Wind-Rain Relationships in Southwestern British Columbia

Alex McGowan, P.Eng. Robert G. Humphries, Ph.D.

ABSTRACT The ongoing focus on building envelope failures in western Canada, especially southwestern British Columbia, have brought to light the strong influence that wind-driven rain can have on the building envelope. Although designers are given some guidance on the concept of designing appropriate envelope assemblies for a given exposure, there are no specific design tools to determine the actual exposure to wind-driven rain for a given location and orientation in this region. The closest design data are the DRWP (driven-rain wind pressure) values in the CSA A440 standard, but these are non-directional. A study (CMHC, 2004) was conducted to correlate wind direction with rain events for southwestern British Columbia, including Vancouver Island and the Lower Mainland. The results of the study provide climatic load data that could be used to modify existing climatic data in building codes, including the revised methodology using moisture indices, which are now incorporated in the National Building Code of Canada. The study suggested that it may be appropriate to consider a directional DRWP guideline to supplement existing information in (for example) the CSA A440 Standard and related design guides. This paper discusses the methodology used to select appropriate weather stations for inclusion in the data set. Criteria for completeness and reliability of the data are discussed, with the intent of providing guidance to others wishing to investigate the seasonal variation of the wind/rain relationship for other locations. Specific topics addressed include assessing the quality of the data, data gaps and length of record, with different criteria used for different record types (e.g., wind speed, wind direction, rainfall intensity). Then, typical examples are used to show how specific stations were selected for the study, and how the data were analysed to show the relation of wind direction, wind speed and rainfall intensity. In the examples, the seasonal variability of wind-driven rain is also investigated (again, this is examined in more depth in the report upon which this paper is based, but this paper provides a reasonable example of that variability).

The possibility of extracting useful rainfall data from existing radar weather data is also discussed.

INTRODUCTION The ongoing focus on building envelope failures in southwestern British Columbia has brought to light the strong influence of wind-driven rain on the building envelope. There are, however, no specific design tools to determine exposure to wind-driven rain for a given location and orientation in this region. The closest design data are DRWP (driven-rain wind pressure) values in the CSA A440 standard, but these are non-directional. A study (CMHC, 2004) was conducted to examine the relationship between wind and rain for all seasons of the year. Wind and rainfall data were used for a number of stations located on Vancouver Island and within the Lower Mainland of British Columbia. Some previous studies of wind-driven rain in Canada (e.g., Welsh et al., 1989; Surry et al., 1995) have ignored the winter months, as most precipitation in the Canadian winter is snow. Also, those studies tended to use Victoria to represent British Columbia: as the recent CMHC study shows (CMHC, 2004), these assumptions are erroneous.

Other statistical data analyzed in the CMHC report include rainfall frequency, seasonal distribution of wet hours, correlation between rainfall and windspeed, correlation between rainfall and wind direction, and local variability in driving-rain wind pressures (DRWP). This paper discusses the methodology used to select appropriate weather stations for inclusion in the data set. Criteria for completeness and reliability of the data are discussed, with the intent of providing guidance to others wishing to investigate the seasonal variation of the wind/rain relationship for other locations. Specific topics addressed include assessing the quality of the data, data gaps and length of record, with different criteria used for different record types (e.g., wind speed, wind direction, rainfall intensity). METHODOLOGY Twelve meteorological stations were determined to be suitable for the analysis: four on Vancouver Island (Figure 1) and the remaining eight in the Lower Mainland (Figure 2). These 12 stations (five Environment Canada Stations and seven stations in the Greater Vancouver Regional District) provide a statistically valid sample of:

Hourly wind speeds and directions at 10 metres above ground. At Environment Canada (EC) stations, two-minute mean wind speed and direction are collected at the top of each hour. Wind directions are stored in tens of degrees. One-hour mean wind speed and direction are computed at Greater Vancouver Regional District (GVRD) stations. There may be differences between top-of-the-hour data and hourly averages, but the study on which this paper is based indicates that trends can still be observed, even accounting for that difference.

Hourly rainfall rates in millimetres per hour. Using this hourly weather data, statistics such as averages, standard deviations, maximum and minimum values were calculated and various subsets of the data were plotted. Wind pressures were calculated using the classic formula P = 0.04991 x (wind speed)2

