Portal de Periódicos da CAPES

Observed soil temperature trends associated with climate change in Canada

2011; American Geophysical Union; Volume: 116; Issue: D2 Linguagem: Inglês

10.1029/2010jd015012

2156-2202

Budong Qian, E. G. Gregorich, Sam Gameda, D. W. Hopkins, Xiaolan L. Wang,

Cryospheric studies and observations

Journal of Geophysical Research: AtmospheresVolume 116, Issue D2 Climate and DynamicsFree Access Observed soil temperature trends associated with climate change in Canada Budong Qian, Budong Qian [email protected] Eastern Cereal and Oilseed Research Centre, Research Branch, Agriculture and Agri-Food Canada, Ottawa, Ontario, CanadaSearch for more papers by this authorEdward G. Gregorich, Edward G. Gregorich Eastern Cereal and Oilseed Research Centre, Research Branch, Agriculture and Agri-Food Canada, Ottawa, Ontario, CanadaSearch for more papers by this authorSam Gameda, Sam Gameda Eastern Cereal and Oilseed Research Centre, Research Branch, Agriculture and Agri-Food Canada, Ottawa, Ontario, CanadaSearch for more papers by this authorDavid W. Hopkins, David W. Hopkins Scottish Crop Research Institute, Dundee, UK School of Biological and Environmental Sciences, University of Stirling, Stirling, UKSearch for more papers by this authorXiaolan L. Wang, Xiaolan L. Wang Climate Research Division, Science and Technology Branch, Environment Canada, Toronto, Ontario, CanadaSearch for more papers by this author Budong Qian, Budong Qian [email protected] Eastern Cereal and Oilseed Research Centre, Research Branch, Agriculture and Agri-Food Canada, Ottawa, Ontario, CanadaSearch for more papers by this authorEdward G. Gregorich, Edward G. Gregorich Eastern Cereal and Oilseed Research Centre, Research Branch, Agriculture and Agri-Food Canada, Ottawa, Ontario, CanadaSearch for more papers by this authorSam Gameda, Sam Gameda Eastern Cereal and Oilseed Research Centre, Research Branch, Agriculture and Agri-Food Canada, Ottawa, Ontario, CanadaSearch for more papers by this authorDavid W. Hopkins, David W. Hopkins Scottish Crop Research Institute, Dundee, UK School of Biological and Environmental Sciences, University of Stirling, Stirling, UKSearch for more papers by this authorXiaolan L. Wang, Xiaolan L. Wang Climate Research Division, Science and Technology Branch, Environment Canada, Toronto, Ontario, CanadaSearch for more papers by this author First published: 21 January 2011 https://doi.org/10.1029/2010JD015012Citations: 114AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Abstract [1] Trends in soil temperature are important, but rarely reported, indicators of climate change. On the basis of the soil temperature data from 30 climate stations across Canada during 1958–2008, trends in soil temperatures at 5, 10, 20, 50, 100, and 150 cm depths were analyzed, together with atmospheric variables, such as air temperature, precipitation, and depth of snow on the ground, observed at the same locations. There was a significant positive trend with soil temperatures in spring and summer means, but not for the winter and annual means. A positive trend with time in soil temperature was detected at about two-thirds of the stations at all depths below 5 cm. A warming trend of 0.26–0.30°C/decade was consistently detected in spring (March–April–May) at all depths between 1958 and 2008. The warming trend in soil temperatures was associated with trends in air temperatures and snow cover depth over the same period. A significant decreasing trend in snow cover depth in winter and spring was associated with increasing air temperatures. The combined effects of the higher air temperature and reduced snow depth probably resulted in an enhanced increasing trend in spring soil temperatures, but no significant trends in winter soil temperatures. The thermal insulation by snow cover appeared to play an important role in the response of soil temperatures to climate change and must be accounted for in projecting future soil-related impacts of climate change. 