Health associations with DTR

A large and growing body of literature has examined associations between DTR and human health across diverse geographic regions and populations. Most studies have focused on mortality outcomes, with consistent evidence of positive associations reported in Asia, North America, and Europe [1,11–25]. Cause-specific analyses indicate that these relationships are particularly evident for cardiovascular [1,24,26–30] and respiratory [1,16,31–34] outcomes, including both mortality and morbidity, although associations have also been identified for cerebrovascular [35–37] and other causes [38–41]. In addition to mortality, a number of studies have examined morbidity outcomes, including hospital admissions, emergency department visits, and ambulance dispatches, which likewise tend to show increased risk on days with elevated DTR [26,27,31,32,35,36,42–45].

Methodologically, most DTR–health studies employ time-series or time-stratified case-crossover designs to evaluate short-term associations between daily temperature variability and health outcomes. These approaches are typically implemented using Poisson or quasi-Poisson regression frameworks and often incorporate distributed lag structures to account for delayed effects (e.g., [14,19,46]). Most studies control for mean temperature and/or other meteorological variables and often stratify analyses by season to account for differing climatological regimes. Adjustments for long-term trends, day-of-week effects, and air pollution are also common, although model specifications vary considerably across studies.

With respect to the exposure–response relationship, most studies report a positive, approximately linear association, indicating that higher DTR is associated with increased mortality or morbidity risk. However, several studies have identified nonlinear relationships, including J-shaped or U-shaped curves, suggesting that both high and low DTR values may be associated with adverse health outcomes [16,24,27,46,47]. These findings imply that the health impacts of DTR may differ depending on the underlying atmospheric conditions, as high vs. low DTR days are driven by markedly different weather situations.

Despite the general consistency in findings across locations, outcomes, and analytical approaches, important uncertainties remain regarding the interpretation of DTR as an exposure variable. In particular, the physiological mechanisms linking short-term temperature variability to adverse health outcomes are not well established [43], and DTR is often evaluated independently of the meteorological factors that influence its magnitude. As a result, it remains unclear whether DTR represents a direct physiological stressor or serves as a proxy for a broader set of environmental conditions that may influence human health.

Physical processes associated with DTR

DTR is the difference between temperature extremes on a given day, and the magnitude and timing of the maximum (T_max) and minimum (T_min) temperatures are influenced by a variety of physical processes that can interact to enhance or suppress either extreme. On a clear day, T_max will typically occur 2–3 hours after the time of maximum solar radiation receipt (solar noon). The time lag between maximum incoming radiation and temperature is related to the thermal responsiveness (heat capacity) of the surface type (soil, vegetation, concrete, water, etc.), which emits infrared (terrestrial) radiation, or heat, to the atmosphere. The difference between incoming solar and outgoing terrestrial radiation defines the net radiation. Although net radiation typically peaks near midday, T_max generally occurs later in the afternoon, after an extended period of positive net radiation and as the surface energy balance approaches zero.

Surface heat is transmitted to the atmosphere through sensible heating and by convective mixing within the daytime boundary layer. Under clear conditions, surface heating promotes turbulent mixing that entrains warmer residual air from aloft (often from the previous day) down to the surface, contributing to the daytime rise in air temperature and the evolution of T_max. Clouds and humidity can significantly impact this timing, such that the hour of T_max tends to be later in the afternoon in humid climates. Water vapor is a greenhouse gas that enhances the absorption of outgoing heat from the surface, thus warming the atmosphere and delaying the rate of afternoon heat loss. Wind shifts during the day can alter temperature advection, so the transport of warmer or colder air into a region can further impact the timing of T_max. Intra-day wind shifts are most often associated with frontal passages, in which a warm resident air mass is displaced by a colder air mass (cold front passage) or vice-versa (warm front passage). On days with cold front passages, it is possible that T_max could occur early in the day and T_min near midnight, as progressively colder air is advected southward by the winds. In mountainous regions, episodic wind events (e.g., Chinook winds) can strongly influence the magnitude of daytime temperatures and DTR as air warms while descending from higher elevations.