for P in Pascal (Pa) and wind speed in kilometres per hour. This equation assumes a 0° C air temperature with an atmospheric pressure of 101.325 kPa. The driving-rain wind pressure (DRWP) is calculated for the 5- and 10-year return periods using the wind pressure, P, for hours when the rainfall exceeds a threshold of 1.8 mm/hr. This is contrasted with the calculation of DRWP values in the CSA A440 national standard for windows (CSA, 1998), which were based on the 5-year and 10-year return periods for use with window systems identified in the CSA A440 standard. Because the data set used in this study covered less than 10 years, the 5-year and 10-year extremes could not be generated. Instead, the wind pressure was calculated for all hours when the rainfall threshold of 1.8 mm/hr was met. DATA ANALYSIS Table 1 shows the average annual rainfall amount at each of the 12 stations, as well as a breakdown of rainfall amounts by season. As shown in the table, Victoria International Airport measured the least amount of annual rainfall at 861 mm, almost one metre less than the highest observed annual rainfall amount (1,858 mm), recorded on Burnaby Mountain. In general, a consistent gradient in annual rainfall is observed, with values increasing as one moves from south to north (either on Vancouver Island or on the mainland), and from west to east for stations in the Fraser River valley. In the Lower Mainland, the stations on Burnaby Mountain and in Port Moody experienced high annual rainfall amounts likely due to their proximity to elevated terrain features.

Table 1 Average Rainfall Amounts by Season at 12 Stations

Station (listed from Average Seasonal Rainfall Amounts (mm) Average Annual east to west) Spring Summer Fall Winter Rainfall Amount (mm) Port Hardy Airport 290 212 691 614 1,807 Comox Airport 204 125 340 421 1,089 Nanaimo Airport 205 107 304 421 1,037 Victoria Int’l Airport 159 88 275 340 861 Vancouver Int’l Airport 242 146 349 412 1,150 Kitsilano 277 140 366 510 1,293 Burnaby Mountain 420 241 511 685 1,858 North Delta 271 164 407 499 1,342 Port Moody 397 182 522 667 1,767 Surrey East 284 146 341 494 1,265 Maple Ridge 337 186 468 476 1,466 Langley Central 269 119 347 585 1,320 All stations observed the highest rainfall amounts in the winter months (December – February), except Port Hardy Airport where the highest rainfall amount occurred in the fall months (September – November). Fall and winter months combined accounted for between 64% and 72% of the annual rainfall amount at the various locations. The study of rainfall and wind direction provides some interesting results, provided in the form of wind roses for the 12 stations studied. Figure 3 shows an example of these results, in this case for Nanaimo Airport. The wind rose at the top in Figure 3 shows direction and intensity of wind at the airport, expressed as a frequency in terms of the percentage of ALL hours for which data are recorded. The lower wind rose in Figure 3 shows the same results, except that the frequency is a percentage of WET hours (i.e., more than 1.8mm of rainfall). Two differences are immediately evident in Figure 3. First, the wind direction changes when it rains. The wind direction for ALL hours is approximately evenly distributed from the four cardinal directions, with some wind recorded from the southeast. During WET hours, the wind is predominantly from the east-southeast, southeast, and south-southeast. Very little wind-driven rain from the north or southwest would be anticipated in Nanaimo, based on these figures. This suggests that the southeast face of a building would be more vulnerable to driving rain conditions in Nanaimo. This is not to say that one should design the wall assembly on this façade differently from the other elevations (for example, rainscreen walls facing southeast and face-sealed stucco everywhere else), but it does provide guidance as to locating vulnerable details. In this case, low-threshold entrance doors for accessibility should not face southeast in Nanaimo, as it will be difficult to prevent water ingress due to prevailing wind-driven rain from this direction. Second, the magnitude of the windspeed shows significant variation when considering wet hours, as opposed to all hours. The wind is calm for almost 52% of all hours, but only for approximately 25% of the wet hours, and the wind speed is more often in the range of 6 – 9 m/s (22 – 32 km/h) during wet weather, as opposed to 4 – 6 m/s (14 – 22 km/h) for all hours. The CMHC report on which this paper is based includes similar wind roses for all twelve locations, for wet and total hours (as shown in Figure 3), and on a seasonal basis. In all cases, the wet-hour wind roses for each season shows that the wind direction varies least during the winter and most during the summer. In addition, the frequency of high wind speeds (greater than 12 m/s) is greatest in the winter. The GVRD shows a great variation in the all-hours and wet-hour wind roses as a function of location, and the Vancouver Airport is not representative of all locations within the GVRD. Thus, design decisions based on information recorded at the Vancouver Airport would be quite incorrect for many other locations.

