Long-term trends in air temperature and precipitation under climate change were analyzed for two meteorological stations on the Island of Montreal: McGill (1872–1986) and Pierre-Elliott-Trudeau (P-E-T, formerly Dorval) Airport (1942–2014). A linear trendline analysis, the Mann–Kendall (MK) test and the two-sample Kolmogorov–Smirnov (KS) test were conducted to assess specific climate trends. On a 100-year basis, temperature increased 1.88°C (34%) and 1.18°C (19%) at the McGill and P-E-T Airport sites, respectively, while annual rainfall increased 23.9 mm y−1 (2.3%) and 138.8 mm y−1 (15%) over the same period. The frequency of 50% (every other year) and 95% (every year) annual maximum daily rainfall events showed decreasing trends for the McGill station, but increasing trends for the P-E-T Airport station. Growing degree-days and growing season length are prone to being influenced by climate change and are critical to managing agricultural activities in the Montreal region; both showed increasing trends. At the same time, the onset of the growing season occurred earlier as time progressed.

ABBREVIATIONS

     
  • GDD

    Growing degree-days

  •  
  • GSE

    Growing season end

  •  
  • GSL

    Growing season length

  •  
  • OGS

    Onset of the growing season

  •  
  • KS

    Kolmogorov–Smirnov

  •  
  • MK

    Mann–Kendall

  •  
  • P-E-T

    Pierre-Elliott-Trudeau

INTRODUCTION

Since the pre-industrial era, greenhouse gas emissions have driven large increases in the atmospheric concentrations of carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O), resulting in extreme weather and climate events, as well as changes in temperature and precipitation (IPCC 2014). According to Berry (1991), model simulations suggest that global temperature increases of 0.4 to 1.3°C should already have occurred in Canada in the past 100 years because of changes in atmospheric composition. Drawing on 50 years of weather data from the South Nation watershed in eastern Ontario, Alodah (2015) found that temperatures had increased in all seasons, while precipitation has increased significantly in the spring, summer and fall, but has decreased in the winter. Ouranos (2015) reported that, over a 62-year term (1950–2011), the annual mean temperature recorded across all regions of Quebec rose by 1.0 to 3.0°C, and that during the same period both maximum and minimum temperatures increased.

Analyzing trends in air temperature and precipitation for Canada and the northeastern United States, Gan (1995) observed that southeastern Canada was experiencing a cooling trend with a strong correlation in October and limited correlation in January and June. Unlike temperature data, precipitation trends were inconsistent, showing obvious patterns. The impacts of climate change on agriculture will differ across regions, in part due to regional variation (Dai et al. 2015). Depending on the location, increasing food demand and changes in water quantity may lead to agricultural production issues and food security may be put at risk.

When comparing or analyzing climate data, it is of great importance to understand the concept of natural variability: small short-term variations (over a few years) are accounted for by natural variability, whereas long-term changes (multiple decades to centuries) are a consequence of climate change (CCDS 2014). Climate change studies aim to provide useful tools for governments, institutions, non-governmental organisations, scientists and the public to understand and predict climate change in order to take appropriate measures to reduce its consequences and better prepare for inevitable future events. Therefore, the results of this study could be a reference for the planning and management of agricultural activities and water resources in the Montreal region under climate change.

For instance, a change in precipitation (increase or decrease) could have a positive or negative effect on agricultural productivity depending on whether the timing, intensity or frequency (or a combination thereof) of precipitation is affected. Changes in rainfall during the growing season may require the installation of appropriate drainage or irrigation systems to reduce the impact of climate change on crops. In fact, intense rainfall events can seriously damage crops due to nutrient leaching, a heightened threat of disease, oxygen depletion and risk of soil waterlogging which may, in turn, increase the risk of compaction (Takle & Hofstrand 2008). In addition, high precipitation in spring may delay planting, which becomes a problem for crops planted early, like corn (Zea mays L.).

The objectives of this study were to: (i) assess whether changes in temperature and precipitation have occurred in the immediate Montreal region through an analysis of historical weather station data from Montreal over recent years, (ii) identify trends occurring on a seasonal or annual basis and (iii) evaluate changes in the growing season. Past climate change trends have been evaluated for Canada, southeastern Canada and the Ottawa region, but these studies were either not recent, did not present long-term analyses or were not specific to Montreal. The present study provides a useful basis of knowledge for further development involving system modelling in order to better evaluate potential impacts of climate change and make recommendations for agriculture, food security, urban drainage and water quality, among other things, for areas around Montreal. More specifically, the rate of changes in rainfall and temperature found from the historical data analysis can be used to predict future climate data (by extracting data); thus, adaptation of new crop or cultivars can be evaluated based on this predicted future weather data.