Similarly, T_min is dictated by the surface radiation balance. On clear days, Earth’s surface typically begins cooling in the mid to late afternoon. Net radiation becomes most negative shortly after sunset, when incoming shortwave radiation approaches zero and outgoing longwave radiation remains high, while T_min generally occurs just after sunrise following prolonged nocturnal cooling. However, the extent of nighttime cooling is greatly influenced by humidity and cloud cover (both of which retard the cooling rate), winds (strong winds mix warmer air from aloft into the boundary layer), and the number of hours of darkness. T_min is particularly influenced by the dew point temperature, as once the air cools to the dew point, water vapor in the air is condensed into liquid (dew or fog) which releases latent heat to the surroundings, quelling any further drop in temperature. Therefore, clear, calm nights will have lower T_min than overcast and/or windy nights. As with maximum temperature, T_min can occur at any hour of the day when there is a frontal passage and/or significant wind shifts.

Among the meteorological variables that modulate the surface radiation balance, humidity and cloud cover have a profound influence on DTR. Everything else being equal, DTR will be suppressed on an overcast day and will be largest on a clear, calm day. Days with very low DTRs are typically overcast (or foggy or extremely polluted), as there is no other mechanism to suppress both the rise in daytime temperature and the decline in nighttime temperature. Therefore, in epidemiological studies that find health responses to low DTR, it is possible that DTR is serving as a proxy for storminess, precipitation, or very poor air quality. DTR will be suppressed in midlatitude climates with warm, humid summers because the high dew point temperatures raise T_min, and on cold and dry winter days, DTR is suppressed by the lower daytime maximum air temperatures in the cold season. Thus, the most extreme DTR values, which are the emphasis of this study, are often found on days with dry air masses in the spring and autumn, when T_max can approach summertime values, but the dry air (i.e., low dew point temperature) and length of night allow for a low T_min to be attained.

Trends in DTR

Decreases in DTR have been documented across many parts of the globe, with some of the earliest evidence presented by Karl et al. [48]. These early findings were based primarily on station-based observational datasets, and subsequent analyses using homogenized station records have consistently confirmed widespread declines in DTR across North America and globally [49–55,56]. Reanalysis-based products, including ERA5 and related datasets, have likewise reproduced these large-scale patterns [57–59]. The widespread decline that emerged during the mid-20th century has been attributed primarily to faster increases in T_min relative to T_max [49–55]. This asymmetry in warming, where nighttime temperatures rise more rapidly than daytime temperatures, has been widely noted in recent decades [49,50,60,61] and has contributed to a long-term global reduction in DTR [55,56,62,63]. Some portion of this asymmetry has also been linked to non-climatic influences, including urbanization surrounding weather stations (a process not explicitly represented in most reanalysis datasets, such as ERA5-Land), which can elevate nighttime temperatures and artificially suppress DTR. Prior to 1950, DTR trends are uncertain because of sparse and uneven station records [56], but longer-term data suggest a shift from increasing to decreasing DTR beginning around 1950, followed by a flattening of the global trend between 1979 and 2004 as increases in T_max began matching those of T_min [52]. Recent studies note a renewed increase in DTR in parts of the world (including North America) linked to reductions in aerosols and shifts in cloud cover and precipitation [64–66]. With respect to future trends in DTR, climate model projections remain mixed and depend strongly on emissions pathways and associated regional climate processes. A recent study by Wang et al. [67] suggests a continued decline in DTR under intermediate and high emissions scenarios, particularly over parts of North America, with a shift towards increasing DTR under low emissions scenarios.

