Access to this full-text is provided by Springer Nature.
Content available from Scientific Reports
This content is subject to copyright. Terms and conditions apply.
1
Vol.:(0123456789)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports
Investigating the eects
of absolute humidity
and movement on COVID‑19
seasonality in the United States
Gary Lin1*, Alisa Hamilton1, Oliver Gatalo1, Fardad Haghpanah1, Takeru Igusa2,3,4 &
Eili Klein1,5,6
Mounting evidence suggests the primary mode of SARS‑CoV‑2 transmission is aerosolized
transmission from close contact with infected individuals. While transmission is a direct result of
human encounters, falling humidity may enhance aerosolized transmission risks similar to other
respiratory viruses (e.g., inuenza). Using Google COVID‑19 Community Mobility Reports, we assessed
the relative eects of absolute humidity and changes in individual movement patterns on daily cases
while accounting for regional dierences in climatological regimes. Our results indicate that increasing
humidity was associated with declining cases in the spring and summer of 2020, while decreasing
humidity and increase in residential mobility during winter months likely caused increases in COVID‑19
cases. The eects of humidity were generally greater in regions with lower humidity levels. Given the
possibility that COVID‑19 will be endemic, understanding the behavioral and environmental drivers
of COVID‑19 seasonality in the United States will be paramount as policymakers, healthcare systems,
and researchers forecast and plan accordingly.
As of October 14, 2021, the coronavirus disease 2019 (COVID-19) pandemic has claimed over 720,000 lives in
the United States alone, with more than 44.7 million conrmed cases1. Current evidence suggests that the primary
mode of transmission of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is close contact with
infected individuals2,3. Aerosols4,5, which are particulates less than 5µm in diameter6,7, likely play a signicant
role in transmission8. Aer the initial rise of cases in the early winter of 2020, cases remained severe through the
spring before dropping in the summer. Given the shelter-in-place order in most states and the rise in humidity,
cases generally decreased in May and stayed in lower ranges through the summer until the fall months. In most
areas of the northern hemisphere, as fall turns to winter, the weather becomes colder and drier. Lower absolute
humidity has been shown to be associated with increased transmission rates of other respiratory viruses (e.g.,
inuenza)9, posing signicant concerns regarding potential increases in the number of COVID-19 cases in the
fall and winter. e surge in cases through the end of 2020 further supports the seasonal eects of COVID-19.
While several studies have suggested a relationship between climatic factors (e.g., temperature and/or humid-
ity) and COVID-1910–18, the exact environmental and biological mechanism behind airborne and droplet trans-
mission and viral survival of SARS-CoV-219 is not yet clear. In inuenza, lower atmospheric moisture has been
shown to increase the production of aerosol nuclei and viral survival time9, which translates to higher risks of
airborne and droplet transmission. Other climatic factors that may impact transmission include temperature
and air quality20,21; nevertheless, absolute humidity can still provide a surrogate measure for indoor air moisture
and temperature22.
Initial eorts to slow the spread of COVID-19 focused on reducing contacts between individuals through
social-distancing measures such as large-scale lockdowns, which were signicantly associated with reductions
in cases23. However, as the initial lockdowns were lied and the movement of individuals increased, the correla-
tion between mobility and case growth rates weakened overall24, though upticks in cases were associated with
OPEN
1Center for Disease Dynamics, Economics & Policy, 962 Wayne Avenue, Suite 530, Silver Spring, MD 20910-4433,
USA. 2Department of Civil and Systems Engineering, Johns Hopkins University, Baltimore, MD, USA. 3Department
of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, MD, USA. 4Center for Systems Science
and Engineering, Johns Hopkins University, Baltimore, MD, USA. 5Department of Emergency Medicine, Johns
Hopkins University, Baltimore, MD, USA. 6Department of Epidemiology, Johns Hopkins University, Baltimore, MD,
USA. *email: lin@CDDEP.org
Content courtesy of Springer Nature, terms of use apply. Rights reserved
2
Vol:.(1234567890)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
increased mobility during national holidays25. During the months of 2020 and 2021 some counties and states saw
increases in cases, while others observed decreases without corresponding increases in movement by any metric.
us, other factors, including environmental factors, must also be considered as important transmission drivers.
Analyses of the factors inuencing COVID-19 have used either climate data21,26–28 or human mobility data23,
but no study to our knowledge has considered changes in both climate and human mobility on COVID-19
outbreaks in the United States. Preliminary studies have investigated these eects in China but did not consider
varying sensitivities to humidity for dierent climatological regimes, leading to a weaker detection of humidity
impacts on transmission risks in areas with higher variations of humidity29. Understanding the potential for
climatic factors to increase transmission in the fall and winter is crucial for developing policies to combat the
spread of SARS-CoV-2. While the interaction between environmental factors and human encounters is complex,
accounting for this relationship is necessary for determining appropriate policies that will be eective at reduc-
ing transmissions. Furthermore, indoor gatherings typically increase in frequency and size in the winter and
are one of the largest risk factors for transmission7,30. erefore, greater understanding regarding the added risk
of weather changes is needed to aid future decisions on restricting gatherings or implementing mandates for
protective face coverings. In this study, we assessed the relative impact of absolute humidity and human mobility
in dierent climatological regimes on reported cases of COVID-19 in the US.
Results
Partitioning climatological regimes. e US is geographically large and encompasses several dier-
ent climatological regimes with varying absolute humidity trends. We partitioned all 3137 US counties into six
exclusive clusters (Fig.1) ranked by average absolute humidity (AH) using a dynamic time warping (DTW)
algorithm which considers both magnitude and functional trends of AH (see “Methods”). e cluster with the
lowest average AH was primarily located in the western region of the US, while the region with the highest aver-
age AH was located on the southern coast bordering the Gulf of Mexico. Large changes of humidity were seen in
clusters High 1 and High 2 which, respectively, includes variances of 26.9 and 30.6g/m3 (see Fig.S1), while Low
1 and Low 2 humidity clusters had a variance of 4.5 and 14.2g/m3.