15%

15%

10%

10%

5%

5%

N

NE

E

SE

S

SW

W

NW

Calm (<=0.5 m/s) = 52.9%

15%

15%

10%

10%

5%

5%

N

NE

E

SE

S

SW

W

NW

24.7%Calm (<=0.5 m/s) =

Figure 3: Nanaimo wind rose for all hours (top) and wet hours (bottom)

> 12 m/s 9 - 12 m/s 6 - 9 m/s 4 - 6 m/s 2 - 4 m/s 0.5 - 2 m/s

11th Canadian Conference on Building Science and Technology Banff, Alberta, 2007

Seasonal wind roses for wet hours only for each station are provided in Appendix A of the report upon which this report is based. In general, the wind direction during wet hours is relatively constant during winter and fall, and quite variable during summer and spring. Thus, the highest frequency of a given wind direction during wet hours will likely occur in winter. Prominent wet wind directions during winter tend to be prominent wet-wind directions for all other seasons, but the frequency of these wind directions decreases as directional variability increases. The most frequent high wind speeds at all stations occur during winter and are associated with the prominent wind direction(s). Wind speeds vary seasonally, with maximum values occurring during the winter and fall and minima occurring during the summer and spring. An analysis of the temporal distribution of rainfall would be useful, as it would indicate whether wetting or drying predominates in a given location. If the wet hours occur all at once, the rest of the year acts as a drying period; in the worst case, wet hours are evenly distributed throughout the year, so that buildings never get a chance to completely dry out before the next rainfall. It would be possible to analyse the data for these trends, but such an analysis was not within the scope of the study on which this paper is based. The nature of the statistical analyses conducted in this study defined the selection criteria for the data used. Hourly wind and rain data were required, so all stations that did not collect hourly data were excluded. For example, Environment Canada stopped archiving hourly rainfall in 1999, and since that time only six-hour accumulated rainfall data are available. Because of the complexities of investigating wind/rain relationships using six-hour accumulated rainfall, only hourly data up until the end of 1998 were considered for the analysis, a total of nine years of data. Also, a study of coincident data requires a relatively complete data set for all data elements. CONCLUSIONS/RECOMMENDATIONS Ocean influence and topography greatly affects local weather conditions within the Lower Fraser Valley and Vancouver Island, and building design decisions based on weather information from a single location such as Vancouver Airport may not be appropriate for the location of the building. Vancouver Airport weather data does not represent wind-rain conditions over all locations within the lower Fraser Valley of BC. Predominant wind directions during rain as well as maximum wind speed vary with location and consequently the DRWP varies with location. From a building-envelope perspective, the findings indicate that using, for example, CSA-A440 DRWP values based on Vancouver Airport data could underestimate conditions around Burnaby Mountain and overestimate conditions in many other locations such as Kitsilano. In the case of remediating a building envelope, cost considerations could mean that one wall, perceived to be most vulnerable to wind driven rain, receives more work or attention than the other walls. If the wind direction during wet hours at the location of interest is significantly different than at the Vancouver Airport, the wrong wall could be upgraded. These results also have implications for new construction. Knowing the prevailing directions for wind-driven rain in a given location provides guidance in selecting appropriate façades to locate entrance doors or sliding patio doors, both of which are notoriously poor at resisting wind-driven rain. Wind roses of the type shown in Figure 3 can also provide some indication about the need for wider overhangs on a given elevation. Also, if (due to an unavoidable interior configuration) an entrance door must be located on an elevation with a high exposure to wind-driven rain, the designer can include a canopy or alcove to reduce the building’s vulnerability to water ingress.

11th Canadian Conference on Building Science and Technology Banff, Alberta, 2007

The analysis shows that all stations in this study exhibit a marked difference in the wind direction and frequency of higher wind speeds for wet hours versus all hours. In addition, the seasonal variation of the wind is such that the most frequent high wind speeds at all stations occur during winter and are associated with the prominent wind direction(s). An early study (Surry et al., 1995) looked at wind-rain relations for several stations across Canada during the summer months (April to September). The present study shows that, for the southwest corner of BC, wind-driven rain is just as important during the winter months when rain events are more common and wind speeds are typically higher. The conclusions of the earlier study (Surry et al., 1995) are valid for this study: that is, "DRWP values derived for a range of directions could provide useful information to sophisticated design approaches. As some building surfaces will receive less wind-driven rain than others, depending on their orientation, it may be possible to design these favourably oriented facades in a more economical manner than the facades which face the prevailing direction of wind-driven rain." The current study also adds to these conclusions that in regions where topography and other factors such as the ocean influence the weather over a small regional scale, the local variation of wind-driven rain and wind direction should be taken into account. One possible outcome of this conclusion is that the Exposure Nomograph in the Best Practice Guide for Coastal BC (CMHC, 2001) could be modified to reflect the difference in exposure categories for various directions. For example, Figure 3 indicates that southeast-facing walls receive the predominant exposure to wind-driven rain in Nanaimo, but it also shows that north-facing walls are only exposed to wind-driven rain for approximately 1% of the hours in the year. For many locations, the average wind speed increased as the rainfall rate increased. It would be preferred for wind speeds to be higher during non-wet hours, to promote drying of the building envelope, but such does not appear to be the case. Building designers should be aware that this emphasizes the need to prevent wetting of the walls during rain events and to promote drying during non-wet hours. Any design index (e.g., the Moisture Index in the National Building Code: see NBCC 2005), used to provide guidance in this regard should highlight this important finding. This study demonstrated that wind-rain relationships can vary significantly over distances as short as 10 km. This is particularly true within the Greater Vancouver Regional District (GVRD) where topography affects rainfall and wind direction. Although the GVRD has a relatively dense network of ambient air quality monitoring stations (see Figure 1), many of which measure rainfall, the spatial variation of rainfall is such that this network cannot adequately map this variation. Rain-gauge measurement of rainfall is precise at a given location, but is often not representative of rainfall over an area. As noted by the World Meteorological Organization (WMO 2000), variables in the hydrologic cycle such as precipitation exhibit large and frequent spatial variations and often exhibit rapid temporal variations. Collier (1990) noted that the use of weather radar offers the only practical approach to measuring rainfall over areas from 0.5 km2 to 5 km2. Weather radar was used to measure precipitation as early as 1948. A panel review paper on precipitation measurement and hydrology by radar is in the proceedings of the American Meteorological Society 1990 Radar Conference (P. L. Smith 1990). Weather radar provides a good measure of the spatial variability of precipitation and when “calibrated” with rain-gauge data can provide reliable measurements of rainfall intensity and amount. Canada has a good network of weather radar, and all major centres of population are well covered (Figure 4). An example of radar coverage over Vancouver is shown in Figure 5.