Materials and methods

Hourly ERA5-Land reanalysis data for the continental United States (CONUS) were accessed through Google Earth Engine for the period January 1950 through December 2022. ERA5-Land is a high-resolution dataset that provides a consistent record of land surface variables over several decades [68]. Although reanalysis datasets are model-based, they are widely used to assess long-term spatial and temporal climate patterns due to their consistency and spatial completeness [57–59]. DTR, particularly its extreme values, can be sensitive to land surface processes and local-scale variability that may not be fully resolved in reanalysis products. However, ERA5-Land has been shown to reproduce observed temperature variability and long-term trends consistent with station-based datasets [57–59]. In addition, comparisons with station observations at representative locations indicate that the seasonal structure of DTR is well captured, supporting the use of ERA5-Land for regional-scale climatological analysis (see S1 Fig). ERA5-Land improves upon the spatial resolution of the original ERA5 dataset by rerunning the land component of the ECMWF ERA5 reanalysis, offering data at a resolution of approximately 9 km. Using the hourly data, we extracted 2-m air temperatures to determine T_max and T_min values for each day (converted from UTC to local time). We also obtained the 2-m dew point temperature and surface pressure at the time of T_max and T_min and calculated the number of hours of measurable precipitation (≥ 0.127 mm) per day. These variables were selected to help determine those meteorological factors that influence the spatio-temporal characteristics and variability of DTR.

Given the large spatial extent of our study area, the native 9 km resolution was finer than necessary for a regional-scale analysis. To better capture broad spatial patterns, we resampled the data by aggregating the 9 km cells to a coarser 150 km resolution, resulting in a spatial field of 480 grid cells. Aggregating to this scale facilitates clearer regional comparisons, aligns with the spatial scale of many climate-health applications, and reduces the influence of localized variability that may obscure larger-scale signals. For each 150 km cell, we calculated daily DTR values and summarized their distribution by month over the period of record. Initial examination of monthly patterns and trends revealed consistent seasonal patterns, so we instead aggregated the daily results by meteorological seasons (winter: DJF; spring: MAM; summer: JJA; autumn: SON). For analysis and visualization purposes, we encoded each grid cell as a node (represented as the centroid of the cell).

Because extreme DTR values have been linked to increased risk of adverse health outcomes, we focused on the upper tail of the DTR distribution. Specifically, we identified the 95th percentile DTR threshold at each grid node over the period of record and then counted the number of days in each season-year when this threshold was equaled or exceeded. To assess long-term trends in extreme DTR, we performed linear regression on the seasonal exceedance counts and evaluated statistical significance at the 95% confidence level. We report standardized regression slopes (β) to enable comparison of trend magnitudes across regions that differ in the variability of seasonal exceedance counts. Standardization provides a scale-independent measure of change, allowing spatial differences in the magnitude of trends to be compared directly. To account for temporal autocorrelation, trend models were fit using generalized least squares (GLS) with a first-order autoregressive [AR(1)] error structure. Mapped effect sizes were expressed as standardized regression coefficients obtained by fitting the GLS model to standardized predictor and response variables at each grid node. Statistical significance of the standardized coefficients was taken directly from the GLS model output. To evaluate the robustness of the results, we repeated the exceedance analysis using alternative extreme DTR thresholds based on the 97.5th and 99th percentiles. Results using these higher thresholds were consistent with those obtained using the 95th percentile, indicating that our conclusions are not sensitive to the specific definition of extreme DTR. The 95th percentile is also a common threshold choice in DTR-health research [26,27,30,39,40].

Climate variables are spatially correlated to varying degrees. Although our primary purpose is to explore general patterns of DTR, we are also interested in the strength of any long-term trends. To account for spatial autocorrelation across grid nodes and the true degrees of freedom in the spatial field, we calculated the False Discovery Rate (FDR; [69]), which results in an adjustment to the critical p value. We calculated the FDR based on the sample of 480 grid nodes and used a type I error rate of 0.10, which is a commonly applied threshold in climatological field significance testing [70].