Figure1. (A) Map of US Counties and their respective absolute humidity clusters. Each county is colored
based on their cluster. Counties that are included in the regression analysis are indicated by a darker shade. e
clustering analysis was conducted using a partitional algorithm that utilized dynamic time warping (DTW) to
measure similarity between absolute humidity proles of 3137 counties in the United States. Expectantly, the
clustering of absolute humidity is related to the geography of the counties which serves as a proxy for regional
weather patterns and dierent climatological regimes. (B) e cross-sectional smoothed mean of human
encounter absolute humidity, and new case per 10,000 people trends for each cluster group of the 497 counties
analyzed in the regression analysis. Map was generated using the ggplot package31 in R.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
3
Vol.:(0123456789)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
Associations between humidity and cases rates. We conducted a regression on counties with more
than 50,000 people using a generalized linear model (GLM) and controlling for individual movement and
behavior with a metric from mobile phone data of visits to non-essential businesses (see Methods), we found
that increases in AH were signicantly negatively associated with cases per 100,000 of COVID-19 in all the non-
high humidity regions (Table1). We found that counties that belong to the least humid clusters, Low 1 and Low
2, had a 1g/m3 increase in AH was associated with an average decrease of 14 percent reduction in cases over
the entire duration, while the most humid clusters (High 1 and High 2) had a decrease of 4 percent in cases. e
largest associations were seen in counties predominantly in the Rocky Mountains (Low 1; 20% decrease in daily
cases), Upper Midwest/Northwest (Mid 1; 12% decrease in daily cases), West Coast/Texas/Northeast (Mid 2;
16% decrease in daily cases), and a region stretching along the western edge of the Midwest down to Texas (Low
2; 8% decrease in daily cases). Small but signicant eects were detected in two high humidity clusters, both
located in the southern region of the US (High 1 and High 2), with respective reductions of 6% and 1% in daily
cases with a 1g/m3 increase in AH.
e overall associations between AH and COVID-19 cases were negatively correlated when disaggregated
across the time periods (Tables2 and 3). e regression showed that AH had strong associations in the Mid
2 cluster, located in West Coast/Texas/Northeast, during the spring and summer months of 2020 (Table2). In
the fall of 2020 and spring of 2021, AH associations were generally stronger in counties from Mid 2 and High 1
clusters, which are in the West Coast, Texas, Northeast and Southern regions of the US (Table3).
Associations between movement and case rates. In general, movement eects on daily cases are
larger than absolute humidity eects, with visits to retail and recreation positively associated with new COVID-
19 cases in most of the clusters (Table1). Mobility trends for retail & recreation and grocery stores & pharmacies
had a larger positive eect during the earlier phase of the pandemic for most clusters (March 10 to September
30, 2020) compared to the later phase spanning from October 1, 2020 to March 1, 2021. e residential mobil-
ity trend was associated with a decrease in new cases in most clusters during the earlier phase of the pandemic
(Table2), while having a positive eect on daily cases during the later phase (Table3).
Detecting multicollinearity between movement and absolute humidity. To understand the col-
linearity of the combined regressions shown in Tables1, 2 and 3, we conducted robustness checks with addi-
tional regressions that included the AH and the mobility trends separately (See TablesS1–S18). Additionally,
we calculated the Generalized Variational Ination Factor (GVIF) for the regressions in our robustness checks.
Workplaces and Residential Mobility Trends were the least collinear with other independent variables (abso-
lute humidity, immunity factor, and previous 14-day caseload) supported by GVIF values less than 2. Mobility
trends in Retail and Recreation Areas and Grocery Stores and Pharmacies were mostly non-collinear with few
exceptions with GVIF values ranging between with a mean of 1.53 (range: 1.15–2.30) and 1.65 (1.28–2.63). And
nally, Transit Stations and Parks demonstrated the most collinearity with mean GVIF values of 2.15 (1.45–3.71)
and 2.01 (1.56–2.83).
Table 1. Untransformed GLM coecient estimates for the entire study period. Untransformed coecient
(β) estimates for GLM Regression against new cases per 100,000 from March 10, 2020 to March 1, 2021. e
95% condence intervals are shown in parenthesis. Estimated coecients for county-level xed eects and
epidemiological terms (immunity factor and lagged daily cases) are not shown. *p < 0.05 **p < 0.01 ***p < 0.001.
