Back to Climate & Biota Table of Contents

The climate and its variability is a critical component for understanding sustainable development and maintaining biodiversity. With the advent of Geographic Information Systems (GIS) technology, the capacity to capture, spatially model and disseminate climate data digitally, is enhanced. The data reconstruction of the climate archive in conjunction with the integration of physiographic and digital elevation modelling was essential towards developing an objective climate interpolation analysis. An adiabatic lapse rate normalization and a Digital Elevation Model (DEM) model were applied to the primary temperature mapping. A linear and an exponentially weighted difference interpolation, collectively, comprise the objective algorithms applied over a variable re-sampling grid.

A collaborative effort by Environment Canada and the Ontario Ministry of Natural Resources (OMNR) was undertaken to develop an integrated objective analysis of bioclimate for the entire province of Ontario, at a regional scale of 1:5 Million. The result is an electronic atlas with which the user is encouraged to interactively display and query a combination of bioclimate maps, the associated data bases, and graphically compare climate station relationships.

1. INTRODUCTION

Sustainable development decisions rely on spatial data sources from land, sea and air for inter-disciplinary studies. These decisions include assessing the viability of existing and potential developments within a variable and changing climate. This requires a data management strategy that will promote the integration of climate data into multi-disciplinary applications.

The atmosphere is a dynamic phenomena, which requires description in the spatial context. The capacity to capture, spatially model and disseminate climate in a digital format has been greatly enhanced with the advent of GIS technology. The Canadian Climate Archive is a large, temporal point dataset, maintained in a tabular format. The spatial nature of climate is an important feature that is not adequately represented by this format. Therefore, by applying GIS techniques to the existing climate archive and by developing ar. integrated climate interpolation methodology, the net value of the archive is enhanced.

The Ontario Ministry of Natural Resources (OMNR) requires the use of regional climate maps for effective forest management. Improved silviculture, genetic sampling and replanting will lead to maximized biomass, hardier trees and a sustained biodiversity. The Bioclimate Division of Environment Canada has undertaken the development of the base mapping of bioclimate parameters (i.e. temperature, precipitation, degree-days, probability of frost etc.) at an improved regional scale. This collaborative effort by Environment Canada and the OMNR will yield a hard copy and an electronic bioclimate atlas for the entire province of Ontario.

The Bioclimate maps are intended to complement the former Bioclimate Profiles by illustrating the respective spatial variability across Ontario.

1.1 Objectives

• delineate the Bioclimate parameters (Primary and Derived)
• generate a complete daily climate record, for input to the primary and derived bioclimate parameters
• develop an objective climate mapping model
• generate climate surfaces of the entire province at a regional scale (approx. 1 : 5,000,000)
• provide digital climate maps for interactive display, query and dissemination on a desktop mapping platform (SPAN MAP)
• provide a hardcopy bioclimate atlas of Ontario

 2. BIOCLIMATE ELEMENTS

The bioclimate elements used in the mapping are averages for the 1968-91 period. There are two primary and twelve derived bioclimate elements described for climate mapping. Each is defined as follows:
 

2.1 Primary Climate Parameters

Primary climate parameters are directly observed and measured by instrumentation, on a daily basis.

2.11 Temperature

• Minimum Temperature 0C
• Maximum Temperature 0 C


Temperatures are measured in a louvred box called a Stevenson screen, mounted 1.5 m above the ground, ideally on a level grassy surface. The maximum temperature is the highest value recorded in a 24-hour period ending on the morning of the next day. The minimum values are also for a 24-hour period beginning on the evening of the previous day. The maps represent the maximum and minimum monthly temperature in degrees Celsius for the period.