Mean DTR by season

The spatial patterns of mean DTR are broadly consistent with prior climatological studies; here, they are presented to provide context for subsequent analyses of trends, extremes, and underlying meteorological drivers. Fundamentally, DTR is largest at dry (low humidity) locations and/or during dry periods. Mean DTR exhibits distinct seasonal and geographic patterns across the CONUS. During winter, the lowest values occur in the Pacific Northwest, Great Lakes, and portions of the Northeast, while the highest values are concentrated in the Southwest, including southern Arizona, eastern New Mexico, western Texas, and Colorado (Fig 1a). In spring, DTR increases across most of the CONUS, with persistent minima in the Pacific Northwest and Great Lakes and an expanded region of high values across the Southwest (Fig 1b). Summer features the broadest spatial contrast, with maximum DTR values across the Intermountain West and West Coast and minimum values across the eastern CONUS, particularly the Great Lakes, mid-Atlantic, and northern Gulf Coast (Fig 1c). In autumn, DTR decreases across much of the CONUS but remains relatively high in the Southwest and Intermountain West, while the lowest values persist across the Great Lakes, Northeast, and along the East Coast (Fig 1d).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Mean DTR (°C) for (A) winter, (B) spring, (C) summer, and (D) autumn.

Locations described in S1 Fig are shown on map (A). Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g001

The influence of dry air on DTR is clear in examining the seasonal signals at four representative locations derived from station observations (S1 Fig). The mean monthly DTR in the desert of Phoenix, AZ, is substantially larger than in subtropical warm and humid Miami, FL. The spring and autumn peaks in Phoenix and Minneapolis, MN (mid-latitude continental interior), and the summer peak in Seattle, WA (temperate maritime), occur during months with lower cloud cover and precipitation. These station-based records exhibit seasonal DTR patterns consistent with those derived from ERA5-Land (Fig 1), providing an independent point of comparison that supports the accuracy of the reanalysis data.

Trends in mean and extreme DTR by season

While long-term declines in DTR associated with asymmetric warming are well documented, the seasonal structure of these trends, their relationship to underlying meteorological drivers, and their behavior in the upper tail of the distribution remain less well characterized, particularly in a framework relevant to health research. Across seasons, DTR trends are characterized by widespread declines across much of the eastern CONUS and more variable, often nonsignificant changes in the western CONUS. In winter (Fig 2a), statistically significant decreases are concentrated across the Great Lakes, Northeast, mid-Atlantic, and Mississippi River Valley, with additional declines in portions of the Southwest. In contrast, the western CONUS exhibits small and mostly nonsignificant increases. Spring exhibits fewer significant trends (Fig 2b). DTR generally decreases across much of the eastern CONUS and parts of the West, though statistically significant changes are limited and scattered, particularly across the Great Lakes, Ozarks, and Pacific Northwest. Summer shows the most extensive and spatially coherent declines (Fig 2c), especially across the eastern CONUS and parts of the Southwest. In contrast, trends in the western CONUS are generally weak and nonsignificant. In autumn, DTR declines across much of the CONUS (Fig 2d), with significant decreases concentrated across the northern tier, mid-Atlantic, and portions of the Southwest.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Trend in mean DTR for (A) winter, (B) spring, (C) summer, and (D) autumn.

Red (blue) circles indicate increasing (decreasing) trends. Filled (open) circles denote statistically significant (nonsignificant) trends where α = 0.05. Circle size is proportional to the standardized regression slope magnitude, such that larger (smaller) circles correspond to larger (smaller) trends. The number of grid nodes with statistically significant trends is provided for each season next to the up-arrow (increasing trend) and down-arrow (decreasing trend). Plus and minus signs indicate those grid nodes that passed the FDR test that accounts for spatial autocorrelation. Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g002

Because previous studies have shown that health impacts tend to occur on days with high DTR, we next examine trends in the seasonal counts of DTR values exceeding the 95th percentile.