Predictors
Low 1 Low 2 Mid 1 Mid 2 High 1 High 2
Log-Mean Log-Mean Log-Mean Log-Mean Log-Mean Log-Mean
Intercept 4.379***
(4.364–4.395) 3.439***
(3.423–3.455) 3.735***
(3.730–3.740) 3.885***
(3.876–3.894) 4.381***
(4.291–4.469) 3.270***
(3.254–3.285)
Absolute Humid-
ity (14-day Lag)
− 0.221***
(− 0.223
to − 0.219)
− 0.084***
(− 0.085
to − 0.082)
− 0.123***
(− 0.124
to − 0.123)
− 0.171***
(− 0.171
to − 0.170)
− 0.060***
(− 0.060
to − 0.059)
− 0.015***
(− 0.015
to − 0.015)
Retail and Recrea-
tion (14-day Lag) 0.826***
(0.815–0.837) 0.839***
(0.829–0.850) 0.925***
(0.920–0.930) 0.515***
(0.511–0.519) 0.950***
(0.941–0.959) 1.299***
(1.293–1.305)
Grocery Stores
and Pharmacies
(14-day Lag)
− 0.354***
(− 0.361
to − 0.348)
− 0.145***
(− 0.152
to − 0.138)
0.040***
(0.037–0.042) 0.171***
(0.169–0.174)
− 0.223***
(− 0.228
to − 0.217)
− 0.130***
(− 0.134
to − 0.126)
Parks (14-day
Lag)
− 0.536***
(− 0.543
to − 0.530)
− 0.123***
(− 0.128
to − 0.118)
− 0.156***
(− 0.159
to − 0.154)
− 0.200***
(− 0.202
to − 0.198)
− 0.379***
(− 0.383
to − 0.374)
− 0.984***
(− 0.988
to − 0.979)
Transit Stations
(14-day Lag)
− 0.134***
(− 0.143
to − 0.125)
− 0.519***
(− 0.528
to − 0.511)
− 0.762***
(− 0.766
to − 0.758)
− 0.602***
(− 0.607
to − 0.598)
− 0.350***
(− 0.356
to − 0.344)
− 0.339***
(− 0.343
to − 0.335)
Workplaces (14-
day Lag)
− 0.592***
(− 0.599
to − 0.585)
− 0.560***
(− 0.569
to − 0.552)
− 0.386***
(− 0.390
to − 0.383)
− 0.683***
(− 0.686
to − 0.680)
− 0.762***
(− 0.767
to − 0.757)
− 0.541***
(− 0.544
to − 0.538)
Residential (14-
day Lag)
− 0.601***
(− 0.611
to − 0.591)
− 0.425***
(− 0.437
to − 0.413)
− 0.166***
(− 0.171
to − 0.161)
− 0.576***
(− 0.580
to − 0.572)
− 0.583***
(− 0.591
to − 0.575)
− 0.269***
(− 0.273
to − 0.265)
Observations 9557 7987 25,568 27,087 16,581 25,916
Content courtesy of Springer Nature, terms of use apply. Rights reserved
4
Vol:.(1234567890)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
Table 2. Untransformed GLM coecient estimates for the 2020 spring to fall period. Untransformed
coecient (β) estimates for GLM Regression against new cases per 100,000 from March 10, 2020 to September
30, 2021. e 95% condence intervals are shown in parenthesis. Estimated coecients for county-level xed
eects and epidemiological terms (immunity factor and lagged daily cases) are not shown. *p < 0.05 **p < 0.01
***p < 0.001.
Predictors
Low 1 Low 2 Mid 1 Mid 2 High 1 High 2
Log-Mean Log-Mean Log-Mean Log-Mean Log-Mean Log-Mean
Intercept 2.064***
(1.982–2.145) 2.198***
(2.159–2.236) 3.064***
(3.053–3.074) 3.033***
(3.014–3.053) 1.020***
(0.952–1.087)
− 2.835***
(− 2.875
to − 2.795)
Absolute Humid-
ity (14-day Lag)
− 0.069***
(− 0.073
to − 0.065)
− 0.038***
(− 0.041
to − 0.035)
− 0.098***
(− 0.099
to − 0.097)
− 0.108***
(− 0.109
to − 0.106)
0.071***
(0.069–0.073) 0.221***
(0.220–0.222)
Retail and Recrea-
tion (14-day Lag) 1.313***
(1.279–1.348) 0.276***
(0.247–0.305) 0.709***
(0.700–0.719) 0.288***
(0.280–0.296) 0.866***
(0.847–0.884) 0.537***
(0.523–0.550)
Grocery Stores
and Pharmacies
(14-day Lag)
− 0.148***
(− 0.166
to − 0.130)
0.096***
(0.079–0.113) 0.352***
(0.347–0.357) 0.492***
(0.487–0.496)
− 0.112***
(− 0.125
to − 0.098)
0.261***
(0.253–0.269)
Parks (14-day
Lag)
− 0.545***
(− 0.567
to − 0.523)
0.098***
(0.085–0.111)
− 0.184***
(− 0.190
to − 0.179)
− 0.239***
(− 0.244
to − 0.235)
− 0.144***
(− 0.153
to − 0.136)
0.528***
(0.514–0.541)
Transit Stations
(14-day Lag)
− 0.463***
(− 0.497
to − 0.430)
− 1.021***
(− 1.052
to − 0.989)
− 0.633***
(− 0.645
to − 0.621)
− 0.589***
(− 0.603
to − 0.576)
− 0.230***
(− 0.247
to − 0.213)
− 0.411***
(− 0.423
to − 0.400)
Workplaces (14-
day Lag)
− 0.650***
(− 0.669
to − 0.631)
− 0.544***
(− 0.574
to − 0.514)
− 0.696***
(− 0.705
to − 0.687)
− 0.900***
(− 0.909
to − 0.892)
− 0.644***
(− 0.663
to − 0.625)
− 0.072***
(− 0.083
to − 0.061)
Residential (14-
day Lag)
− 0.256***
(− 0.278
to − 0.233)
− 0.579***
(− 0.615
to − 0.543)
− 0.356***
(− 0.367
to − 0.345)
− 0.782***
(− 0.791
to − 0.773)
− 0.233***
(− 0.256
to − 0.209)
0.176***
(0.165–0.188)
Observations 3604 2903 9781 11,260 6446 11,460
Table 3. Untransformed GLM coecient estimates for the 2020 winter and 2021 spring seasons.
Untransformed coecient (β) estimates for GLM Regression against new cases per 100,000 from October
1, 2020 to March 1, 2021. e 95% condence intervals are shown in parenthesis. Estimated coecients for
county-level xed eects and epidemiological terms (immunity factor and lagged daily cases) are not shown.
*p < 0.05 **p < 0.01 ***p < 0.001.