The weather data available for the two weather stations (McGill University, downtown Montreal, QC, and P-E-T Airport, Dorval, QC) from which data were drawn were unique in terms of their quality, homogeneity and length. Therefore, despite the spatial variability of weather data, it is believed that this study can be useful in providing a guideline for all agricultural regions in Montreal and in its surroundings. On the island of Montreal itself there remain active farmlands, particularly on the western end of the island. McGill University's 205 ha Macdonald Campus Farm (McGill 2016), an experimental and demonstration farm, situated in Sainte-Anne-de-Bellevue, at the western tip of the island, is dedicated to research but also sells its products locally, as well as providing farm-fresh produce to university food establishments. Moreover, 4% of the total area of the island of Montreal, some 2,000 ha, constitutes a permanent agricultural zone. These farmlands are mainly located in the western portion of Montreal's Pierrefonds-Roxboro and l’Île Bizard-Sainte-Geneviève boroughs (Ville de Montréal 2015), and in the independent towns of Seneville and Sainte-Anne-de-Bellevue.

MATERIALS AND METHODS

Study area and weather data

Daily temperature and precipitation data, recorded in °C and mm respectively, were obtained from two climatological stations on Montreal island: Dorval/P-E-T Airport (P-E-T; Dorval, QC), and McGill (Ville-Marie borough, Montreal, QC) (Table 1, Figure 1).
Table 1

Location and ID no. of climatological stations from which data was obtained

  Location
 
  ID no.
 
Name City Latitude (N) Longitude (W) Elevation (m AMSL)   Station WMO Canadian climate 
Dorval/P-E-T Airport Dorval, QC 45°28′00″ 73 °45′00″ 35.97  CYUL 71627 7025250 
McGilla Montreal, QC 45 °30′16″
45 °30′17″ 
73 °34′41″
73 °34′30″ 
56.90
50.00 
1964
>1964 
— — 7025280 
Sainte Anneb Sainte-Anne-de-Bellevue, QC 45 °25′38″ 73 °55′45″ 39.00  1993
≥1994 

CWVQ 

71377 
7026839
702FHL8 
  Location
 
  ID no.
 
Name City Latitude (N) Longitude (W) Elevation (m AMSL)   Station WMO Canadian climate 
Dorval/P-E-T Airport Dorval, QC 45°28′00″ 73 °45′00″ 35.97  CYUL 71627 7025250 
McGilla Montreal, QC 45 °30′16″
45 °30′17″ 
73 °34′41″
73 °34′30″ 
56.90
50.00 
1964
>1964 
— — 7025280 
Sainte Anneb Sainte-Anne-de-Bellevue, QC 45 °25′38″ 73 °55′45″ 39.00  1993
≥1994 

CWVQ 

71377 
7026839
702FHL8 

aThe climatological station moved roughly 200 m in 1964.

bClimatological station until 1994, thereafter a regularly reporting automatic station.

Figure 1

Map of Montreal Island showing the location of the three stations: 1 McGill, 2 Dorval/P-E-T, 3 Sainte-Anne-de-Bellevue (Google Maps).

Figure 1

Map of Montreal Island showing the location of the three stations: 1 McGill, 2 Dorval/P-E-T, 3 Sainte-Anne-de-Bellevue (Google Maps).

For the McGill station, data were available from 1872 to 1986 (115 years), while from Dorval/P-E-T Airport (Dorval) data were available from 1942 to 2014 (74 years) (Government of Canada 2015). It is important to point out that these two stations did not move significantly over the years, as this reduces the margin of error and station non-homogeneity.

Weather station data may be missing for a number of causes, the most frequent being the breakage or malfunction of instruments over a certain time period (Allen et al. 1998). However, the two selected stations have very consistent data, such that only a few days' data were missing over the many decades of recording. In this case, it is appropriate to replace the missing data using nearby stations and a correction factor (Table 2). The correction factor was calculated from the average difference between the yearly data over a common period of time. The missing data for the McGill station was replaced by data from the P-E-T station. Similarly, the missing data for P-E-T was replaced with data from the Sainte-Anne-de-Bellevue station, which was selected for its proximity to P-E-T, many years of data (45 years) and current activity. Overall, our weather data were very complete and reliable.

