagnoses,
mortality, medical histories, or individual radiation exposure, this
study cannot estimate cancer risk directly. Instead, it represents
a small exploratory analysis of environmental radiation exposure patterns.
Data Quality and Observation Periods
Before comparing radiation levels between locations, I evaluated the size,
completeness, and observation periods of each dataset. I examined the number
of observations, missing radiation measurements, missing timestamps, negative
measurements, and the dates and times covered by each monitoring station.
Figure 1. Number of radiation observations included
in each location's 2025 dataset. The datasets contain different
numbers of observations, which should be considered when comparing
locations.
Figure 2. Assessment of missing dose-equivalent rate
measurements across the four study locations. The datasets contained
no missing radiation measurements.
Figure 3. Assessment of missing timestamps across the
four study locations. No missing timestamps were identified in the
datasets used for analysis.
Figure 4. Assessment of negative radiation measurements
across the four study locations. No negative dose-equivalent rate
measurements were identified.
Figure 5. Observation periods for the four monitoring
locations, showing the beginning and ending dates and times of the
available measurements. Edison began monitoring in March 2025, while
Phoenix, Miami, and Seattle began in January 2025.
The monitoring periods were not identical across all four locations.
Phoenix, Miami, and Seattle had observations beginning in January 2025,
while Edison began in March 2025. All four datasets extended through
December 31, 2025. This difference is an important limitation because
Edison contains fewer months of observations.
Descriptive Results
After evaluating the datasets, I examined the distribution of environmental
radiation measurements at each location. The primary descriptive measure
was the mean dose-equivalent rate, measured in nanosieverts per hour (nSv/h).
Figure 6. Mean environmental dose-equivalent rate by
location. Phoenix had the highest mean measurement at 85.21 nSv/h,
followed by Miami at 47.84 nSv/h, Edison at 40.81 nSv/h, and Seattle
at 27.07 nSv/h.
Figure 7. Descriptive statistics for environmental
radiation measurements by location, including mean, standard deviation,
median, interquartile range, minimum, maximum, and upper percentiles.
The statistics show differences in the typical values and ranges of
radiation measurements among the four locations.
Figure 8. Distribution of environmental radiation
measurements across the four locations. The box plot shows differences
in the central values, spread, and ranges of the measurements, with
Phoenix generally showing higher measurements than the other locations.
Phoenix had the highest mean dose-equivalent rate at 85.21 nSv/h,
followed by Miami at 47.84 nSv/h, Edison at 40.81 nSv/h, and
Seattle at 27.07 nSv/h. The distributions also differed in their
ranges and overall spread.
Why might radiation levels differ?
Background radiation is not exactly the same everywhere. Possible factors
include altitude, geology, naturally radioactive materials in soil and
rocks, atmospheric conditions, and geography. However, the dataset alone
cannot determine exactly why one location had higher measurements than
another.
Phoenix
85.21 nSv/h
Miami
47.84 nSv/h
Edison
40.81 nSv/h
Seattle
27.07 nSv/h
85.21
nSv/h · Phoenix, Arizona
47.84
nSv/h · Miami, Florida
40.81
nSv/h · Edison, New Jersey
27.07
nSv/h · Seattle, Washington
Location
Mean Dose-Equivalent Rate
Phoenix, Arizona
85.21 nSv/h
Miami, Florida
47.84 nSv/h
Edison, New Jersey
40.81 nSv/h
Seattle, Washington
27.07 nSv/h
Methods
The study was designed as a cross-sectional, exploratory comparison of
environmental radiation measurements across four U.S. locations. The
primary exposure-related variable was dose-equivalent rate, while
geographic location was the primary comparison variable.
Because the study used environmental monitoring data rather than
individual-level health records, the analysis focuses on differences
in measured radiation rather than individual cancer outcomes.
The primary variable was dose-equivalent rate, measured in nanosieverts per
hour (nSv/h). Descriptive statistics included the mean, standard deviation,
median, interquartile range, minimum, maximum, and upper percentiles.