Predictors
Low 1 Low 2 Mid 1 Mid 2 High 1 High 2
Log-Mean Log-Mean Log-Mean Log-Mean Log-Mean Log-Mean
Intercept 5.411***
(5.391–5.431) 6.039***
(6.013–6.066) 5.782***
(5.772–5.791) 4.553***
(4.540–4.566) 6.410***
(6.318–6.498) 5.188***
(5.168–5.207)
Absolute Humid-
ity (14-day Lag)
− 0.141***
(− 0.144
to − 0.138)
− 0.093***
(− 0.096
to − 0.091)
− 0.151***
(− 0.152
to − 0.150)
− 0.220***
(− 0.221
to − 0.219)
− 0.159***
(− 0.160
to − 0.157)
− 0.093***
(− 0.093
to − 0.092)
Retail and Recrea-
tion (14-day Lag) 0.329***
(0.314–0.344) 0.780***
(0.764–0.795) 0.567***
(0.559–0.575)
− 0.167***
(− 0.175
to − 0.158)
0.450***
(0.436–0.464) 0.511***
(0.501–0.521)
Grocery Stores
and Pharmacies
(14-day Lag)
0.312***
(0.302–0.322) 0.380***
(0.369–0.391) 0.501***
(0.497–0.506) 0.782***
(0.777–0.788) 0.200***
(0.191–0.209) 0.367***
(0.359–0.374)
Parks (14-day
Lag)
− 0.518***
(− 0.527
to − 0.509)
− 0.030***
(− 0.037
to − 0.022)
0.267***
(0.263–0.270) 0.277***
(0.274–0.281)
− 0.249***
(− 0.254
to − 0.243)
− 0.736***
(− 0.743
to − 0.729)
Transit Stations
(14-day Lag) 0.244***
(0.230–0.257) 0.129***
(0.116–0.142)
− 0.171***
(− 0.178
to − 0.165)
− 0.023***
(− 0.027
to − 0.019)
0.079***
(0.068–0.089)
− 0.038***
(− 0.046
to − 0.031)
Workplaces (14-
day Lag) 0.017***
(0.007–0.027) 0.469***
(0.458–0.480) 0.514***
(0.509–0.519) 0.347***
(0.342–0.352)
− 0.145***
(− 0.152
to − 0.138)
− 0.081***
(− 0.086
to − 0.077)
Residential (14-
day Lag) 0.545***
(0.529–0.561) 1.322***
(1.304–1.340) 1.273***
(1.265–1.281) 0.931***
(0.923–0.938) 0.218***
(0.207–0.229) 0.223***
(0.216–0.230)
Observations 5953 5084 15,787 15,827 10,135 14,456
Content courtesy of Springer Nature, terms of use apply. Rights reserved
5
Vol.:(0123456789)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
Discussion
As the COVID-19 epidemic continues in the US and given the surge of COVID-19 in the winter seasons, there
is renewed interest in understanding the relationship between outbreaks and seasonal changes, especially cli-
matological factors related to outdoor and indoor humidity. is is not the rst study to investigate humidity
impacts on transmission, which been associated with increased transmission of respiratory pathogens (e.g.,
inuenza) and SARS-CoV-2. While SARS-CoV-2 is a novel human virus, other pandemic coronaviruses (e.g.,
MERS-CoV and SARS-CoV-1)9,32–35 have also been associated with increased transmission in the winter, thus
suggesting similar implications for SARS-CoV-2. Here, we found that the relative eect of absolute humidity
on transmissions has so far been signicant and was greatest in the Western, upper Midwest, and Northeast
regions of the United States, which were clustered into the driest climatological regimes. ese results support
the hypothesis that falling rates of absolute humidity magnify the transmission risk of SARS-CoV-2, particularly
in regions that are more arid and dry36. is eect was less noticeable for more humid regions, such as the coastal
and southern counties of the US (Fig.2).
e eects of behavior and nonpharmaceutical interventions (NPI) are observed in our analysis when we
disaggregate the analysis between the early and later phases of the pandemic. In the early phase of the pandemic,
we see that an increase in mobility trends for retail & recreation resulted in an increase in daily cases, which
measures visits to restaurants, cafes, shopping centers, theme parks, museums, libraries, and movie theaters.
While in the later stages during the fall and winter of 2020, retail & recreation mobility had a lesser eect since
many of those establishments were closed due to NPI policies. Furthermore, increases in residential mobility
played a larger role in transmission, especially during the winter holidays when travel between residential homes
occurred at a higher incidence.
e relationship between humidity and transmission is not fully clear, but several studies have shown that
as absolute humidity decreases, survival times for enveloped viruses increase nonlinearly, including other
coronaviruses9,22,37,38. Our ndings support the hypothesis of a nonlinear relationship since the log-linear eects
between humidity and case growth varied between climatological regimes. Our stratied regression and Fig.2
show that dierent climatological regimes have dierent sensitivities to humidity changes. e increased survival
of the virus in lower AH may be compounded by increased binding capacity, thereby enhancing the potential
infectivity of the virus39. As AH falls, relative humidity indoors also decreases, which may increase susceptibility
to airborne diseases40. is association suggests that increased humidication of indoor air in high transmission
settings may help decrease the burden of COVID-19.
Figure2. e average daily new cases per 100,000 people plotted against the average Google Mobility Measure
of 497 counties for the entire study duration. e plots are organized by type of movement and cluster group.
For each plot, we added a simple linear trend line with shaded standard errors.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
6
Vol:.(1234567890)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
Given that our results suggest COVID-19 cases will increase signicantly during winters, areas where humid-
ity typically falls earlier in the fall (e.g., the upper Midwest) are likely to see cases increase earlier. In contrast,
more humid regions (e.g., Gulf Coast areas) will likely observe outbreaks later in the winter. However, the results
demonstrate that mobility had a larger and signicant impact on cases, particularly when humidity was unchang-
ing in the summer. Consequently, falling temperatures and holiday celebrations are likely to increase the risk
of people gathering in indoor spaces for longer durations, resulting in a surge of COVID-19 cases through the
winter, given that there are no substantial changes in population immunity and behavior.
e prior inuenza pandemic in 2009 is instructive here, as increased contact patterns that occurred in the fall
likely combined with falling humidity to drive transmission, which resulted in the peak of infections occurring
signicantly earlier than other years. Given the uncertainty and nonlinear eects of humidity on transmission,
increasing vaccination, proper social distancing, and improving healthcare capacities can potentially reduce the
toll of the COVID-19 pandemic. In addition, the uncertainty regarding the role of children in transmission41–43,
who remain largely unvaccinated, suggests that proper precautions related to opening schools is warranted as
the potential for transmission increases. While studies linking schools to outbreaks to date have been limited,
few have occurred during the winter when transmission is higher.