1. Introduction [2] The Intergovernmental Panel on Climate Change (IPCC) has reported that the 100 year linear warming trend (1906–2005) of 0.74°C [0.56–0.92°C] in global surface temperature is larger than the corresponding trend of 0.6°C [0.4–0.8°C] for 1901–2000 given in the IPCC TAR (Third Assessment Report) [Intergovernmental Panel on Climate Change (IPCC), 2007]. The linear warming trend over the 50 years from 1956 to 2005 (0.13°C [0.10–0.16°C] per decade) is nearly twice that for the 100 years from 1906 to 2005. This temperature increase is widespread over the globe and is greater at higher latitudes. Furthermore, land has warmed faster than the oceans. Zhang et al. [2000] found a distinct pattern of temperature trends across Canada for the period 1950–1998, showing warming in the south and west, and cooling in the northeast; these trends were most evident in winter and spring. They also found that precipitation had increased by 5–35% across Canada, although significant negative trends occurred in southern regions during winter. The ratio of snowfall to total precipitation also increased overall, but significant negative trends in the ratio occurred mostly in southern Canada during spring. Decreases in snow cover were observed over a large, well-defined area extending from the west coast of Canada, across the southern Prairies and northern Great Plains, to the Great Lakes [Brown and Goodison, 1996]. Decreased snow cover occurred in the second half of the snow cover season, while the spring temperatures were significantly higher in the Prairies [Brown and Goodison, 1993]. The strong positive feedback between spring snow cover and temperature through the radiative balance [Groisman et al., 1994] suggests that decreased snow cover has an important role in the process of climate warming. Moreover, projected warming in the 21st century shows scenario-independent geographical patterns similar to those observed over the past several decades. Warming is expected to be greatest over land and at higher latitudes, continuing recent observed trends [IPCC, 2007]. [3] Soil temperatures have not been analyzed as much as other climate variables, such as air temperature and precipitation, because data are not widely available for either spatial or temporal coverage. Zhang et al. [2001] analyzed long-term changes over the 1890s to 1990s in soil temperatures at Irkutsk, Russia, finding that soil temperature at 40 cm depth increased by up to 9°C during the winter, while air temperature increased about 4–6°C. They suggested that an increase in snowfall during early winter (October and November) and early snowmelt in spring might play a major role in the increase of soil temperatures through the effects of insulation and albedo changes. García-Suárez and Butler [2006] showed an increasing trend of 0.04–0.25°C/decade in soil temperatures over the past century for three sites in Ireland. Du et al. [2007] found an even greater trend (0.43–0.66°C/decade) in seasonal mean soil temperatures at near-surface layers (top 40 cm) observed at Lhasa in Tibet, China, during 1961–2005, especially in spring. They also indicated that the increasing trend in soil temperatures was stronger than that for air temperature. [4] Ground surface temperature (GST), based on borehole temperature, has been used as a proxy to reconstruct surface air temperature (SAT) [Mann and Schmidt, 2003]. In light of the importance of soil temperature at shallow depths as the source of deep soil temperature variations, Hu and Feng [2005] examined how soil temperature has been affected by the surface air temperature and precipitation in the Eurasian continent, based on historical soil temperature data at 153 stations in Russia, Mongolia, and China. At some locations it was found that the relationship between air and ground temperatures was not as straightforward as previously thought [Beltrami and Kellman, 2003; Beltrami et al., 2003, 2006]. Woodbury et al. [2009] examined long-term SAT and GST changes since the 1960s at eight sites west of the Canadian cordillera, concluding that GST observations showed no obvious climate-induced perturbations, even though all sites showed