2.12 Precipitation

• Total Precipitation (mm)


Precipitation comprises several types including rain, drizzle, freezing rain, freezing drizzle, hail and snow. The Canadian rain gauge (Type B), a cylindrical container 40 cm high and 11.3 cm in diameter, is the standard instrument for measuring rainfall. The precipitation is funneled into a plastic container, which has graduated markings. Precipitation amounts represent an accumulation of the water equivalents of all types of precipitation and are averaged for a given month of the year. At most climate stations the water equivalent of snowfall is computed by dividing the measured amount by ten. At principal stations, precipitation may be measured from a standard rain gauge and/or a Nipher snow gauge and/or a tipping bucket rain gauge. The maps represent the mean total monthly precipitation in millimetres for the period.

2.2 Derived Climate Parameters

Derived climate parameters are computed from the primary climate parameters and reconstructed daily elements.

2.21 Mean Temperature ^oC

Daily mean temperature is not directly observed, but calculated as the average of the minimum and maximum temperatures for the day. The maps represent the mean monthly temperature, in degrees Celsius, for the period.

2.22 Probability of Frost ( %)

Probability of Frost is the percentage (%) of number of days during the growing season when the daily minimum temperature is <0.0 ^oC. Note: That the temperatures are measured at the standard height of 1.5 m above ground.

2.23 Total Degree-Days Above 0,5 & 10 ^oC

Degree-days for a given day, represent the number of Celsius degrees that the mean temperature is above a given base temperature (e.g. 0, 5, 10 0C). Values above 50C are often called growing degree-days, and are used in agriculture as an index of crop growth. Totals, represent the average accumulation of degree-days for a given month or year.

2.24 Annual Total Heating Degree-Days Below 18 ^oC

Heating degree-days are the number of Celsius degrees that the daily mean temperature is below the base of 18 ?C. These values are used to estimate the heating requirements at a location. The annual total represents the average accumulation of degree-days for the period.

2.25 Annual Total Cooling Degree-Days Above 18 ^oC

Cooling degree-days are the number of Celsius degrees that the daily mean temperature is above the base of 1 80C. Values above the base of 1 8?C are used to estimate the air-conditioning requirements at the location. The annual total represents the average accumulation of degree-days for the period.

2.26 Annual Total Corn Heat Units (CHU)

Corn Heat Units are in degrees Celsius. This accumulated value is calculated using the following conditions.

• Start-up of corn growing season:  where the mean daily temperature is> 12.8 ^oC for three consecutive days during the period May 11 to July 31.
• End of the season:  the first occurrence when the minimum daily temperature drops below -2.0 ^oC during the period August 1 to October 15.

 
2.27 Total Water Deficit Above a WHC of 100 mm

Water deficit is derived from the Thornthwaite water balance calculation (Thornthwaite and Mather, 1955; Johnstone and Louie, 1983). This model uses an empirical method to compute the changes in water storage as a function of monthly mean temperature, total precipitation, latitude (duration of sunlight) and soil texture (water holding capacity). A sandy loam soil was assumed for all water holding capacities.

2.28 Growing Season, Mean Start, End & Length (Julian days)

The growing season length is the annual mean number of days between the last frost (0^oC) in the spring and the first frost in the autumn. The growing season start and end are defined as the annual mean start/end date, in Julian days, of the last occurrence of 0 ^oC in the spring and the first frost in the autumn.


3. CLIMATE ARCHIVE DATA RECONSTRUCTION OF PRIMARY ELEMENTS

The data values utilized in the project are primarily derived from data in the national climate archive of Environment Canada. While every effort is made to ensure the accuracy of these data, no guarantee can be given that they are completely error free (Canadian Climate Normals 1960-90).
 

3.1 Period of Record of Climate Mapping

The original intent of the Bioclimate of Ontario project (phase II) is to support the Bioclimate Profiles of Ontario project (phase I) which covers the period 1968-88 inclusive. This constitutes the primary data source for the mapping project. This raises the question of deviating from a 30-year normal period. This question is addressed by the American Association of State Climatologists (AASC) in a published statement entitled "Climatic Means and Normals" directed primarily at the National Climatic Data Centre (NCDC). They recommend that the "inclusion of additional averaging periods other than 30 years in published statistics would provide flexibility in responding to climatic needs". Further to this statement there is no a-priori reason for a 30 year mean to be optimal for all applications (Kunkel and Court, 1990). Climate fluctuations occur over months, years and centuries and thus, 30 year means have no inherent stability.