We suspected that a relationship between human behavior and climate might exist which can cause variations
in encounters. During winter months, the likelihood of being indoors increases especially in colder climates. To
investigate this potential interaction, we conducted a collinear analysis. We can interpret this collinear analysis
as residential and workplace movement patterns not being collinear with meteorological conditions (absolute
humidity) and epidemiological factors (immunity factor and new cases per 100,000 (14-day Lag)). Retail/recrea-
tion and grocery/pharmacies are moderately collinear, while transit stations and parks were the most collinearly
related to meteorological and epidemiological variables.
One limitation of this study includes changing social distancing dynamics and masking adherence between
counties. We attempted to account for county-level heterogeneities using xed eects for each county, but these
are static eects. Furthermore, it is dicult to disentangle the epidemiological dynamics that cause exponential
growth of cases. Events related to evacuation in natural disasters or mass-gatherings during the summer of
2020 that were not reected in the Google Mobility Data44 may bias the analysis. Also, as with many COVID-19
analyses on retrospective data, the dierences in testing rates at the county-level will result in varying detection
rates of actual cases. Potential variations around vaccination ecacy for variants and within-host changes will
impact the magnitude and exact timing of outbreaks45.
Transmission of SARS-CoV-2 will likely increase during the winters in the United States and other temper-
ate regions in the northern hemisphere due in part to falling humidity. Studies of prior viruses and preliminary
studies of the SARS-CoV-2 virus underpin the theoretical connection between humidity and transmission of
droplet and aerosols. Nevertheless, mobility is still a signicant driver of transmission.
Methods
Study design. e United States is geographically large, and the timing and magnitude of changes in abso-
lute humidity can vary widely across regions. In order to account for regional dierences in humidity, we utilized
a partitional clustering algorithm with dynamic time warping (DTW) similarity measurements46 to classify the
absolute humidity temporal prole for all observed counties into six exclusive clusters that are ranked based
on average humidity. e clustering algorithm was implemented using the dtw package in R47. ese clusters
are ranked from lowest to highest as Low 1, Low 2, Mid 1, Mid 2, High 1, and High 2. Clustering allowed us to
designate groups of counties based on temporal, climatological regimes and to stratify dierent absolute humid-
ity patterns, which reduces group-level eects and enhances the independence of the data points. e DTW
clustering of absolute humidity was conducted on a larger set of 3,137 counties. In the regression analysis, we
included data from a subset of counties that had more than twenty cumulative conrmed cases and a population
of more than 50,000 people. We excluded any days with fewer than 20 cumulative conrmed cases within each
county because early transmission dynamics had a high rate of undetected cases48, making the data unreliable
for this analysis. e nal dataset used in the regression analysis included 497 counties, where separate panel
data GLM was conducted on counties in each cluster (NLow1 = 39, NLow 2 = 42, NMid1 = 118, NMid2 = 108, NHigh1 = 78,
and NHigh2 = 105). We assessed the results of the model over the entirety of the dataset and two time periods in
2020–2021: (1) the entire duration of the dataset (March 10, 2020 to March 1, 2021), (2) spring and summer
when humidity increases (March 10, 2020 to September 30, 2020), and (3) the fall and winter months when
humidity decreases to its lowest point (October 1, 2020 to March 1, 2021).
Data sources. Conrmed case data were extracted from the Johns Hopkins Center for Systems Science and
Engineering1 for each county. Population data were obtained from the US Census Bureau49 for 3,137 counties
from March 10, 2020 to March 1, 2021. Daily cases were obtained from the conrmed case count by taking a
simple dierence between the days. Any data incongruencies, such as negative case counts, were omitted in our
analysis.
Daily average absolute humidity for each US county, excluding territories, was calculated using tempera-
ture and dewpoint data from the National Centers for Environmental Information50 at the National Oceanic
and Atmospheric Administration (NOAA). Time series data for the year 2020 from US weather stations were
acquired from the NOAA Global Summary of the Day Index51. Weather stations were mapped using latitude
and longitude to corresponding counties using the Federal Communications Commission (FCC) Census Block
API52. For counties without a weather station, we used data from the nearest station, which was calculated based
on distance from the county’s spatial centroid using the haversine formula. In cases where counties contained
Content courtesy of Springer Nature, terms of use apply. Rights reserved
7
Vol.:(0123456789)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
multiple stations, data were averaged across all stations in a county. Absolute humidity was calculated using
average daily temperature and average daily dew point (see Alduchov and Eskridge53).
Data on mobility from March 10, 2020 to March 1, 2021 was obtained from the Google COVID-19 Com-
munity Mobility Reports54. We specically utilized the metric that measures visits to grocery stores & pharma-
cies, parks, transit stations, retail & recreation, residential, and workplaces by comparing the median rate on the
county-level to a 5-week period Jan 3–Feb 6, 2020. e measure was calculated as the percent dierence from
before policy interventions (e.g., shelter-in-place orders) began to impact movement. is temporal measure
allowed us to compare movement dierences across counties.
Statistical analysis. For each humidity cluster that was classied using the DTW algorithm, we conducted
three multivariate regressions using a generalized linear model (GLM) that assessed the time-weighted associa-
tion between absolute humidity and non-essential visits with the number of new coronavirus cases (Eqs.1–3).
e GLM regression results in Tables1, 2 and 3 are described in the following equation,
where Yit, is the number of daily COVID-19 cases for county i at time t, log(N) is an oset term to control for
population-size, and α is the intercept. In order to account for population immunity and exponential growth
dynamics, we added the independent variables cumulative cases per 100,000, IMt, and lagged daily cases per
100,000, yi(t-δ) to the regression models. Absolute humidity, AHi(t-δ) is smoothed using a 7-day moving average
and lagged by δ days. Google mobility trends to retail and recreation, RRi(t-δ), grocery and pharmacies, GPi(t-δ),
parks, PKi(t-δ), transit stations, TSi(t-δ), workplaces, WPi(t-δ), residential places, RDi(t-δ), are smoothed using a 7-day
moving average, lagged by δ days, and rescaled and centered on the mean. Fixed eects γi for each county were
added to capture unobserved heterogeneities between counties. For our study, we assumed that the lag length
δ was equal to 14days, which is based on previous studies investigating lagged eects due to the incubation
period of COVID-1955. As our outcome variable was daily cases, we modeled the variable as a Poisson distrib-
uted random variable with a log-transformed link function. Standard errors were calculated for the estimated
linear coecients β.
