Climate data obtained by observers can be affected by a number of issues, including observer bias, time of observation bias, station moves, instrument changes, and missing data to name just a few. To deal with these difficulties as much as possible, metadata for each data set used in this study were carefully evaluated, and only the most appropriate data for climate studies were used. Missing data are also an important problem, especially when daily thresholds in temperature or precipitation are being explored. Unfortunately, daily precipitation data are more greatly affected than temperature, because in many instances observers fail to record any observation on days with no precipitation falling. A lack of observation of precipitation for a given day must then be treated as missing data because the intent of the observer cannot be presumed. For this study, annual data were considered complete for temperature when 95% of the days in a year were available and when 90% were available for precipitation.
The National Climatic Data Center (NCDC) makes available statewide and divisional temperature, precipitation, and drought index data for the period 1895 through 2012. When necessary, observations have been adjusted to account for the effects introduced into the climate record by factors such as instrument changes, station relocations, changes in observer or observing practice, urbanization, etc. Analogous data are available for the state’s two climate divisions: division 1 (New Castle County) and division 2 (Kent and Sussex Counties). These data were used in the analysis of temperature and precipitation variability and large-scale drought (PDSI, PDHI, PMDI, and Palmer-Z) for the entire State and for each of Delaware’s two climate divisions for the period 1895-2013. Some possible problems in the divisional data associated with stations coming into and out of the divisional calculation over time have been discussed in the literature (see Keim et al. 2003 and 2005), especially if the station changes include large elevation differences. Given Delaware’s small elevation differences and the small size of the climate division areas, this problem was judged to be insignificant for this study. The statewide data are calculated from National Weather Service Cooperative Station data from 1931 to the present. Prior to 1931, statewide data are derived from United States Department of Agriculture data.
NCDC’s DSI-3200 data set, includes 23 National Weather Service Cooperative stations that have been located in Delaware at some point since the late 19th century. Daily observations of maximum temperature, minimum temperature, total liquid precipitation, snowfall, and snow cover on the ground are available in this data set. Although the data have been quality controlled, care must still be taken in their use to account for time of observation biases, poor sensor placement, etc. Metadata for all stations were collected and reviewed to ascertain those stations and period of record that are suitable for the analysis of climate variability. The data identified as appropriate for further analyses are used in the evaluation of temperature and temperature extremes, precipitation and precipitation extremes, and potential asymmetric changes in temperature (changes in maximum compared to minimum temperatures). Out of the initial list of 23 possibilities, nine stations were retained for further analyses. A complete list of the climate change indicators calculated from the cooperative station data is given in this link. It is important to note that only climate indicators that evidenced significant trends at a number of locations are discussed in this report. All 41 climate indicators analyzed in this study were investigated at all nine stations (369 separate trends were analyzed). Four high-quality stations including Wilmington Porter Reservoir, Wilmington NCC Airport, Dover, and Lewes are used in this report to illustrate the trends across the state for specific climate indicators.
The most common analysis technique used to identify the existence of statistically significant trends is simple linear regression. This technique describes the linear relationship between two variables, typically time and the meteorological variable of interest. Linear regression techniques were used in the current study to determine the existence of statistically significant trends between the independent variable (time) and the dependent climatological variable. A variety of statistical tests were used to determine the significance of the relationship.
An examination of Delaware statewide mean annual and mean seasonal temperatures shows a statistically significant increasing trend during the period 1895 through 2012 annually and for all seasons. An increasing trend of 0.2°F per decade was identified for annual, winter, spring, and summer temperatures (Table 1). Autumn season temperatures have seen a significant increase, at a rate of 0.1°F per decade. A modest increasing trend in statewide mean annual temperatures is detectable before 1960, with a more apparent trend after that year. Individual seasons show a more monotonic long-term upward trend from 1895 through the present. Significant increasing trends were found for statewide cooling degree-days annually and during the summer, and significant decreasing trends were found in heating degree-days for all seasons except winter. These results are expected, because cooling- and heating-degree day data are calculated directly from mean temperature statistics (Table 1).
Several temperature-dependent climate indicators also display statistically significant trends during the period of record. These include growing season length, the annual number of days with minimum temperatures below 32°F and 20°F, the annual number of days with minimum temperatures above 75°F, and seasonal mean maximum and minimum temperatures. Four of the Cooperative stations have data that extend into the most recent decade. Of these four, three (Wilmington Porter Reservoir, Dover, and Lewes) show significant rising trends in growing season length associated with an earlier “last freeze” date in the spring and a later “first freeze” date in the fall. All four stations (Wilmington Porter Reservoir, Wilmington Airport, Dover, and Lewes) show significant decreasing trends in the number of days with minimum temperatures below freezing. Days with minimum temperatures below 20°F, have seen significant reductions at both Wilmington Porter Reservoir and Lewes. For minimum temperatures greater than 75°F, Wilmington Porter Reservoir, Wilmington Airport, and Lewes all show significant increasing trends in warm nighttime temperatures. An examination of mean seasonal temperatures at the four high-quality Cooperative stations shows statistically significant increasing trends in seasonal mean minimum temperatures for each season. Increasing trends in seasonal mean maximum temperatures were apparent for several stations across the seasons. The majority of these increasing trends were associated with temperature increases early in the 20th century. There has been less warming of maximum temperatures in recent decades.
No significant trends were identified in annual, winter, spring, or summer season statewide precipitation for the period 1895-2012. Only autumn season statewide precipitation was found to have a statistically significant increasing trend of 0.27” per decade. During the observational record since 1895, the most noteworthy characteristic of Delaware precipitation has been large annual and seasonal precipitation variability. For example, Statewide annual precipitation values have varied between 28.29” in 1930 and 62.08” in 1948. In addition, there has been a tendency for decadal-scale variations in annual precipitation, including a continuously wet period from 1932 through 1939 and an exceptionally dry period during the 1960s.
Cooperative Station Results
An examination of Cooperative station precipitation data shows no significant long-term trends in any of the climate indicators based on daily precipitation thresholds. Only a few of the nine stations for which data were examined show significant trends for any precipitation variables. The exception being the autumn season in which three stations of the nine showed significant increasing trends in precipitation.
|Statewide Temperature||Statewide Precipitation||Statewide HDD||Statewide CDD|
The analysis of historical temperature data suggests that temperatures across Delaware have been increasing at a rate of approximately 0.2°F per decade since 1895. An analysis of temperature data from nine Cooperative stations, used in the statewide values, suggests that much of the long-term trend in annual and seasonal temperatures is being driven by increasing minimum temperatures, especially later in the period of record. Mean seasonal maximum temperatures have increased at many stations, but the primary focus of that warming has bee earlier in the 20th century at most locations. The analysis also shows that cold days, with minimum temperatures below 32°F and 20°F (cold nighttime low temperatures), are decreasing, while days with minimum temperatures above 75°F (warm nighttime low temperatures) are increasing in recent decades. Put more simply, nighttime low temperatures are asymmetrically increasing across Delaware compared with daytime maximum temperatures, especially in the most recent decades.
Statewide precipitation has shown no significant changes since 1895, except for a significant upward increasing trend during the autumn season. The most distinguishing characteristic of precipitation across Delaware is large interannual and intra-annual variability. An analysis of Cooperative station precipitation data indicated no homogeneous trends in precipitation thresholds for the stations analyzed, except for an upward increasing trend in precipitation at several locations during the autumn season.