Table 2

Replacement of missing data using a correction factor

     Missing data
 
Data from → to Years in common Distance between stations (km) Daily data Correction factor Before [days] After [days and (%)] 
P-E-T to McGill 1942–1986 13.76 Temperature +0.983 °C 30 4 (<0.01) 
Precipitation +0.193 mm 38 0 (0.0) 
Ste-Anne to P-E-T 1970–2014 15.03 Temperature −0.159 °C 60 57 (0.21) 
Precipitation −0.0945 mm 44 38 (0.14) 
     Missing data
 
Data from → to Years in common Distance between stations (km) Daily data Correction factor Before [days] After [days and (%)] 
P-E-T to McGill 1942–1986 13.76 Temperature +0.983 °C 30 4 (<0.01) 
Precipitation +0.193 mm 38 0 (0.0) 
Ste-Anne to P-E-T 1970–2014 15.03 Temperature −0.159 °C 60 57 (0.21) 
Precipitation −0.0945 mm 44 38 (0.14) 

Plot trendline

The first method to test for trends in temperature or precipitation over a long period of time consisted of plotting the average yearly and seasonal data in SigmaPlot (Systat Software, Inc.) and obtaining a linear equation of the parameter versus year. The strength of the relationship was determined using the coefficient of determination. As recommended in other studies (Gibbons et al. 2009; Ouranos 2015), the analysis was also done by seasons: (i) winter, from December to February, (ii) spring, from March to May, (iii) summer, from June to August and (iv) fall, from September to November.

Mann–Kendall test

The Mann–Kendall (MK) test is a statistical test widely used for the analysis of trends in climatologic time series. The advantage to using this test is that it is non-parametric and therefore does not require the data to be normally distributed. Moreover, it has low sensitivity to abrupt breaks due to non-homogeneous time series (Karmeshu 2012). The MK tests whether to reject the null hypothesis (H0) or accept the alternative hypothesis (Ha), where H0 represents the case where no monotonic trend exists (the data are independent and randomly ordered), and Ha represents the case where a significant (p ≤ 0.05, or p ≤ 0.01) monotonic trend exists (variables consistently increase or decrease through time depending on the situation) (Gilbert 1987; Pacific Northwest National Laboratory 2005). The test followed the methods of Gibbons et al. (2009) in obtaining the Kendall slope. This method was used to test for trends in temperature and precipitation on a seasonal or growing season basis. The seasonal Kendall test was used to test trends over the years as well. When data are influenced by seasonal effects, the previously described estimators can yield biased results, requiring a trend estimator adjustable to seasonal variation. The seasonal Kendall test is very similar to the regular MK test, except that the x values are the seasonal averages instead of the annual averages (Gibbons et al. 2009). The computation by this method was elaborated through a Matlab code (R2015b, The MathWorks, Inc., USA) enabling the calculation of large datasets very efficiently.

Two-sample Kolmogorov–Smirnov test

The two-sample Kolmogorov–Smirnov (KS) test is one of the most useful and general nonparametric methods for comparing two samples to determine whether they follow the same distribution (Wang et al. 2008). In this case, the method aims to test the difference between two subsamples before and after a change point, with an equivalent subsample size; in the case that the data set is odd, the extra value goes into the first group. The change point was taken to be the year in the middle of the data set, and this method was used in testing annual and seasonal data. A written-in function for the two-sample KS test in Matlab, known as the kstest2, was used to implement the KS test; p-values indicate whether the numbers from each group differ significantly; thus, a small p-value (p ≤ 0.05) rejects the null hypothesis.

Frequency analysis of rainfall data

Frequency analysis of historical rainfall data is a statistical method to estimate the return period, Tx, of a rainfall event. It is expressed as the number of years in which the annual observation is expected to return. The return period is the reciprocal value of the probability (Px) given by the following equation: . This analysis follows the method of Chow et al. (1988). The annual maximum daily rainfall event was computed for a 2-year return period (p = 0.50) and for a 1.05-year return period (p= 0.95). The number of daily precipitation events exceeding the specific return period rainfall event was plotted versus year.