We conducted additional regressions on the absolute humidity and mobility measures as predictors indi-
vidually to test for robustness. Specically, we t a GLM with absolute humidity for each humidity cluster and
one measure from rescaled Google COVID-19 Community Mobility as linear predictors for new daily cases, as
described in Eqs. (2) to (8).
To demonstrate robustness in the coecient estimates, the coecients in the combined regression analyses
with absolute humidity and all mobility trends (Eq.(1)) were compared to the regression coecients for absolute
humidity and each mobility trend (Eqs. (2)–(8)). e analysis using GLM was conducted using the stats package
in R (Version 4.0.2). All untransformed coecient estimates are located in (Tables1, 2 and 3). In the main text,
we reported the logit-transformed estimates as relative change in cases per unit increase (1g/m3) of absolute
humidity. Given the log-linear relationship in a Poisson regression between the covariates and response variable,
we can calculate the percent change in daily cases for a unit increase of a covariate to be equal to exp (β) − 1. For
example, if β = − 0.112 for absolute humidity, we would state that there is a 9% (= exp (− 0.112) − 1) reduction
for 1g/m3 increase in absolute humidity. To verify that mulicollinearity is not a major issue, we conducted a
collinearity analysis by calculating the Generalized Variational Ination Factor (GVIF) for all regressions, which
are listed in TableS19.
In addition to running a GLM regression, we also discretized the data based on months for each humidity
cluster and calculated the Pearson correlation coecient for absolute humidity and Google Mobility Trends
against new cases (Fig.S2). Stationarity was checked for absolute humidity and Google mobility trends using the
Levin-Lin-Chu unit-root test for unbalanced panel data for the three periods that were analyzed aforementioned
regressions. Results for the stationarity are listed in TableS20 in the supplement.
We tested for robustness and externally validated our regressions by conducting additional analysis using
K-folds cross-validation. We validated the coecient estimation of all the GLMs mentioned previously by show-
ing that the relative eect size for each regression was similar. e analysis was conducted over 100 folds or
(1)
log
(
Y
it )=
log
(
N
)+
α
+
β
1
IM
t+
β
2
y
i(t−δ)+
β
3
AH
i(t−δ)+
β
4
RR
i(t−δ)+
β
5
GP
i(t−δ
)
+β
6PKi(t
−
δ)
+β
7TSi(t
−
δ)
+β
8WPi(t
−
δ)
+β
9RDi(t
−
δ)
+γ
i
+ǫ
it
(2)
log (Yit )
=
log (N)
+α+β
1IMt
+β
2yi(t−δ)
+β
3AHi(t−δ)
+γ
i
+ǫ
it
(3)
log
(
Yit
)=
log
(
N
)+
α
+
β1IMt
+
β2yi(t
−
δ)
+
β3AHi(t
−
δ)
+
β4RRi(t
−
δ)
+
γi
+
ǫit
(4)
log
(
Yit
)=
log
(
N
)+
α
+
β1IMt
+
β2yi(t
−
δ)
+
β3AHi(t
−
δ)
+
β4GPi(t
−
δ)
+
γi
+
ǫit
(5)
log
(
Yit
)=
log
(
N
)+
α
+
β1IMt
+
β2yi(t
−
δ)
+
β3AHi(t
−
δ)
+
β4PKi(t
−
δ)
+
γi
+
ǫit
(6)
log
(
Yit
)=
log
(
N
)+
α
+
β1IMt
+
β2yi(t
−
δ)
+
β3AHi(t
−
δ)
+
β4TSi(t
−
δ)
+
γi
+
ǫit
(7)
log
(
Yit
)=
log
(
N
)+
α
+
β1IMt
+
β2yi(t
−
δ)
+
β3AHi(t
−
δ)
+
β4RDi(t
−
δ)
+
γi
+
ǫit
(8)
log
(
Yit
)=
log
(
N
)+
α
+
β1IMt
+
β2yi(t
−
δ)
+
β3AHi(t
−
δ)
+
β4WPi(t
−
δ)
+
γi
+
ǫit
Content courtesy of Springer Nature, terms of use apply. Rights reserved
8
Vol:.(1234567890)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
iterations with separate training and test sets derived from a subset of the county-level data. We used test sets
for each fold where the mean square error (MSE) was calculated for each t and shown in TableS22 in the sup-
plement. In order to minimize overtting, we also excluded county-level xed eects in our cross-validation
analysis. Additionally, we show the 95% condence intervals of all parameter estimations using the GLM model
that includes all variables in TableS23.
Data availability
e data that support the ndings of this study are openly available through the Johns Hopkins Center for
Systems Science and Engineering, Unacast Social Distancing Scorecard, and NOAA National Centers for Envi-
ronmental Information. Population data can be found through the US Census Bureau Website. All input data
and code used to conduct the analysis and generate gures are also available on Github at https:// github. com/
CDDEP- DC/ COVID- Humid ity- Mobil ity- GAM.
Received: 5 October 2021; Accepted: 6 September 2022
References
1 . Dong, E., Du, H. & Gardner, L. An interactive web-based dashboard to track COVID-19 in real time. Lancet Infect. Dis. 20, 533–534
(2020).
2. Liu, J. et al. Community transmission of severe acute respiratory syndrome coronavirus 2, Shenzhen, China, 2020. Emerg. Infect.
Dis. 26, 1320–1323 (2020).
3. Chan, J.F.-W. et al. A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person
transmission: A study of a family cluster. e Lancet 395, 514–523 (2020).