The 30 year statistical period is adopted primarily to monitor climate change. In the prediction of next year?s seasons, the classic "normal climate" has been found to have less predictive accuracy than other period means (Lamb and Changnon, 1981). Beaumont (1957) and Enger (1959) found that a 15-25 year average produces the smallest root mean square errors. Sabin and Shulman (1985) suggest up to 40 years as an optimal period and yet found little difference for periods as short as 10 years. According to Landsberg (1951), 15 years for temperature and 30-40 years for precipitation is required to reach a stable frequency distribution for annual and monthly climate values at the regional scale such as that used in Ontario. The above studies found optimum results for averaging periods between 11 and 20 years. In summation, the addition of averaging periods other than the 30 year normals is not only valid but will improve the capture of spatial variability within the maps.

3.2 Data Reconstruction

A climate network occasionally has missing data and a history of gaps in the observational record for a variety reasons. Whether down time are due to equipment failure, relocation, observer issues or budget constraints, all long term volunteer programs experience this reality. Ideally, maps should be based on data from homogeneous macroclimatic stations, but this is not realistically possible. Reconstruction of data becomes necessary when the network density is inadequate and must be supplemented by valid short duration records across time and space (McKay and Thomas, 1969). Short term climate periods must be referenced as an aid in filling the large data sparse areas of the climatological network.

Therefore, a goal was set to maximize the use of climate observational data (quality checked) regardless of the completeness of the individual station record within the period. A subset of all stations within the 1968-91 period with a minimum of 11 years of daily consecutive observations was identified. The 1961-90 Climate Normals apply a rigid "Three-five rule" in which an entire month would be discounted if more than five intermediate or three consecutive temperature observations were missed. With respect to precipitation, an entire month would be excluded if one or more days were not observed.

To avoid this loss, a data reconstruction model was developed to enhance the Climate Normals procedures. Missing daily temperature or precipitation observations were estimated from all available (Canadian Climate Archive) concurrent station data. Temperature data were corrected for elevation effects inherent at station locations through a lapse rate normalization procedure. All neighbouring climate station data concurrently observed on a given day within a two degree radius of the missing observation were used in the weighted reconstruction of the climate element. An inverse squared distance weighting algorithm was utilized to account for the irregular spacing. This preprocessing was repeated on a daily basis for each climate station, over the entire 1968-91 period.

The reconstructed daily climate archive, as described, provided the complete dataset for the calculation of summary statistics of monthly means, totals and probabilities.

Table 3-1. Climate Stations and Network Reconstruction Totals for the 1968-91 Period.

 
The derived climate dataset included 300 stations from Ontario. In the west, 74 stations from Manitoba and 134 stations from Quebec were included, to optimize data analysis across provincial boundaries. The dominant influence of the Great Lakes along the southern boundary of Ontario, eliminated the need for climate data in these areas. The inclusion of these stations would have negatively biased results along the adjacent Ontario shoreline. Along land borders with Minnesota and New York State, 7 USA climate stations were added to the dataset. A total of 515 derived climate station summaries were compiled and input into the quality control phase of the mapping project. Refer to Appendix B for a listing of the climate stations used in the final modelling and mapping.

To improve on the spatial network for mapping, selected stations from the 1951-80 period were utilized. A difference between the means of the neighbouring stations over the two periods was used as the adjustment procedure. Adjusted monthly means were derived for the 1968-91 period in those data sparse geographic areas.