Changes in growing season

The methodology adopted in this study follows closely that of Bootsma (1994) in analyzing growing season data. To begin with, the growing degree-days (GDD), a measure of heat accumulation used by farmers to predict crop development rates, was calculated on a daily basis by subtracting a given base temperature – in this case 5° – from the mean daily temperature. A base temperature of 5°C was chosen as being a suitable value for the most common Canadian crops, and was the same value used in Bootsma (1994) and Qian et al. (2012). If the mean daily temperature is lower than the Tbase, then GDD = 0 (Bootsma 1994). Thereafter, the total annual GDD can be obtained by the summation of the daily GDD values. The growing season length (GSL) starts at the date when the daily mean temperature stays above 5.5°C for 5 consecutive days (onset of growing season, OGS) and ends when the weighted temperature falls below 5.5°C (growing season end, GSE) (Qian et al. 2009). Therefore, the GSL corresponds to the number of days between the particular onset- and end-days.

RESULTS

Trend analysis by linear regression

The mean annual temperature, precipitation and snowfall for both stations, along with trendlines to indicate the tendency of the data over time, are presented in Figure 2. The slopes of the trendline regression equations (Table 3) show that, in general, temperature increased for both locations and in all seasons, and annual total precipitation increased, whereas total winter precipitation and snowfall depth decreased.
Table 3

Rate of change, over a 100-year period, of seasonal/annual temperature means and precipitation totals analyzed through linear regression and Kendall slope analysis, along with KS test of differences in trends over different portions of the historical records

  Slope
 
 Linear
 
Kendall
 
 
°C/century
 
Precipitation
mm/y/century
 

°C/century
 
Precipitation
mm/y/century
 
Season McGill P-E-T McGill P-E-T McGill P-E-T McGill P-E-T 
Winter prec. 2.25 2.36 −34.2 −117 2.12**†† 2.57* −28.8*† −15.3 
Winter snow – – −20.7 −18.9 – – −20.7** −18.9*†† 
Spring 2.44 1.25 +20.2 +62.6 2.54**†† 1.05 +16.6 +58.9 
Summer 1.36 0.67 +15.6 +51.5 1.08**†† 0.68 +26.7 +63.5 
Fall 2.04 0.08 +10.0 +47.3 1.76**†† 0.24 +10.0 +31.9 
Full year 2.00 1.13 +14.6 +149.6 1.88**†† 1.18** +23.9 +138.7* 
  Slope
 
 Linear
 
Kendall
 
 
°C/century
 
Precipitation
mm/y/century
 

°C/century
 
Precipitation
mm/y/century
 
Season McGill P-E-T McGill P-E-T McGill P-E-T McGill P-E-T 
Winter prec. 2.25 2.36 −34.2 −117 2.12**†† 2.57* −28.8*† −15.3 
Winter snow – – −20.7 −18.9 – – −20.7** −18.9*†† 
Spring 2.44 1.25 +20.2 +62.6 2.54**†† 1.05 +16.6 +58.9 
Summer 1.36 0.67 +15.6 +51.5 1.08**†† 0.68 +26.7 +63.5 
Fall 2.04 0.08 +10.0 +47.3 1.76**†† 0.24 +10.0 +31.9 
Full year 2.00 1.13 +14.6 +149.6 1.88**†† 1.18** +23.9 +138.7* 

*, ** — significant p 0.05, p 0.01, respectively, for MK test for significance of Kendall slope.

†, †† — significant p 0.05, p 0.01, respectively, for KS test of differences in trends over different portions of the temperature/precipitation records.

Figure 2

Mean annual temperature, total annual precipitation and snowfall at McGill and Dorval/P-E-T Airport climatological stations. Refer to Table 3 for the slopes.

Figure 2

Mean annual temperature, total annual precipitation and snowfall at McGill and Dorval/P-E-T Airport climatological stations. Refer to Table 3 for the slopes.

The slope of plots of temperature and precipitation data over time (Table 3) were calculated by linear regression and Kendall slope analysis, which were consistent with each other. The Kendall slope is more reliable, since it is known to give unbiased results and compares every value to all others. For both stations, the seasonal and annual temperature data showed a consistent increase over the years, while in the case of precipitation a general increase was seen in all seasons but winter.