Patterns in extreme DTR largely mirror those observed for mean DTR, with the most pronounced and widespread declines occurring in summer and autumn. In winter (Fig 3a), declines in extreme DTR are observed across portions of the Northeast and Southeast, with more limited and scattered changes elsewhere. Spring shows relatively few significant trends (Fig 3b), with only isolated increases and decreases across the Midwest, Great Lakes, and western CONUS. In contrast, summer exhibits widespread and significant declines in extreme DTR (Fig 3c), particularly across the eastern CONUS and extending into parts of the High Plains and Southwest. Autumn also shows numerous and often large decreases (Fig 3d), especially across the northern tier of the CONUS.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Same as Fig 2 but for the trend in extreme DTR for (A) winter, (B) spring, (C) summer, and (D) autumn.

Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g003

Trends in maximum and minimum temperatures

To evaluate the potential drivers of observed DTR trends, we next examine seasonal trends in T_max and T_min. We also calculate correlation coefficients between DTR and T_max and T_min to characterize how the strength and direction of these relationships vary seasonally and spatially (see Supporting Information). In winter, increasing trends in many parts of the CONUS are noted for both T_max and T_min (Fig 4a and 4b). However, increases in T_min are generally larger in magnitude than those in T_max, resulting in declining DTR. Nearly all grid nodes show strong, statistically significant negative correlations between T_min and DTR, indicating that increases in nighttime temperatures strongly reduce DTR (S2 Fig). Correlations between DTR and T_max are mostly positive, but fewer are statistically significant, and most are near zero.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Trends in (A) maximum and (B) minimum temperatures during winter, and (C) maximum and (D) minimum temperatures during spring.

Plus and minus signs indicate those grid nodes that passed the FDR test that accounts for spatial autocorrelation. Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g004

In spring, T_max and T_min are also increasing across much of the western CONUS (Fig 4c and 4d). Correlation coefficients are positive there, with the strongest associations between DTR and T_max (S3 Fig). Across the eastern CONUS, where spring DTR has declined, T_min is increasing more rapidly than T_max, and correlations between T_min and DTR are strongly negative, similar to the winter pattern.

Summer shows a more complex spatial structure (Fig 5a and 5b). T_max and T_min are increasing across many regions, including the western CONUS and the Northeast, but decreasing trends in T_max are observed across portions of the southern CONUS and Midwest, particularly in Mississippi and Alabama, corresponding to the well-documented “Warming Hole” [71]. This region also exhibits increases in T_min, though trends are smaller than those in the Northeast. As in other seasons, correlations between DTR and T_max are generally positive, while correlations with T_min are mostly negative (S4 Fig). The largest decreases in summer DTR occur where T_max is declining and T_min is increasing, particularly across the eastern CONUS. In contrast, across much of the West, T_max and T_min are increasing at similar rates, resulting in minimal or nonsignificant summer DTR trends.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Same as Fig 4 but for (A,B) summer and (C,D) autumn.

Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g005

In autumn, T_max is increasing across most grid nodes, although significant trends are limited to the western CONUS and Florida (Fig 5c). T_min is also increasing, with larger magnitudes and widespread statistical significance (Fig 5d). These combined associations align with the widespread significant decreases in DTR observed in autumn, particularly across the northern tier of the CONUS. Correlations between DTR and T_min are strongly negative, consistent with patterns found elsewhere. Correlations with T_max are also negative across many areas in autumn (S5 Fig).