3.3 Quality Control and Error Analysis

The quality of the climate data used, directly influences the quality of the final map. Environment Canada uses several standard quality control checks to identified data outside the realm of possibility for normal and extreme field observational data. Random errors due to instruments, observations and sample bias are less significant with respect to mean statistics but are certainly relevant for event or extreme occurrence identification (McKay and Thomas, 1971). Non-random errors within the realm of possibility, i.e., standard deviations, are very difficult to identify. Spatial variability changes along with the interaction of temperature, precipitation, physiography, season, time period/duration and the scale of the meteorological phenomena. Climate is often assumed to be constant, yet in reality it is both changing and variable. This assumption has more basis in the relative short time span of humans than it does in a geologic time frame. Errors introduced by combining data from different time periods (that meet the 11 yr. minimum length of record criteria) are negligible in comparison to normal climate variability. Spatial scale can be an issue when local micro climates become dominate due to poorly located stations and misrepresent the regional macro climate. Therefore, it is important to identify such stations and situations.

GIS technology enables one to illustrate spatial anomalies within the regional climate record, which would otherwise go undetected in the alphanumeric tabular format of the climate archive. The method of interrogation involves applying a linear interpolation to the annual and monthly climate dataset. Several advantages are gained by applying this linear quality control test. First, an exact interpolator forces the climate surface to pass through all data points. Second, spatial artifacts generated by the linear Triangulated Irregular Network (TIN) algorithm are eliminated. Outlier climate anomalies are easily identified for further scrutiny. At this stage, a subjective climatological interpretation is made of the test climate surfaces to identify spatial anomalies. Suspect stations have their historical inspection files investigated for evidence supporting erroneous data observations/recordings. Poor instrument exposure, dominant local effects, observer procedures and station relocation are all checked for possible errors. Exceptions are made for urban heat island effects.

Stations which produce anomalies and in turn vary from their surrounding climate stations greater than +/-1^oC for temperature and/or 20 % for precipitation, from the mean local climate surface, are removed from the mapping phase. Stations with less than 15 years of record, which demonstrated this level of variability, are also eliminated.

4. DEVELOPMENT OF A CLIMATE SENSITIVE MAPPING MODEL

Spatial variability changes according to the element, season, topography and the duration of observation. Station data are by their nature biased and not always representative of the macroclimate of an area. Variability can be empirically determined to predict areal change when the climate is controlled by the physiography, i.e., orography, surface relief, exposure and marine influences (McKay and Thomas, 1969).

The national climate archive can be described as an irregularly spaced network of observations biased towards human habitation. An important consideration towards spatially interpolating an acceptable regional climate map is to capture the character of the surface in the x, y, element point data. This reason alone calls for the maximizing of the climate network in the reconstructed bioclimate dataset.
 

4.1 Run First Order Linear TIN Surface

GIS technology uses a Triangular Irregular Network (TIN) method as one of several methods to describe a surface geometry (climate surface) via planar triangulation connected at the vertices of the triangles (climate station sites). A continuous, raw, unclassified, climate surface is generated for each bioclimate parameter, on a monthly basis. An annual surface is also generated. A linear TIN is used to project the primary climate surface while providing input data for the subsequent climate zone re-sampling.

4.11 Adiabatic Lapse Rate Normalization of Temperature

Each climate station has the inherent imprint of altitude imparted on the value of the temperature observations. It is therefore necessary to normalize the adiabatic impact of elevation on the existing climate network, before correct interpolation of the continuous climate surface can be modeled. This is achieved by reducing all the existing station temperature data down to sea level.

Table 4-1: Adiabatic Lapse Rates

(From: Pal Arya, 1988)

A range of adiabatic lapse rates (Table 4-1) are chosen as corrective coefficients for altitude changes below cloud levels within the boundary layer. The climate site elevations are used for the normalization adjustment to sea level.

4.2 Grid Re-sampling

The number and quality of climate data imposed limitations on the map scale, placing of isolines and the significance of derived estimates. The data sampling utilized the following spatial mapping features to describe the geographic variability of the climate of Ontario.

A buffer zone map was generated from the climate station sites at radii of 25, 60, 125, 250 and 500 kilometres. The buffer map enabled a more equitable surface sampling in all directions. This diminished the bias of the irregularly spaced station locations.