Frequency analysis of rainfall data

Maximum daily precipitation was plotted by year for both sites (Figure 3). Following the frequency analysis, the 50% (2-year return period) daily maximum rainfall for McGill was found to be 48.67 mm, compared to 46.19 mm for P-E-T. The 95% daily maximum rainfall (1.05-year return period) corresponded to 31.3 mm for both stations. From these reference values, the total number of daily precipitations above the 50% and 95% daily maximum rainfalls in each decade were plotted (Figure 4). For both stations, the annual maximum daily precipitation increased slightly but not significantly over the years (Figure 3). The number of precipitation events above the 50% and 95% daily maximum rainfall events decreased slightly by decades for McGill, but increasing considerably for P-E-T. However, conducting the MK and KS tests on the number of precipitation events above the n-year return period data would be inappropriate, since the number of decades is insufficient. Therefore, the significance of these data can only be interpreted by using the linear regression analysis.
Figure 3

Annual maximum daily precipitation by year for McGill and P-E-T Airport sites.

Figure 3

Annual maximum daily precipitation by year for McGill and P-E-T Airport sites.

Figure 4

Total number of daily precipitation events above the 50% and 95% daily maximum rainfalls (2 year and 1.05 year return events) in each decade for McGill and P-E-T Airport climatological stations.

Figure 4

Total number of daily precipitation events above the 50% and 95% daily maximum rainfalls (2 year and 1.05 year return events) in each decade for McGill and P-E-T Airport climatological stations.

Changes in growing season

Increases of 315 and 157 GDD were observed over a 100-year period at the McGill and P-E-T sites, respectively Figure 5. Over the same period, the length of the growing season at McGill increased by 13.4 days, with the season's onset being 3.3 days earlier and its end 10 days later than 100 years ago, on average. Similarly at the P-E-T site, the growing season was extended by 6.5 days, with the season's onset occurring 8.8 days earlier and season's end occurring 2.1 days earlier.
Figure 5

Growing season parameters at both stations: GDD, OGS, GSE and GSL.

Figure 5

Growing season parameters at both stations: GDD, OGS, GSE and GSL.

Both MK and KS tests were performed on growing season parameters to test if any trends were significant. Only the GDD at the McGill site showed a highly significant trend (p ≤ 0.01), showing a change of 3.154 GDD/yr; for all other parameters p > 0.10. As for the KS test, it showed a significant increase between 1930–1986 and 1872–1929 at McGill for the end of season (p= 0.10), season length (p= 0.10) and GDD (p= 0.01). All the parameters at P-E-T showed no significant change.

DISCUSSION

Variation in the results of the McGill and P-E-T trend analysis are expected for the following reasons: spatial variability (a distance of 13.76 km between the stations), temporal variability (data from different time periods: McGill 1872–1986 and P-E-T 1942–2014) and microclimate difference (effects of being close to an urban center versus an airport). These differences can have significant effects on temperature and rainfall readings, as well as homogeneity in station data (Government of Canada 2015). Finally, it is important to be aware of the differences between the two stations and to analyze the data accordingly.

Temperature

The temperatures at McGill station are increasing significantly (p= 0.01 for MK and KS tests) for every season and annually, at a rate of 1.88°C/century. Gan (1995) showed an overall warming of 0.5°C/century for Canada and the northeast United States, while Berry (1991) suggested a global temperature increase of 0.4°C/century to 1.3°C/century. Thus, in comparison, our rate of temperature change was much higher than the observed global temperature change, indicating that the Montreal station (1872–1986) showed a more rapid warming trend than the global mean.

A change in temperature of 1.18°C/century was found at the P-E-T station, representing a significant increasing trend (p 0.01). The value of change over a 100-year period is an extrapolation, since the data only cover a range of 74 years. In comparison, data from Ouranos (2015) showed an increase of 1–3°C for all Québec regions over the period of 1950–2011, corresponding to a range of 1.6–4.8°C/century, assuming a linear increasing trend. The latter results can be compared to P-E-T's data (1942–2014) which cover about the same period of time. Indeed, the observed change in temperature does not fit in the range reported by Ouranos; thus, it seems that Ouranos overestimated the change for southern Québec and the island of Montreal.

The biggest change in temperature at P-E-T occurred in winter, a significantly (p 0.05, MK test) increasing trend of 2.57°C/century. A more detailed analysis detected a change in October temperatures, which were seen to be decreasing by 0.87°C/century (p 0.01 for KS test). Gan (1995) reported that southeastern Canada was entering a cooling trend, with a strong correlation in October and limited correlation in January and June. These results concur with the October cooling trend found in the present study, but not for January and June.