significant increasing trends in SAT. Their comparison of GST and SAT temperatures suggested that any trend in increased SAT temperatures was masked by freeze–thaw and latent energy effects in the winter and spring. [5] As an alternative to soil temperature observations, simulated soil temperatures derived from process-based models have been used to study the change in soil temperatures in response to climate change as characterized by warmer air temperatures and precipitation variability. Isard et al. [2007] studied soil temperature trends in the Great Lakes region by examining the simulated vertical profiles of soil water content and temperature, calculated using a modified form of a soil water and temperature algorithm [Schaetzl and Isard, 1991, 1996; Isard and Schaetzl, 1993, 1995]. They modeled soil temperatures at 50 cm in the soil profile, using 1951–2000 air temperature and precipitation data from 194 National Weather Service stations in Wisconsin and Michigan. Their modeling results suggested that, even though winter air temperatures increased, winter soil temperatures decreased, especially at sites downwind from the Great Lakes, many of which are in snowbelt locations. Rising winter air temperatures over the past 50 years coincide with (and probably have led to) more variable and thinner snowpacks, lessening their insulating effect and contributing to decreasing winter time soil temperatures. [6] Zhang et al. [2003] developed a process-based model of northern ecosystem soil temperature (NEST) to simulate the transient response of soil thermal regime to climate change in Canada. Their results [Zhang et al., 2005] show that, depending on the location, changes in annual mean soil temperature during the 20th century differed from those in air temperature by −3°C to +3°C, and that the difference was more significant in winter and spring than in summer. They found that on average, for the whole of Canada, the annual mean soil temperature at 20 cm depth increased by 0.6°C while the annual mean air temperature increased by 1.0°C. [7] Changes in soil temperature as a result of a warmer climate will have profound effects on terrestrial ecosystems. Increases in soil temperatures will likely thaw permafrost in high latitudes, change the distribution and growth of plants, including agricultural crops, and enhance decomposition of soil organic matter. Permafrost melting, enhanced decomposition and some changes in plant growth generate greater CO2 emissions from the soil to the atmosphere, thereby setting up a positive feedback to climate change [Trumbore et al., 1996; Goulden et al., 1998; Nelson, 2003; Davidson and Janssens, 2006]. [8] Given the evidence from several studies for increasing soil temperatures during the second half of the 20th century, it is now important to understand the controls and feedbacks on soil temperature, how it interacts with other climate variables, and quantify the magnitude of the increases. Here we present trends observed from soil temperatures at six soil depths: 5, 10, 20, 50, 100 and 150 cm, measured at 30 stations across Canada for the period 1958–2008. Trends in other climate variables, such as air temperature, precipitation, and depth of snow cover at the same locations, were also observed for the same period. We also examined the associations between trends in soil temperature and other climate variables in Canada to better understand the effects of future climate change on soils and the associated biophysical and biochemical processes. 