A Kruskal-Wallis test was used to compare radiation measurements among the
four locations, followed by pairwise Mann-Whitney U tests with Bonferroni
correction.
07 · Results
Statistical Results
Statistical tests were used to determine whether the observed radiation
measurements differed significantly among the four study locations.
Figure 9. Kruskal-Wallis test comparing dose-equivalent
rate distributions among Phoenix, Miami, Edison, and Seattle, with
the Miami zero measurements included. The test produced a large H
statistic and p < 0.001, providing strong statistical evidence that
the four distributions were not all the same.
Figure 10. Pairwise Mann-Whitney U tests comparing
dose-equivalent rate distributions between each pair of locations.
The extremely small p-values provide strong evidence of differences
between the location pairs.
Figure 11. Pairwise Mann-Whitney U comparisons with
Bonferroni-adjusted p-values. All six location comparisons remained
statistically significant after correction for multiple comparisons.
Figure 12. Summary of the Kruskal-Wallis analysis,
including the H statistic, total sample size, number of groups,
and epsilon-squared effect size. The epsilon-squared value of
approximately 0.928 indicates a very large difference among the
distributions in this environmental dataset.
28,315.01Kruskal-Wallis H statistic
p < 0.001Kruskal-Wallis significance
≈ 0.928Epsilon-squared effect size
Kruskal-Wallis Test
The Kruskal-Wallis test provided strong statistical evidence that the
distributions of radiation measurements were not the same across all
four locations.
The test provides evidence that at least one location had a different
pattern of radiation measurements from the others.
Pairwise Mann-Whitney U Tests
Pairwise Mann-Whitney U tests were then performed between all six pairs
of locations. These tests compare locations two at a time and are useful
when the data may not follow a normal distribution.
All pairwise comparisons produced extremely small p-values, and all
remained statistically significant after Bonferroni correction.
How should the p-value be interpreted?
An extremely small p-value does not mean that the true p-value is
literally zero. It indicates that the observed differences would
be very unlikely under the assumption that the groups had the same
distributions.
Statistical Significance vs. Practical Significance
The extremely small p-values are partly related to the very large number
of observations. With thousands of measurements per location, even
relatively small differences can produce very small p-values.
These findings support an association between geographic location and the
distribution of measured environmental radiation in this dataset. They
should not be interpreted as evidence that geographic location itself
causes differences in radiation levels, or that the differences directly
translate into differences in cancer risk.
Bonferroni Correction
Because six pairwise statistical tests were performed, there was a greater
chance of obtaining a statistically significant result simply by chance.
The Bonferroni correction makes the significance requirement stricter to
account for these multiple comparisons.
Epsilon-Squared Effect Size
Because statistical significance does not explain the magnitude of the
difference, I calculated epsilon-squared as an effect-size measure for
the Kruskal-Wallis test.
The estimated epsilon-squared effect size was approximately 0.928,
indicating a very large difference in the distributions of measurements
among the locations.
This suggests that the location groups were very different within this
environmental radiation dataset.
Important interpretation:
The value of 0.928 does not mean that 92.8% of cancer
risk is caused by location or radiation. The effect size describes
the differences observed in this environmental dataset.
08 · Data Investigation
The Miami Data Mystery
Figure 13. Descriptive statistics for the Miami
radiation measurements. The minimum value of 0 nSv/h was substantially
lower than the other measurements and prompted further investigation
of the Miami dataset.
Figure 14. Number and percentage of zero radiation
measurements identified in the Miami dataset. Sixteen zero measurements
were identified among 7,349 observations, representing approximately
0.218% of the Miami dataset.
Figure 15. Comparison of the Miami mean dose-equivalent
rate with zero measurements included and excluded. Removing the 16 zero
measurements changed the mean by approximately 0.10 nSv/h, indicating
little effect on the overall Miami average.
Miami contained 16 measurements of 0 nSv/h out of 7,349 observations.
These represented approximately 0.218% of the Miami dataset.
16zero measurements
7,349total Miami observations
0.218%of Miami observations
While analyzing the data, Miami had a minimum measurement of 0 nSv/h.