Table 4-2: Climate Zones

A climate zone map was compiled from reports on the Climate of Southern Ontario by Brown et al. (1968) and the Climate of Northern Ontario by Chapman and Thomas (1968). These regions were integrated with local and regional climate phenomena described in the climate zone and marine corridor sampling tables zl-2 and 4-3. The local climate was captured in near shore lake effect areas, local topography, inversions, cool-air drainage, transitional zones (steep gradients) and heat island features in urban areas. Regional climate was modeled by describing transitional inland lake effect of the Great Lakes, combined with upslope and rugged relief areas. The Algonquin Highlands, Dundalk Uplands, OakRidge Moraine, Iguace, Wawa and the Niagara Escarpment were identified. Continental scale climate influences via Maritime, Continental and Polar air masses were amply described in the broad climate network.

Table 4-3: Marine Corridor Grid Re-sampling

The addition of re-sampled points in describing climate zones, models the use of breaklines where vertices and triangle edges can better describe the continuous surface. In this way, a variable grid density was selected, based on the climate sites, in order to maintain the original sampling integrity and allow for the spatial re-sampling of local and regional climate zones within the province.

Table 4-4: Grid Re-sampling

 
5. OBJECTIVE SPATIAL ANALYSIS

A GIS climate interpolation analysis was developed and applied systematically across the province of Ontario. This provided a common objective platform from which the spatial interpolation of the regional climate could be assessed with an assured level of consistency.

5.1 Inverse Squared Weighted Difference Climate Mapping

The input data for the interpolation algorithm were re-sampled from the normalized and adiabatically adjusted climate surface at a defined variable grid spacing. The distribution and climate characteristics of these data points were critical to the success of the spatial interpolator. The re-sampled grid was the sole source for a weighted distance interpolation algorithm. Via an empirical process, the combination of a minimum 50 kilometre and maximum250 kilometre sphere of influence, in conjunction with an Inverse-squared, weighted distance algorithm, limited to 5 neighbouring input points, was selected. Through an iterative process, the selected parameters formed the best model of consistent objective analysis across the province. The primary temperature parameters were re-adjusted to the local orography by adding back the appropriate adiabatic lapse rate correction across the entire province. A regional course- gridded five-minute Digital Elevation Model (DEM) was used. The source of the DEM was the U.S. Geophysical Centre in Boulder, Colorado and the Great Lakes Atlas. The modal height was chosen as the most representative correction factor for regional climate purposes. This stage of the temperature mapping significantly improved spatial resolution of the climate surface beyond previous efforts by readjusting the climate surface to the local and regional elevation influence.

5.11 Marine Corridor Adjustment

Within the marine corridor zones, the temperature model was adjusted perpendicular to the shore for those months of the year in which representative near shore stations showed significantly warmer marine moderation in contrast to inland stations (Table 5-1). Water bodies cause a near parallel influence on isotherms along shoreline, which significantly changes during seasons of open water verses ice cover. The moderation can be represented by a simple arithmetic, linear gradient.

a = gd, where

a = net adjustment (^oC)
g= near shore station temperature - inland station temperature (^oC) perpendicular distance from near shore station to inland station (km)
d = distance of grid cell from marine shore (km)

Table 5-1: Temperature Moderated in Marine Corridor

(m-moderate, - negligible)