2. Data and Methods Soil Temperature and Other Climate Data [9] Soil temperature is not observed as widely as other climate variables across Canada. Measurement of soil temperature started as early as 1958, with 62 stations recording these data in 1990. In contrast, air temperature and precipitation were recorded at more than 2000 stations that year, and records go back as far as 1840 [Phillips, 1990]. [10] For this study, Environment Canada provided daily soil temperature data at depths of 5, 10, 20, 50, 100, and 150 cm, recorded at about 80 stations across Canada, beginning between 1958 and 1991 and ending between 1969 and 2008. Almost all stations had some missing data so we selected 30 stations from this data set (Figure 1) which the criterion of having soil temperature data for at least 25 years between 1958 and 2008. The stations are not evenly distributed across Canada, but they represent Canada's regional climate regimes well in the west coast, Prairies, central Canada, and Atlantic Canada, with one station in the Arctic (Resolute). Although soil temperatures were recorded twice daily, morning (0800 h) and afternoon (1500 h), we used only morning observations to calculate monthly mean soil temperatures because there were more missing data in the afternoon data. A monthly value was retained if soil temperature readings were available for at least 25 d in the month; otherwise, a missing value was assigned for the month. Seasonal and annual means were calculated from monthly values if all monthly values were available and, as above, missing values were assigned when there were monthly values missing. For interpreting the results it is important to note that most of the stations are located in southern Canada and that only morning observations were used in the analyses. Figure 1Open in figure viewerPowerPoint Map showing the 30 stations with soil temperature data across Canada (Ottawa is at 45°23′N, 75°43′W). [11] Monthly means of daily maximum and minimum air temperatures, monthly rainfall, snowfall, total precipitation, and snow cover depth at the 30 stations were derived from daily climate data archived at Agriculture and Agri-Food Canada (AAFC) and provided by Environment Canada for 1958–2008. Seasonal and annual series were then created from monthly data. Homogenization of the Data Series [12] Homogeneity of climate data is often a concern, especially in detecting and analyzing long-term trends. For example, it has been noticed that linear-trend estimates are reliable only when the data series are homogeneous in time [Easterling and Peterson, 1995; Lu et al., 2005; Hanesiak and Wang, 2005; Wang, 2006]. A lack of data homogeneity is often related to discontinuities, or change points, in climate data series, which can be introduced by changes in measurement instruments, observation locations, and measurement practices and procedures. Some change points can be identified from climate metadata, and are referred to as documented shifts; however, metadata are often incomplete and inaccurate. Change points may still exist in climate data that cannot be identified from metadata; these are referred to as undocumented shifts [Wang et al., 2007]. Since no metadata were available for this study, we first had to detect undocumented shifts in the climate data series and then adjust the data series to eliminate the detected shifts by using statistical methods. Several statistical methods have been proposed for detecting undocumented shifts [Reeves et al., 2007]. We used a statistical algorithm of the penalized maximal F test accounting for autocorrelation (PMFred) [Wang, 2008a, 2008b; Wang and Feng, 2010]. Statistical Analysis [13] A nonparametric Kendall's tau-based slope estimator [Sen, 1968] was used to estimate trends. An iterative procedure, originally proposed by Zhang et al. [2000] and refined by Wang and Swail [2001], was adopted to take the effect of lag-1 autocorrelation into account when testing the significance level of a trend. A p value was obtained in each test, and a trend was considered to be statistically significant if the p value was smaller than the significance level α (e.g., 0.05). We maximized the power conditional on the Type I error rate being at or below some level α when testing a single hypothesis. However, a compound error rate should be controlled when multiple testing is conducted [Storey, 2002]. In our study, the same test was conducted at different locations. The null hypothesis that there is no trend in a series can be rejected at a rate higher than the specified significance level α even if the hypothesis is true in this case. Therefore, we applied a false discovery