4. CDC. Coronavirus Disease 2019 (COVID-19). Centers for Disease Control and Prevention. https:// www. cdc. gov/ coron avirus/
2019- ncov/ more/ scien tic- brief- sars- cov-2. html (2020).
5. Tang, S. et al. Aerosol transmission of SARS-CoV-2? evidence, prevention and control. Environ. Int. 144, 106039 (2020).
6. Ma, J. et al. Coronavirus disease 2019 patients in earlier stages exhaled millions of severe acute respiratory syndrome coronavirus
2 per hour. Clin. Infect. Dis. https:// doi. org/ 10. 1093/ cid/ ciaa1 283 (2021).
7. de Man, P. et al. Outbreak of coronavirus disease 2019 (COVID-19) in a nursing home associated with aerosol transmission as a
result of inadequate ventilation. Clin. Infect. Dis. https:// doi. org/ 10. 1093/ cid/ ciaa1 270 (2021).
8 . Rahman, H. S. et al. e transmission modes and sources of COVID-19: A systematic review. Int. J. Surg. Open 26, 125–136 (2020).
9. Shaman, J. & Kohn, M. Absolute humidity modulates inuenza survival, transmission, and seasonality. Proc. Natl. Acad. Sci. 106,
3243–3248 (2009).
10. Wu, Y. et al. Eects of temperature and humidity on the daily new cases and new deaths of COVID-19 in 166 countries. Sci. Total
Environ. 729, 139051 (2020).
11. Kato, M., Sakihama, T., Kinjo, Y., Itokazu, D. & Tokuda, Y. Eect of Climate on COVID-19 Incidence: A Cross-Sectional Study in
Japan. https:// papers. ssrn. com/ abstr act= 36121 14. https:// doi. org/ 10. 2139/ ssrn. 36121 14 (2020).
12. Christophi, C. A. et al. Ambient Temperature and Subsequent COVID-19 Mortality in the OECD Countries and Individual United
States. https:// papers. ssrn. com/ abstr act= 36121 12. https:// doi. org/ 10. 2139/ ssrn. 36121 12 (2020).
13. Meyer, A., Sadler, R., Faverjon, C., Cameron, A. R. & Bannister-Tyrrell, M. Evidence that higher temperatures are associated with
a marginally lower incidence of COVID-19 cases. Front. Public Health https:// doi. org/ 10. 3389/ fpubh. 2020. 00367 (2020).
14. Steiger, E., Mussgnug, T. & Kroll, L. E. Causal analysis of COVID-19 observational data in German districts reveals eects of
mobility, awareness, and temperature. medRxiv https:// doi. org/ 10. 1101/ 2020. 07. 15. 20154 476 (2020).
15. Kifer, D. et al. Eects of environmental factors on severity and mortality of COVID-19. medRxiv https:// doi. org/ 10. 1101/ 2020. 07.
11. 20147 157 (2020).
16. Sehra, S. T., Salciccioli, J. D., Wiebe, D. J., Fundin, S. & Baker, J. F. Maximum daily temperature, precipitation, ultraviolet light, and
rates of transmission of severe acute respiratory syndrome coronavirus 2 in the United States. Clin. Infect. Dis. https:// doi. org/ 10.
1093/ cid/ ciaa6 81 (2020).
17. Mecenas, P., da Rosa Moreira Bastos, R. T., Vallinoto, A. C. R. & Normando, D. Eects of temperature and humidity on the spread
of COVID-19: A systematic review. PLOS ONE 15, e0238339 (2020).
18. Aragão, D. P. et al. Multivariate data driven prediction of COVID-19 dynamics: Towards new results with temperature, humidity
and air quality data. Environ. Res. 204, 112348 (2022).
19. Yuan, S., Jiang, S.-C. & Li, Z.-L. Do humidity and temperature impact the spread of the novel coronavirus?. Front. Public Health
https:// doi. org/ 10. 3389/ fpubh. 2020. 00240 (2020).
20. Lolli, S., Chen, Y.-C., Wang, S.-H. & Vivone, G. Impact of meteorological conditions and air pollution on COVID-19 pandemic
transmission in Italy. Sci. Rep. 10, 16213 (2020).
21. Sajadi, M. M. et al. Temperature, humidity, and latitude analysis to estimate potential spread and seasonality of coronavirus disease
2019 (COVID-19). JAMA Netw. Open 3, e2011834 (2020).
22. Marr, L. C., Tang, J. W., Van Mullekom, J. & Lakdawala, S. S. Mechanistic insights into the eect of humidity on airborne inuenza
virus survival, transmission and incidence. J. R. Soc. Interface 16, 20180298 (2019).
23. Badr, H. S. et al. Association between mobility patterns and COVID-19 transmission in the USA: A mathematical modelling study.
Lancet Infect. Dis. 20, 1247–1254 (2020).
24. Gatalo, O., Tseng, K., Hamilton, A., Lin, G. & Klein, E. Associations between phone mobility data and COVID-19 cases. Lancet
Infect. Dis. 21, e111 (2020).
25. Aragão, D. P., dos Santos, D. H., Mondini, A. & Gonçalves, L. M. G. National holidays and social mobility behaviors: Alternatives
for forecasting COVID-19 deaths in Brazil. Int. J. Environ. Res. Public Health 18, 11595 (2021).
26. Wang, J. et al. High temperature and high humidity reduce the transmission of COVID-19. SSRN https:// doi. org/ 10. 2139/ ssrn.
35517 67 (2020).
27. Qi, H. et al. COVID-19 transmission in Mainland China is associated with temperature and humidity: A time-series analysis. Sci.
Total Environ. 728, 138778 (2020).
28. Xie, J. & Zhu, Y. Association between ambient temperature and COVID-19 infection in 122 cities from China. Sci. Total Environ.
724, 138201 (2020).
29. Poirier, C. et al. e role of environmental factors on transmission rates of the COVID-19 outbreak: An initial assessment in two
spatial scales. Sci. Rep. 10, 17002 (2020).