5.2 Exponential Weighted Difference Precipitation Mapping

Precipitation does not easily lend itself to simple linear or parametric based modelling. A direct physical model, based on elevation, can not be applied to precipitation, as is the case for temperature. The non-parametric character of precipitation resulted in a procedure chosen based on observations and spatial re-sampling Therefore, a model incorporating some scale re-sampling aspects of the physical approach, while maintaining the objective analysis of the linear and weighted difference numerical approaches, was applied. Generally, increases in precipitation with elevation approximate a linear form and in other conditions can be described as log-linear or exponential, according to Daly et Al. Precipitation maximums in mid latitudes usually occur at or near the crest of topological barriers (Hanson et al. 1980). In large scale situations, displacement upwind of the crest may occur over very broad barriers as a result of lifting in the upslope flow (Smith 1979). An optimal scale for orographic effects has not been definitively answered, yet evidence suggests that 2 - 15 km broad scale topographic features demonstrated a higher correlation with precipitation than actual point station elevations (Hilbbert 1977). Evaluation of a 5 ?lat./ long. (approx. 5 ?9 km) DEM (National Geophysical Data Centre 1989) grid suggested that the elevation of the DEM can better approximate a station?s regional influence (Daly et al. 1994). Local variance was slightly smoothed by increasing the maximum number of neighbours to 6 while accounting for precipitation?s non-linear behaviour at scales less than the sample density with an exponential decay rate of 0.1.

6. SMOOTH FILTERING

The intermediate climate surfaces were produced at a grid scale of 1 kilometre. A post- processing, smoothing algorithm was applied to reduce random noise errors, dampen undulation in the continuous surface and better represent the regional climate. The common smoothing algorithm entailed moving a modal, statistical matrix (11 x 11 km) across each climate map.

Appendix A and B

Assel, R.A., Quinn, F.H., Leshkevich, G.A., Bolsenga, SF., 1983. Great Lakes Ice Atlas. United States Department of Commerce, NOAA, Environment research laboratories. Great Lake Environmental Research Laboratory, Ann Arbor, Michigan, l-1 15.

Basist, A., Bell, G.D., Meentemeyer, V., 1994. Statistical relationship between topography and precipitation patterns. Journal of Climate, 7: 1305- 13 15.

Beaumont, R.T., 1957. A criterion for selection of length of record for a moving arithmetic mean for hydrologic data. Trans. American Geophysl. Union, 38: 198-200

Brown, D.M., McKay, G.A., Chapman, L.J., 1968. The climate of southern Ontario. Climatological studies, no. 5. Meteorological Branch, Department of Transport.

Bruce, J.P., Clark, R.H., 1966. Introduction to hydrometeorology. Oxford, Pergamon Press Ltd.

Burrough, P.A., 1986. Principles of geographical information systems for land resources assessment. Monographs on soil and resources survey, no. 12. New York, Oxford University Press.

Canadian Climate Normals, 195 l-l 980 : Temperature and precipitation. Canadian Climate Program, Atmospheric Environment Service, Environment Canada.

Canadian Climate Normals, 196 1- 1990,1993. Canadian Climate Program, Atmospheric Environment Service, Environment Canada.

Chapman, L.J., Thomas, M.K., 1968. The climate of Northern Ontario. Climatological Studies, no. 6. Meteorological Branch, Department of Transport.

Climatic Atlas Canada : A series of maps portraying Canada?s climate. Map series 1 - Temperature and degree days, 1984. Canadian Climate Program, Atmospheric Environment Service, Environment Canada.

Climatic Atlas Canada : A series of maps portraying Canada?s climate. Map series 2- Precipitation, 1986. Canadian Climate Program, Atmospheric Environment Service, Environment Canada.

Daly, C., Neilson, R.P., Phillips, D.L., 1993. A statistical-topographic model for mapping climatological precipitation over mountainous terrain. Journal of Applied Meteorology, 33: 140-l 58.

Enger, I., 19.59. Optimum length of record for climatological estimates of temperature. Journal Geophys. Research, 64: 779-787.

Greenland, D., 1977. Aspects of the mesoclimatology of the Toronto area. Chmatological Studies, no. 29. Atmospheric Environment, and Fisheries and Environment Canada.

Hanson, CL., Morris, R.L., Engleman, R.L., Coon, D.L., Johnson, C. W., 1980. Spatial and seasonal precipitation distribution in southwest Idaho. Agricultural Reviews and Manuals, ARM-W- 13, USDA Dept. of Agriculture, Science and Education Administration.