rate (FDR) controlling procedure [Benjamini and Hochberg, 1995] when a statistical test was performed at multiple sites. In this procedure, considering testing H1, H2, …, Hm based on the corresponding p values P1, P2, …, Pm, let P(1) ≤ P(2) ≤ … ≤ P(m) be the ordered p values, and denoted by H(i) the null hypothesis corresponding to P(i). To define the following Bonferroni-type multiple-testing procedure, let k be the largest i for which P(i) ≤ (i/m)q, then reject all H(i) i = 1, 2, …, k for the FDR at q. We chose the FDR at q = 0.05. [14] Correlation analysis was performed to identify the relationships between soil temperature variability and other climate variables at the same locations. The FDR controlling procedure was also taken in the tests of correlation significance. 3. Results and Discussion [15] Change points were not very common in the data; they were detected in less than one-third of the monthly data series tested, for soil temperature, air temperature and precipitation. To account for the potential effects of these change points, trend analysis and correlation analysis were applied to the original data series and those that were adjusted. Significant trends were detected at slightly more of the stations in some cases when the adjusted series were used. However, our conclusions are not different whether the original or adjusted data series were used. Here we present only results from the adjusted data, but it is important to note that most data series were not adjusted, because change points were not detected in the two-thirds of the data series. Trends in Soil Temperatures [16] Only a few stations had sufficient data to estimate trends over the period 1958–2008 in annual mean soil temperature and there were only 8 stations with data at the 5 cm depth (Table 1). It appears however that seasonal mean soil temperatures increased in spring (March–April–May), especially in surface layers (e.g., 10 cm) (Table 1). A positive trend in soil temperature over time was estimated at about two-thirds of the stations at all depths from 5 to 150 cm. A map of the trends across Canada of spring mean soil temperature at 10 cm depth is shown in Figure 2a. Significant soil warming was found at six stations across the country, but a significant soil cooling was seen in Nova Scotia, in the Maritimes. Soil cooling was observed at more locations in eastern Canada than anywhere else; this observation is related to several factors, one of the most important of which is reduced snow cover in this region (discussed below). A similar distribution of trends is shown in Figure 2b for spring soil temperature at 100 cm, but with fewer significant trends. Soil warming during summer (June–July–August) was more often found in deeper layers (at 100 and 150 cm depths, Figures 2c and 2d) than near the surface, compared to the spring (Figures 2a and 2b). This implies that the warming occurring in surface soil layers could take some time to reach the deeper soil layers, but once warmed, the deeper layers remain warmer later in the year. Significant trends were found at fewer stations in autumn (September–October–November) (Table 1). Figure 2Open in figure viewerPowerPoint Maps showing the trends across Canada of spring (March–April–May) mean soil temperature at the depths of (a) 10 cm and (b) 100 cm and summer (June–July–August) at the depths of (c) 100 cm and (d) 150 cm for the period 1958–2008. Upward and downward triangles show positive and negative trends, respectively. Solid triangles indicate trends significant at the 5% level. Table 1. Number of Stations Having Statistically Significant Trends in Annual and Seasonal Soil Temperature at Different Depths During 1958–2008 Time Perioda Soil Depth 5 cm 10 cm 20 cm 50 cm 100 cm 150 cm Annual N 8 18 17 18 16 14 NPT 4 12 9 8 10 9 NST 4 3 0 2 4 0 NSPT 3 2 0 1 4 0 Dec, Jan, Feb N 22 27 26 27 28 26 NPT 13 14 11 14 13 15 NST 0 0 0 3 2 4 NSPT 0 0 0 1 2 1 Mar, Apr, May N 21 27 27 26 28 26 NPT 15 20 20 19 20 17 NST 4 7 3 3 3 5 NSPT 3 6 2 1 3 3 Jun, Jul, Aug N 22 27 27 25 27 25 NPT 15 20 18 15 18 18 NST 0 4 4 5 8 8 NSPT 