In the Government Strategy for Climate Change Adaptation report, the Government of Québec (2012) predicted an increase of 2.5–3.8°C in winter and 1.9–3.0°C in summer by 2050. It is interesting to compare these results to ours on a 38-year basis (2012–2050). An increase of 2.61°C in winter and 1.54° in summer was observed in the last 38 years at P-E-T, fitting in the Quebec winter range, slightly below the summer values, but still consistent overall.

Precipitation

Yearly cumulated daily precipitation at the McGill station increased by 23.9 mm/y/century or 2.3%; however, this trend was not significant under the MK and KS tests (p > 0.10). The winter precipitation and snow decreased significantly (p 0.05, p 0.01, respectively); more specifically, precipitation for January decreased significantly by 27.0 mm/y/century. Annual precipitation increased significantly (p 0.05) at P-E-T, with a change of 138.7 mm/y/century, i.e., an increase of 15%. Winter precipitation was also observed to have decreased, but the trend was not significant; on the other hand, the snow depth was decreasing significantly (p 0.05). There are few results from other studies concerning the change in precipitation in southern Québec. A study from the University of Ottawa supports our results, since it observed that precipitation increased significantly for all seasons except for winter, which decreased over the past 50 years (Alodah 2015). The present results seemingly contradict the Government of Québec's prediction (Government of Québec 2012) of an increase in precipitation of 8.6% to 18.1% for winter with no significant change in summer.

Changes in growing season parameters

It is evident that all growing season parameters are directly related to the temperature; thus, our first prediction was that with increasing temperatures, the GDD and the GSL are expected to be greater and longer, respectively. As discussed in Bootsma's (1994) study analyzing similar trends for Ottawa, these two variables showed little evidence of being affected by climatic change, although there had been gradual increases since the 1930s. This is supported by the results of the KS test, which showed an increase during 1930–1986 compared to 1872–1929 at McGill for season length (p 0.10) and GDD (p 0.10). Despite the increasing trend of temperature at P-E-T, supported by the results of MK and KS tests, none of the growing season parameters showed a significant change. This may be explained by the fact that the warming trend did not occur during the growing season, but mostly happened during winter, and that fall temperatures have been observed to be declining over the last 40 years. Moreover, the distributions for GDD and GSL showed a significant rising trend (p0.10) in Ottawa, which did not concur with any of our results except GDD at McGill (p 0.01).

The warming trend in spring is observed to be much larger than the warming trend in fall, as the onset of the growing season became earlier and the end of the growing season did not change as much. Indeed, warming only in spring results in an earlier start to the growing season, and thus an extended growing season, since most crops are usually mature before the end of summer. Qian et al. (2012) found the last spring frost to occur as early as April in southern Québec. Following the trend for the season's onset at P-E-T, the first day of the growing season is expected to occur in the second week of April for the 2010–2014 period.

CONCLUSIONS

The analyses of the 115 years and 74 years of historical weather data at the McGill and P-E-T stations, respectively, showed evidence of climate change. With some exceptions, temperature trends showed significant increases. On the other hand, the trends for precipitation were not as significant except the decreasing trend observed in winter precipitation and snowfall. Overall, the use of Kendall's test provided a good insight into the temporal distribution of temperature and precipitation trends in Montreal. Moreover, the KS test was useful to detect any change between two periods of time; indeed, an important increase in precipitation was observed for P-E-T.

Frequency analysis of the rainfall data was useful in determining precipitation for specific events. An increase in the number of 50% (2-year return period) and 95% (1.05-year return period) daily maximum precipitation was observed at the P-E-T station, whereas the annual maximum rainfall amount showed no evident increasing trend. This suggests that although the maximum amount of daily precipitation does not change, the number of heavy rain events (i.e., exceeding a 2-year return period daily maximum) is increasing. The GSL has been increasing at both stations, with the OGS being earlier than 100 years ago; thus, climate change has helped increase crop production by lengthening the growing season. Thus, this overall change in weather patterns may result in more frequent flood events (nutrient leaching) and heat waves (higher evapotranspiration rate) which will require appropriate measures, such as the use of additional nutrients, installation of irrigation and drainage systems and crop monitoring to determine plant health.