This was unusual because nearly all of the surrounding measurements were
around 48 nSv/h.
Instead of immediately removing these measurements, I tested how much
they affected the results.
Removing the zero measurements changed the Miami average by approximately
0.10 nSv/h. Therefore, removing the zero measurements had very little
effect on the overall Miami average.
The dataset alone cannot determine whether these zero values represent
missing measurements, invalid measurements, a detector problem, or a
data-recording issue.
Importantly, the Kruskal-Wallis result also remained highly significant
when the Miami zero measurements were removed.
09 · Patterns
Monthly, Hourly & Gamma Measurements
Figure 16. Monthly variation in environmental radiation
measurements across the four study locations during 2025. Radiation
measurements varied throughout the year, with the magnitude and pattern
of variation differing among locations.
Figure 17. Monthly dose-equivalent rates by location.
Phoenix showed monthly averages ranging from approximately 82.74 to
89 nSv/h, while Seattle ranged from approximately 25.97 to 28.15 nSv/h.
The differences demonstrate that radiation measurements varied across
both location and month.
Figure 18. Mean environmental dose-equivalent rate by
hour of the day across the four study locations. The hourly patterns
show variation in measurements throughout the day but do not establish
that a specific environmental factor caused those differences.
Figure 19. Dose-equivalent rate and gamma count-rate
measurements across the four study locations. The figure shows how
the measured dose-equivalent and gamma count-rate values differed
among locations, providing a comparison of the radiation measurements
recorded by the monitoring detectors.
Figure 20. Correlations between dose-equivalent rate
and gamma count-rate measurements in Edison. Several gamma count rates
showed strong positive relationships with dose-equivalent rate, including
R05 (r = 0.789), R06 (r = 0.784), and R07 (r = 0.791). These correlations
indicate that the measurements tended to vary together but do not
establish causation.
Figure 21. Descriptive statistics and upper-percentile
measurements for the four study locations, including the 95th and 99th
percentiles. Phoenix had the highest typical and upper-percentile
measurements, while Phoenix and Edison also showed maximum values that
were substantially higher than their typical measurements.
Monthly Variation
I also examined the average radiation measurement by month. The analysis
showed that radiation levels varied over the year, but the pattern differed
among locations.
Phoenix monthly averages ranged from approximately 82.74 to 89 nSv/h,
while Seattle ranged from approximately 25.97 to 28.15 nSv/h.
This shows that environmental radiation is not exactly the same throughout
the year. However, these monthly patterns alone cannot tell us what
environmental factor caused the changes.
Hourly Variation
I also examined the average radiation measurement by hour of the day.
This was done to identify possible patterns in radiation levels throughout
the day.
However, since environmental conditions, monitoring stations, and sampling
patterns can be different at each location, these hourly differences should
only be viewed as patterns in the data. They do not prove that a specific
environmental factor caused the changes.
Gamma Count Rates
The datasets also included gamma count-rate measurements from the detector
variables. I examined the relationship between dose-equivalent rate and
gamma count rates.
For example, Edison showed fairly strong positive relationships with
several gamma count-rate measurements, including R05 (r = 0.789),
R06 (r = 0.784), and R07 (r = 0.791).
Seattle also showed positive relationships with most of the gamma
count-rate measurements.
These results suggest that dose-equivalent rate and gamma count rates
often increased or decreased together. However, correlation does not
prove that one measurement caused the other to change.
Differences in correlation between locations also suggest that detector
measurements may provide additional information that could be studied
in future research.
Upper-Percentile Measurements
I also examined higher-percentile measurements to identify the upper
portion of each location's distribution.
These results show that Phoenix had the highest typical radiation
measurements as well as the highest measurements in the upper portion
of the data. Edison and Phoenix also had maximum measurements that
were much higher than their usual measurements.
These higher measurements are worth investigating, but one high measurement
does not necessarily mean that the exposure was dangerous or that it
increased cancer risk.
Factors such as how long the exposure lasted, the accuracy of the
measurement, and the total amount of radiation exposure would need
to be considered.