Hevesi, J.A., Istok, J.D., Flint, A.L., 1992. Precipitation estimation in mountainous terrain using multivariate gedstatistics. Part I: Structural Analysis. Journal of Applied Meteorology, 3 1: 66 1-676.

Hevesi, J.A., Flint, A.L., Istok, J.D., 1992. Precipitation estimation in mountainous terrain using multivariate geostatistics. Part II: Isohyetal Maps. Journal of Applied Meteorology, 3 1: 677-688.

Hibbert, A.R., 1977. Distribution of precipitation on rugged terrain in central Arizona. Hydrological Water Resources Arizona Southwest, 7: 163- 173.

Johnstone, K., Louie, P.Y.T., 1983. Water balance tabulations for Canadian climate stations. Canadian Climate Centre, Environment Canada., 8-83.

Kunkel, K.E., Court, A., 1990. Climatic means and normals- A statement of the American Association of State Climatologists (AASC). Bulletin American Meteorological Society, 7 l(2): 201-204.

Landsberg, H., 19.5 1. "Applied Climatology". Compendium of Meteorology, American Meteorology Society, 976-992. ,

Lamb, P.L., Changnon, S.A., 198 1. On the "best" temperature and precipitation normals: the Illinois situation. Journal of Applied Meteorology, 20: 13 83- 13 90.

MacIver, D.C., Isaac, J.L., 1989. Bioclimate profiles for Canada - 195 l-1 980. Canadian Climate Centre, Atmospheric Environment Service, Environment Canada.

MacIver, D.C., Master-ton, J., 1984. Clirnatograms of Ontario. In MacIver, D.C., Street, R.B., Auclair, A.N. (Eds.) Proceedings of climate applications in forest renewal and forest production. Forestry Canada, 282-286.

MacIver, D.C., Whitewood, R.D., 1991. Forest Climates of Ontario: Part I- Bioclimates of Ontario. Ontario Ministry of Natural Resources, Sault Ste. Marie, Ontario.

McKay, G.A., Thomas, M.K., 1969. Mapping of climatological elements. In Hydrological Mapping. Proceedings of Work Shop Seminar. Canadian National Committee, International Hydrological Decade, Ottawa, p. 73-82.

McKay, G.A., Thomas, M.K., 197 1. Mapping of climatological elements. The Canadian Cartographer, 8: 1:27-40.

Oke, T.R., 1987. Boundary layer climate. 2nd Edition, London and New York, Metheun Press.

Pal Arya, S., 1988. Introduction to micrometeorology. New York: Academic Press, Inc., 5 l-56.

Phillips, D.L., Dolph J., Marks, D., 1992. A comparison of geostatistical procedures for spatial analysis of precipitation in mountainous terrain. Agricultural and Forest Meteorology, 58: 119- 14 1.

Sabin, T.E., Shuhnan, M.D., 1985. A statistical evaluation of the efficiency of the climatic normal as a predictor. Journal of Climatology, 5: 63-77.

Savleslaja, A., 1986. Great Lakes climatological atlas, Atmospheric Environment Service, Environment Canada.

Smith, R.B., 1979. The influence of mountains on the atmosphere. Vol. 2 1, Advances in Geophysics, Academic Press, 87-230.

Tabios, G.Q., Salas, J.D., 1985. A comparative analysis of techniques for spatial interpolation of precipitation. Water Resources Bulletin, 2 l(3): 365-380.

The Great Lakes : An environment atlas and resource book. 1987. Jointly produced by Environment Canada, United States Environmental Protection Agency, Brock University, Northwestern University.

Thomthwaite, C.W., Mather, J.R., 1955. The water balance. Drexel Institute of Technology. Publications in Climatology, 8( 1).

Whitewood, R.D., MacIver, D.C., 1992. Forest Climates of Ontario: Part 2 - GE-Generated Climate Atlas of Ontario, Ontario Ministry of Natural Resources, Sault Ste. Marie, Ontario.

about imap | map collection | integrated mapping assessments | imap papers | contacts