Having a good knowledge and understanding of how climate change has affected a region is very important for further studies. In past years, crop yields have considerably increased in Québec; however, it is still not clear how much climate change accounts for the improvement in agricultural yields, compared to the effects of implementing better cultural techniques. Thus, the results obtained can be useful for further studies involving system modelling that would use the measured historical and predicted future weather datasets as inputs in order to get the crop yield as the output. Furthermore, a comparison of the actual observed crop yield (influenced by climate change and farming technology progress) and the simulated crop yield (only influenced by climate change) could be made to analyze the impact of climate on agriculture.

In general, the temperature and precipitation analyses provided a better understanding of climate change and the growing season over the last 100 years and more in Montreal. The results were not totally in agreement with other studies, but they were consistent with Ouranos (2015), Government of Québec (2012) and Bootsma (1994); by contrast, this study is more specific to Montreal and detailed in its analysis of historical weather data.

ACKNOWLEDGEMENTS

This research was funded by the National Sciences and Engineering Council (NSERC) of Canada and the Faculty of Agricultural and Environmental Sciences, McGill University.

REFERENCES

REFERENCES
Allen
R. G.
Pereira
L. S.
Raes
D.
Smith
M.
1998
Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements
.
FAO Irrigation and Drainage Paper 56. FAO, Rome. http://www.fao.org/docrep/x0490e/x0490e00.htm
(accessed 1 April 2016).
Alodah
A.
2015
Development of Climate Change Scenarios for the South Nation Watershed
.
Masters of Applied Science Thesis
,
Dept. of Civil Engineering, Ottawa University
.
http://hdl.handle.net/10393/32103 (accessed 1 April 2016)
.
Berry
M. O.
1991
Recent temperature trends in Canada
.
Operational Geographer
9
,
9
13
.
Canadian Climate Data and Scenarios (CCDS)
2014
Climate Change Concepts: Canadian Climate Data and Scenarios
,
Government of Canada
. .
Chow
V. T.
Maidment
D. R.
Mays
L. W.
1988
Applied Hydrology
.
McGraw-Hill International Editions
,
Singapore
, pp.
390
398
.
Gibbons
R. D.
Bhaumik
D. K.
Aryal
S.
2009
Trend analysis
. In:
Statistical Methods for Groundwater Monitoring
.
2nd edn
.
Wiley
,
New York
, pp.
205
215
.
Gilbert
R. O.
1987
Detecting and estimating trends
. In:
Statistical Methods for Environmental Pollution Monitoring
.
Van Nostrand Reinhold
,
New York
, pp.
204
224
.
Government of Quebec
2012
Government Strategy for Climate Change Adaptation 2013–2020
.
Bibliothèque et Archives nationales du Québec
,
Quebec City, QC
. .
IPCC
2014
Climate change 2014: synthesis report
. In:
Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
(
Core Writing Team
,
Pachauri
R. K.
Meyer
L. A.
, eds).
IPCC
,
Geneva
,
Switzerland
.
http://www.ipcc.ch/report/ar5/syr/ (accessed 1 April 2016)
.
Karmeshu
N.
2012
Trend Detection in Annual Temperature & Precipitation Using the Mann Kendall Test – A Case Study to Assess Climate Change on Select States in the Northeastern United States
.
Masters of Environmental Studies Thesis
,
University of Pennsylvania
. .
McGill
2016
Macdonald Campus Farm
. .
Ouranos
2015
Vers l'adaptation. Synthèse des connaissances sur les changements climatiques au Québec
.
Édition 2015
.
Ouranos
,
Montréal, QC
. .
Pacific Northwest National Laboratory
2005
Mann-Kendall Test for Monotonic Trend
.
)
.
Qian
B.
Zhang
X.
Chen
K.
Feng
Y.
O'Brien
T.
2009
Observed long-term trends for agroclimatic conditions in Canada
.
J. Appl. Meteorol. Climatol.
49
,
604
618
.
Qian
B.
Gameda
S.
Zhang
X.
De Jong
R.
2012
Changing growing season observed in Canada
.
Climatic Change
112
,
339
353
.
Takle
E.
Hofstrand
D.
2008
Global Warming – Impact of Climate Change on Global Agriculture
.
Iowa State University
.
(accessed 1 April 2016)
.
Ville de Montréal
2015
Preserve and Enhance Rural Character and Agricultural Activities in Certain Areas of the West Island
. .
Wang
W.
Chen
X.
Shi
P.
van Gelder
P. H. A. J. M.
2008
Detecting changes in extreme precipitation and extreme streamflow in the Dongjiang River Basin in southern China
.
Hydrol. Earth Syst. Sci.
12
,
207
221
.