10 · Spaceflight
Why Does This Matter for Mars?
A future Mars mission could expose astronauts to the deep-space radiation
environment for months during the journey to Mars, during time on the
Martian surface, and during the return trip.
Shielding
Spacecraft materials can reduce radiation exposure, but additional
shielding increases spacecraft mass.
Water
Hydrogen-rich materials such as water can potentially be positioned
around crew areas as part of radiation protection.
Supplies
Food and other materials already required for a mission could also
contribute to shielding around crew living areas.
Radiation Shelters
Specially protected areas could provide additional protection during
intense solar particle events.
11 · Limitations
What This Study Cannot Tell Us
The environmental analysis demonstrates geographic differences in measured
radiation. It does not directly measure cancer outcomes or individual
radiation exposure.
01 · Astronaut Population
There are relatively few astronauts with long-duration deep-space
exposure.
02 · Radiation Type
Occupational radiation exposure is not identical to deep-space
radiation.
03 · Environmental Data
Monitoring stations measure environmental conditions rather than
individual human exposure.
04 · Other Risk Factors
Cancer is influenced by age, genetics, smoking, UV radiation,
lifestyle, occupation, and other factors.
05 · Long Latency
Cancer can develop many years after radiation exposure, requiring
long-term follow-up.
06 · Unequal Observation Periods
Edison began in March 2025 while the other three locations began
in January 2025.
07 · No Cancer Outcomes
The environmental dataset does not contain cancer diagnoses,
mortality, or individual medical information.
References
References
U.S. Environmental Protection Agency.
Why Do I See Higher Levels of Radiation at Some Monitor Locations?
EPA RadNet
NASA.
Space Radiation Is Risky Business for the Human Body.
NASA
Centers for Disease Control and Prevention.
Health Effects of Radiation.
CDC
PMC Article: Radiation Exposure and Cancer Risk.
PubMed Central.
PubMed Central
PMC Article: Radiation, Cancer Risk, and Epidemiological Evidence.
PubMed Central.
PubMed Central
American Association for Cancer Research.
Cancer Risk Estimates: What Do Those Numbers Mean?
AACR
The evidence provides scientific reason to be concerned about radiation
exposure while also showing why the exact cancer risk from deep-space
radiation remains difficult to determine.
One of the most important outcomes of this project was methodological.
I began with the goal of exploring radiation and cancer risk in the
context of astronauts, but the limited number of astronauts with
long-duration deep-space exposure made a direct epidemiological analysis
difficult. Rather than treating a small sample as sufficient evidence, I
changed the approach and developed an exploratory analysis using
environmental radiation data.
This environmental study found substantial differences in measured
radiation levels among four U.S. locations during 2025. Phoenix had
the highest average dose-equivalent rate, while Seattle had the lowest.
The statistical analysis also provided strong evidence that the distributions
of measured environmental radiation differed among the four locations.
Pairwise comparisons remained statistically significant after Bonferroni
correction.
However, these results do not demonstrate that people living in locations
with higher environmental measurements have higher cancer rates. The
dataset does not contain individual radiation exposure, cancer outcomes,
age, sex, smoking history, occupation, medical history, or long-term
follow-up.
The bigger question:
How can scientists better estimate and reduce the long-term health
risks of radiation during missions beyond Earth's protective
magnetic field?
The findings of this project suggest that radiation measurements can vary
substantially across environments, but environmental measurements on Earth
cannot be directly substituted for measurements of deep-space radiation
exposure.
Future research will require better astronaut health datasets, improved
models of space-radiation exposure, and continued investigation of how
different types of radiation affect human biology.
Ultimately, protecting astronauts from radiation will require both better
evidence about long-term health effects and practical strategies for
reducing exposure during missions to the Moon, Mars, and beyond.
Final Takeaway
This study does not show that higher environmental radiation causes
higher cancer risk. Instead, it demonstrates how epidemiological and
statistical methods can be used to investigate radiation patterns
while carefully distinguishing statistical evidence from conclusions
about human health.