11th Canadian Conference on Building Science and Technology Banff, Alberta, 2007

Figure 4. Environment Canada’s weather radar sites across Canada. Black dots show the location of the radar station, and the grey circles identify the radar’s range of observation.

FFigure 5. Radar coverage for the GVRD, indicating rain showers over the area.

11th Canadian Conference on Building Science and Technology Banff, Alberta, 2007

Using weather radar for measuring rainfall is not a trivial exercise. Just as rain gauges have sources of error such as wind deformation around a gauge, splashing of raindrops and wetting of internal walls of a gauge, radar has sources of errors. These include, ground clutter when the transmitted microwaves are reflected by terrain or buildings, anomalous propagation when atmospheric conditions bend or reflect the beam to the ground, and attenuation when precipitation attenuates the beam so that precipitation further out looks weaker than it is. There are, however, techniques to correct for these and other sources of error. Hence, it is feasible to use weather radar in conjunction with existing rain gauges to provide spatially detailed measurements of precipitation for use in determining wind-rain relations. Just as precipitation can vary over distances of a few kilometres, wind can also vary as a result of local topography and other influences such as the ocean and the depth of the boundary layer. Diagnostic models exist that can take local wind observations and other weather parameters and calculate the wind direction and speed at other locations. One such model, CALMET, is a diagnostic 3-dimensional meteorological model used in air quality analysis (see Figure 6 for an example of a CALMET derived wind field).

Figure 6 Example of wind fields generated by CALMET for three vertical levels, for the hour ending at 9 AM on September 9, 1995. Wind speed is denoted by the length of the arrows, and covers a range of 0.0 - 4.6 m/s

11th Canadian Conference on Building Science and Technology Banff, Alberta, 2007

This and other diagnostic models can provide information about the wind over large areas for which standard wind measurements are too sparse to be representative. Combining wind information from diagnostic models with weather radar observations of rainfall provides an opportunity to develop wind-rain relations over regions such as the GVRD where wind speed and direction, and rainfall varies significantly over short distances because of the local topography and the influence of the ocean. REFERENCES CSA, 1998. User Selection Guide to CSA Standard A440-98, Windows. 1998, Canadian

Standards Association, Rexdale, ON (more recent editions are available). CMHC, 2001. Best Practice Guide to Wood-Frame Envelops on the Coastal Climate of British

Columbia. CMHC, 2004. Wind-Rain Relationships in Southwestern British Columbia. Report written by

Levelton Consultants. 2004 Canada Mortgage and Housing Corporation, Ottawa, ON. Collier, C.G., 1990: Assessing and forecasting extreme rainfall in the United Kingdom, Weather,

45,4, pp. 103-112. National Research Council, 2005. National Building Codes of Canada 2005. Canadian

Commission on Building and Fire Codes. Smith, P.L., 1990. Precipitation Measurement and Hydrology: Panel Report Radar in

Meteorology: Battan Memorial and 40th Anniversary Radar Meteorology Conference held in Boston in 1990 , pp 607-618.

Surry, D., P.F. Skerlj and M.J. Mikitiuk, 1995. An Exploratory Study of the Climatic Relationships Between Rain and Wind. Boundary Layer Wind Tunnel report BLT-2230-1994 for Canada Mortgage and Housing Corporation.

Welsh, L.E., W.R. Skinner and R.J. Morris, 1989. A Climatology of Driving Rain Wind Pressures for Canada (Draft Report).

WMO, 2000: Precipitation Estimation and Forecasting, WMO Operational Report No. 46 by C.G. Collier.