0 4 3 2 8 7 Sep, Oct, Nov N 21 29 28 26 29 26 NPT 14 19 15 11 17 16 NST 0 4 2 2 3 2 NSPT 0 2 1 1 3 2 a N, number of stations (out of 30) available for trend analysis; NPT, number of stations with a positive (either statistically significant or not); NST, number of stations with a trend (positive or negative) statistically significant at the 0.05 level; NSPT, number of stations with a significant positive trend. [17] Examining monthly mean soil temperatures, a significant trend was detected at more stations in March, April, and May than other months at the near surface layers, as well as in May, June, July, and August in the deeper layers (Table 2). This is consistent with the results from seasonal means. A positive trend in soil temperature at 5 cm was observed at most locations across the country in March (Figure 3a). However, a significant negative trend was also seen in eastern Canada. Several factors may contribute to this observation. The depth of snow cover affects the near-surface soil temperature because of high albedo, insulation effects, and soil-water content related to snowmelting. The cooling trends could also be associated with declining air temperature in eastern Canada observed by Zhang et al. [2000]. Significant soil warming in April soil temperatures at 10 and 20 cm depths were found mostly in western Canada (Figures 3b and 3c). A significant soil warming trend in May at 10 cm was also found at stations in eastern Canada (Figure 3d). Nevertheless, a significant warming was detected at more stations across the country at deeper layers, for example, in May and June at 100 and 150 cm depths (Figures 3e–3h). It is possible that a significant trend in soil temperatures may be more easily detected at deeper soil layers than at near-surface layers because soil temperatures at deeper layers are less variable. Figure 3Open in figure viewerPowerPoint Trends of monthly mean soil temperatures in (a) March at 5 cm, (b) April at 10 cm and (c) April at 20 cm, (d) May at 10 cm, (e) May at 100 cm, (f) May at 150 cm, (g) June at 100 cm, and (h) June at 150 cm for the period 1958–2008. Upward and downward triangles show positive and negative trends, respectively. Solid triangles indicate trends significant at the 5% level. Table 2. Number of Stations Having Statistically Significant Trends in Monthly Soil Temperature Across Canada During 1958–2008a Time Periodb Soil Depth 5 cm 10 cm 20 cm 50 cm 100 cm 150 cm Jan N 24 28 28 28 29 28 NPT 13 14 16 14 12 15 NST 3 0 0 2 2 5 NSPT 2 0 0 1 2 2 Feb N 24 29 30 30 30 29 NPT 13 16 13 15 16 15 NST 5 3 3 4 7 5 NSPT 2 2 2 2 6 2 Mar N 24 29 30 29 30 28 NPT 16 18 21 16 17 15 NST 7 5 4 5 8 6 NSPT 4 3 2 3 6 3 Apr N 24 28 29 27 29 27 NPT 17 23 23 18 19 17 NST 4 5 6 8 10 7 NSPT 4 5 5 6 8 4 May N 24 28 28 27 29 27 NPT 17 20 20 20 21 20 NST 3 5 2 7 10 11 NSPT 1 5 2 5 10 9 Jun N 25 29 29 28 29 27 NPT 15 19 19 19 25 23 NST 0 2 6 4 8 9 NSPT 0 2 4 2 8 8 Jul N 24 27 27 27 27 25 NPT 15 17 17 16 19 18 NST 0 0 4 5 10 5 NSPT 0 0 1 2 9 3 Aug N 23 29 27 28 28 26 NPT 16 19 17 15 21 18 NST 0 2 2 5 11 2 NSPT 0 2 1 3 10 2 Sep N 23 29 28 26 30 27 NPT 15 23 17 14 20 18 NST 2 0 2 2 6 3 NSPT 1 0 1 1 6 2 Oct N 24 29 30 28 30 28 NPT 14 16 17 15 17 18 NST 5 0 2 4 3 4 NSPT 4 0 1 2 3 3 Nov N 23 29 30 28 30 28 NPT 13 18 15 14 17 18 NST 2 3 4 2 2 4 NSPT 2 2 2 1 2 3 Dec N 23 28 28 28 29 27 NPT 15 16 12 14 18 13 NST 0 2 4 5 4 4 NSPT 0 2 2 4 4 2 a Out of 30 stations. b N, number of stations (out of 30) available for trend analysis; NPT, number of stations with a positive (either statistically significant or not); NST, number of stations with a trend (positive or negative) statistically significant at the 0.05 level; NSPT, number of stations with a significant positive trend. [18] The median values of the trends across the country in seasonal and annual mean soil temperatures are shown in Table 3. The median values may be more representative than the means with respect to the magnitude of the trends, as they are less influenced by an occasional extreme values. A warming rate of 0.26–0.30°C/decade was consistently seen in spring (March–April–May) at all depths (5, 10, 20, 50, 100, and 150 cm). The warming rate was even higher when evaluated on a monthly scale, e.g., the median value of monthly mean soil temperatures ranged from 0.27 to 0.47°C/decade at different depths in May (Table 4). Table 3. Median of the Magnitudes of the Trends Estimated From Seasonal and Annual Mean Soil Temperature at Different Depths and Daily Maximum and Minimum Air Temperatures (Tmax and Tmin) Across Canada During 1958–2008 Time Perioda Soil Depth Air Temperatures 5 cm 10 cm 20 cm 50 cm 100 cm 150 cm Tmax Tmin Annual N 8 18 17 16 18 14 30 30 M 0.10 0.08 0.01 0.16 −0.07 0.11 0.16 0.28 Dec, Jan, Feb N 22 27 26 28 27 26 30 30 M 0.17 0.07 −0.15 −0.07 0.01 0.04 0.29 0.48 Mar, Apr, May N 21 27 27 28 26 26 30 30 M 0.30 0.27 0.28 0.29 0.26 0.29 0.35 0.39 Jun, Jul, Aug N 22 27 27 27 25 25 30 30 M 0.30 0.28 0.22 0.34 0.07 0.30 0.02 0.23 Sep, Oct, Nov N 21 29 28 29 26 26 30 30 M 0.12 0.15 0.03 0.09 −0.04 0.19 −0.02 0.00 a N, number of stations (out of 30) with sufficient data to estimate a trend; M, the median magnitude (°C/decade) of the trends from all available stations. Table 4. Median of the Magnitudes of the Trends Estimated From Monthly Mean Soil Temperature at Different Depths and Daily Maximum and Minimum Air Temperatures (Tmax and Tmin) Across Canada During 1958–2008 Montha Soil Depth Air Temperatures 5 cm 10 cm 20 cm 50 cm 100 cm 150 cm Tmax Tmin Jan N 24 28 28 28 29 28 30 30 M 0.25 0.03 0.06 0.02 −0.05 0.07 0.34 0.69 Feb N 24 29 30 30 30 29 30 30 M 0.04 0.10 −0.05 0.00 0.06 0.07 0.34 0.60 Mar N 24 29 30 29 30 28 30 30 M 0.18 0.25 0.10 0.09 0.20 0.08 0.40 0.52 Apr N 24 28 29 27 29 27 30 30 M 0.29 0.34 0.26 0.26 0.22 0.19 0.44 0.32 May N 24 28 28 27 29 27 30 30 M 0.28 0.37 0.32 0.36 0.47 0.34 0.17 0.33 Jun N 25 29 29 28 29 27 30 30 M 0.28 0.40 0.28 0.24 0.30 0.37 0.01 0.23 Jul N 24 27 27 27 27 25 30 30 M 0.14 0.19 0.19 0.17 0.32 0.27 −0.06 0.16 Aug N 23 29 27 28 28 26 30 30 M 0.25 0.18 0.07 0.17 0.30 0.24 0.11 0.17 Sep N 23 29 28 26 30 27 30 30 M 0.18 0.20 0.16 0.08 0.15 0.21 0.08 0.10 Oct N 24 29 30 28 30 28 30 30 M 0.10 0.07 0.04 0.03 0.13 0.17 −0.16 −0.13 Nov N 23 29 30 28 30 28 30 30 M 0.27 0.09 −0.01 −0.01 0.09 0.10 −0.09 0.10 Dec N 23 28 28 28 29 27 30 30 M 0.17 0.04 −0.02 0.01 0.10 0.00 0.18 0.31 a N, number of stations (out of 30) with sufficient data to estimate a trend; M, the median magnitude (°C/decade) of the trends. Trends in Other Climate Variables [19] Significant increases were detected more often in annual mean daily minimum air temperatures (Tmin) than in daily maximum air temperatures (Tmax) (Table 5). The significant increase in Tmin was often found in spring and summer (Figure 4). These findings are consistent with those of Zhang et al. [2000], who observed that air temperatures were higher everywhere in the country except for northern Quebec and Labrador, where it was significantly cooler. Although no significant trends have been detected in snowfall, a negative trend was seen in most parts of the country in the annual and the seasonal snowfall totals in winter and spring, accompanied by a decrease in the average snow depth (Table 5), which occurred almost everywhere across the country (Figure 5). A trend could not be estimated in regions where snow did not accumulate on the ground in most years, such as the west coast and the southernmost part of Ontario. Figure 4Open in figure viewerPowerPoint Trends in annual mean (a) daily minimum air temperature (Tmin) and (b) daily maximum air temperature (Tmax) and (c) spring (March–April–May) mean Tmin and (d) summer (June–July–August) mean Tmin for the period 1958–2008. Upward and downward triangles show positive and negative trends, respectively. Solid triangles indicate trends significant at the 5% level. Figure 5Open in figure viewerPowerPoint Trends in seasonal mean snow depth on ground in (a) winter (December–January–February) and (b) spring (March–April–May) and monthly mean in (c) February and (d) March for the period 1958–2008. Upward and downward triangles show positive and negative trends, respectively. Soli