Climatological composites of extreme DTR days

To complement the physical interpretation of high DTR conditions, we constructed composite anomaly fields of selected meteorological variables on days with extreme DTR (≥95th percentile) minus the nodal mean value. Across seasons, extreme DTR days are characterized by systematically lower dew point temperatures, particularly at the time of T_min, indicating drier atmospheric conditions. These differences are especially pronounced in summer (Fig 6), when large negative dew point anomalies coincide with widespread reductions in precipitation frequency, consistent with reduced cloud cover on high DTR days. Surface pressure is uniformly higher on extreme DTR days, reflecting the presence of synoptic-scale high-pressure systems associated with subsidence and clear skies. These results support the underlying physical arguments that extreme DTR events occur under a consistent set of meteorological conditions—dry days with low humidity, low cloud cover, and high pressure. (Composite anomaly maps for the other seasons are provided in S6–S8 Figs).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Summer anomalies in (A) dew point temperature at time of maximum temperature, (B) dew point temperature at time of minimum temperature, (C) precipitation frequency (hours), and (D) surface pressure (Pa) at the time of minimum temperature.

Anomalies are the difference in the values of each variable on high DTR days (DTR ≥ 95th percentile value) and the seasonal mean of that variable at each grid node. Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g006

Trends in dew point temperatures and precipitation

To further assess the meteorological controls shaping DTR trends, we examined seasonal trends in dew point temperature at the time of daily T_max and T_min, along with trends in precipitation frequency. These variables provide insight into how atmospheric moisture and humidity modulate short-term temperature variability.

In winter, dew point temperatures have increased at both T_max and T_min, with larger increases occurring at T_min (Fig 7a and 7b). These changes are consistent with observed increases in winter T_min and the associated decline in DTR. Spring exhibits a similar pattern, with increasing dew points at both T_max and T_min, the latter exhibiting larger increases (Fig 7c and 7d). These conditions correspond with reductions in DTR across the central Plains and Pacific Northwest.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Trends in dew point temperature at the time of (A) maximum and (B) minimum temperatures during winter, and (C) maximum and (D) minimum temperatures during spring.

Plus and minus signs indicate those grid nodes that passed the FDR test that accounts for spatial autocorrelation. Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g007

Summer shows widespread and significant increases in dew point at both T_max and T_min (Fig 8a and 8b). In the “Warming Hole” region of the southeastern and south-central CONUS, increasing dew point appears to contribute to both increasing T_min and suppressed T_max, resulting in declining DTR. Autumn also exhibits increasing dew point trends, except in the western CONUS (Fig 8c and 8d). In this region, decreasing dew point at the time of T_max paired with increases at T_min reflects daytime drying and nighttime moistening patterns that correspond with weak or positive DTR trends. Elsewhere, the seasonal moistening in autumn aligns with modest trends in T_max, widespread increases in T_min, and declines in DTR across the northern tier and parts of the southern tier of the CONUS. Both summer and autumn also show significant decreases in extreme DTR values, which align more strongly with changes in T_min than T_max, as indicated by stronger correlations between DTR and T_min. Where T_max exhibits little trend, the decline in extreme DTR values appears tied to nighttime moisture increases and rising temperatures.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Same as Fig 7 but for (A,B) summer and (C,D) autumn.

Base map from ESRI (VGIN, Esri, HERE, Garmin, FAO, NOAA, USGS, EPA, NPS; url: https://www.arcgis.com/home/item.html?id=979c6cc89af9449cbeb5342a439c6a76).

https://doi.org/10.1371/journal.pone.0352866.g008

Analysis of precipitation frequency supports these relationships. Winter precipitation frequency trends are generally weak across the eastern CONUS, though mostly positive and consistent with declining DTR, whereas decreasing trends in the Northwest align with localized increases in DTR (S9 Fig). In spring, decreasing precipitation frequency in parts of the western CONUS are consistent with increasing DTR, while significant increases across the interior CONUS correspond with increasing dew point and declining DTR. Summer exhibits strong regional contrasts. Decreasing precipitation frequency in the West corresponds with daytime drying and increasing DTR, whereas significant increases across the eastern CONUS coincide with elevated dew point and decreasing DTR. In autumn, precipitation frequency increases across the High Plains, Great Lakes, and lower Mississippi River Valley, supporting decreases in both mean DTR and counts of extreme DTR values, while decreases in precipitation frequency are noted in the western CONUS.