30. Bhagat, R. K., Wykes, M. S. D., Dalziel, S. B. & Linden, P. F. Eects of ventilation on the indoor spread of COVID-19. J. Fluid Mech.
https:// doi. org/ 10. 1017/ jfm. 2020. 720 (2020).
31. Wickham, H. ggplot2: Elegant Graphics for Data Analysis (Springer-Verlag, 2016).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
9
Vol.:(0123456789)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
32. Gardner, E. G. et al. A case-crossover analysis of the impact of weather on primary cases of Middle East respiratory syndrome.
BMC Infect. Dis. 19, 113 (2019).
33. Chan, K. H. et al. e eects of temperature and relative humidity on the viability of the SARS coronavirus. Adv. Virol. 2011, 1–7
(2011).
34. Yuan, J. et al. A climatologic investigation of the SARS-CoV outbreak in Beijing, China. Am. J. Infect. Control 34, 234–236 (2006).
35. Altamimi, A. & Ahmed, A. E. Climate factors and incidence of Middle East respiratory syndrome coronavirus. J. Infect. Public
Health 13, 704–708 (2020).
36. AntonyAroulRaj, V., Velraj, R. & Haghighat, F. e contribution of dry indoor built environment on the spread of Coronavirus:
Data from various Indian states. Sustain. Cities Soc. 62, 102371 (2020).
37. McDevitt, J., Rudnick, S., First, M. & Spengler, J. Role of absolute humidity in the inactivation of inuenza viruses on stainless
steel surfaces at elevated temperatures. Appl. Environ. Microbiol. 76, 3943–3947 (2010).
38. Shaman, J., Goldstein, E. & Lipsitch, M. Absolute humidity and pandemic versus epidemic inuenza. Am. J. Epidemiol. 173, 127–135
(2011).
39. de la Noue, A. C. et al. Absolute humidity inuences the seasonal persistence and infectivity of human norovirus. Appl. Environ.
Microbiol. 80, 7196–7205 (2014).
40. Ahlawat, A., Wiedensohler, A. & Mishra, S. K. An overview on the role of relative humidity in airborne transmission of SARS-
CoV-2 in indoor environments. Aerosol Air Qual. Res. 20, 1856–1861 (2020).
41. Qiu, H. et al. Clinical and epidemiological features of 36 children with coronavirus disease 2019 (COVID-19) in Zhejiang, China:
An observational cohort study. Lancet Infect. Dis 20, 689–696 (2020).
42. Kelvin, A. A. & Halperin, S. COVID-19 in children: e link in the transmission chain. Lancet Infect. Dis 20, 633–634 (2020).
43. Viner, R. M. et al. School closure and management practices during coronavirus outbreaks including COVID-19: A rapid systematic
review. Lancet Child Adolesc. Health 4, 397–404 (2020).
44. Salas, R. N., Shultz, J. M. & Solomon, C. G. e climate crisis and Covid-19—A major threat to the pandemic response. N. Engl.
J. Med. 383, e70 (2020).
45. Kronfeld-Schor, N. et al. Drivers of infectious disease seasonality: Potential implications for COVID-19. J. Biol. Rhythms 36, 35–54
(2021).
46. Berndt, D. J. & Cliord, J. Using dynamic time warping to nd patterns in time series. 12.
47. Giorgino, T. Computing and visualizing dynamic time warping alignments in R : e dtw package. J. Stat. Sow. https:// doi. org/
10. 18637/ jss. v031. i07 (2009).
48. Silverman, J. D., Hupert, N. & Washburne, A. D. Using inuenza surveillance networks to estimate state-specic prevalence of
SARS-CoV-2 in the United States. Sci. Transl. Med. https:// doi. org/ 10. 1126/ scitr anslm ed. abc11 26 (2020).
49. 2018 ACS 1-year Estimates. e United States Census Bureau https:// www. census. gov/ progr ams- surve ys/ acs/ techn ical- docum
entat ion/ table- and- geogr aphy- chang es/ 2018/1- year. html.
50. National Centers for Environmental Information (NCEI). https:// www. ncei. noaa. gov/.
51. Global Surface Summary of the Day—GSOD—NOAA Data Catalog. https:// data. noaa. go v/ datas et/ datas et/ global- surfa ce- summa
ry- of- the- day- gsod.
52. Census Block Conversions API. Federal Communications Commission https:// www. fcc. gov/ census- block- conve rsions- api (2011).
53. Alduchov, O. A. & Eskridge, R. E. Improved Magnus form approximation of saturation vapor pressure. J. Appl. Meteorol. 35,
601–609 (1996).
54. Google LLC. COVID-19 Community Mobility Report. COVID-19 Community Mobility Report https:// www. google. com/ covid 19/
mobil ity? hl= en.
55. Lauer, S. A. et al. e incubation period of coronavirus disease 2019 (COVID-19) from publicly reported conrmed cases: Estima-
tion and application. Ann. Intern. Med. 172, 577–582 (2020).
Acknowledgements
is work was funded by the Centers for Disease Control and Prevention (CDC) MInD-Healthcare Program
(Grant Numbers U01CK000589, 1U01CK000536, and contract number 75D30120P07912). e funders had no
role in the design, data collection and analysis, decision to publish, or preparation of the manuscript.
Author contributions
E.K. conceived the research, G.L. designed the study, A.H. and O.G. collected and processed the data, G.L., E.K.,
F.H., T.I. analyzed and interpreted the data. All authors contributed to interpretation of results and manuscript
writing.
Competing interests
e authors declare no competing interests.
Additional information
Supplementary Information e online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 022- 19898-8.
Correspondence and requests for materials should be addressed to G.L.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
10
Vol:.(1234567890)
Scientic Reports | (2022) 12:16729 | https://doi.org/10.1038/s41598-022-19898-8
www.nature.com/scientificreports/
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2022
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
Available via license: CC BY 4.0
Content may be subject to copyright.