Hung Chak Ho

Estimating daily maximum air temperature from MODIS in British Columbia, Canada

Air temperature (Ta) is an important climatological variable for forest research and management. Due to the low density and uneven distribution of weather stations, traditional ground-based observations cannot accurately capture the spatial distribution of Ta, especially in mountainous areas with complex terrain and high local variability. In this paper, the daily maximum Ta in British Columbia, Canada was estimated by satellite remote sensing. Aqua MODIS (Moderate Resolution Imaging Spectroradiometer) data and meteorological data for the summer period (June to August) from 2003 to 2012 were collected to estimate Ta. Nine environmental variables (land surface temperature (LST), normalized difference vegetation index (NDVI), modified normalized difference water index (MNDWI), latitude, longitude, distance to ocean, altitude, albedo, and solar radiation) were selected as predictors. Analysis of the relationship between observed Ta and spatially averaged remotely sensed LST indicated that 7 × 7 pixel size was the optimal window size for statistical models estimating Ta from MODIS data. Two statistical methods (linear regression and random forest) were used to estimate maximum Ta, and their performances were validated with station-by-station cross-validation. Results indicated that the random forest model achieved better accuracy (mean absolute error, MAE = 2.02°C, R2 = 0.74) than the linear regression model (MAE = 2.41°C, R2 = 0.64). Based on the random forest model at 7 × 7 pixel size, daily maximum Ta at a resolution of 1 km in British Columbia in the summer of 2003–2012 was derived, and the spatial distribution of summer Ta in this area was discussed. The satisfactory results suggest that this modelling approach is appropriate for estimating air temperature in mountainous regions with complex terrain.

A comparison of urban heat islands mapped using skin temperature, air temperature, and apparent temperature (Humidex), for the greater Vancouver area

Apparent temperature is more closely related to mortality during extreme heat events than other temperature variables, yet spatial epidemiology studies typically use skin temperature (also known as land surface temperature) to quantify heat exposure because it is relatively easy to map from satellite data. An empirical approach to map apparent temperature at the neighborhood scale, which relies on publicly available weather station observations and spatial data layers combined in a random forest regression model, was demonstrated for greater Vancouver, Canada. Model errors were acceptable (cross-validated RMSE=2.04 °C) and the resulting map of apparent temperature, calibrated for a typical hot summer day, corresponded well with past temperature research in the area. A comparison with field measurements as well as similar maps of skin temperature and air temperature revealed that skin temperature was poorly correlated with both air temperature (R(2)=0.38) and apparent temperature (R(2)=0.39). While the latter two were more similar (R(2)=0.87), apparent temperature was predicted to exceed air temperature by more than 5 °C in several urban areas as well as around the confluence of the Pitt and Fraser rivers. We conclude that skin temperature is not a suitable proxy for human heat exposure, and that spatial epidemiology studies could benefit from mapping apparent temperature, using an approach similar to the one reported here, to better quantify differences in heat exposure that exist across an urban landscape.

Delineation of Spatial Variability in the Temperature-Mortality Relationship on Extremely Hot Days in Greater Vancouver, Canada.

Background: Climate change has increased the frequency and intensity of extremely hot weather. The health risks associated with extemely hot weather are not uniform across affected areas owing to variability in heat exposure and social vulnerability, but these differences are challenging to map with precision. Objectives: We developed a spatially and temporally stratified case-crossover approach for delineation of areas with higher and lower risks of mortality on extremely hot days and applied this approach in greater Vancouver, Canada. Methods: Records of all deaths with an extremely hot day as a case day or a control day were extracted from an administrative vital statistics database spanning the years of 1998–2014. Three heat exposure and 11 social vulnerability variables were assigned at the residential location of each decedent. Conditional logistic regression was used to estimate the odds ratio for a 1°C increase in daily mean temperature at a fixed site with an interaction term for decedents living above and below different values of the spatial variables. Results: The heat exposure and social vulnerability variables with the strongest spatially stratified results were the apparent temperature and the labor nonparticipation rate, respectively. Areas at higher risk had values ≥ 34.4°C for the maximum apparent temperature and ≥ 60% of the population neither employed nor looking for work. These variables were combined in a composite index to quantify their interaction and to enhance visualization of high-risk areas. Conclusions: Our methods provide a data-driven framework for spatial delineation of the temperature-–mortality relationship by heat exposure and social vulnerability. The results can be used to map and target the most vulnerable areas for public health intervention. Citation: Ho HC, Knudby A, Walker BB, Henderson SB. 2017. Delineation of spatial variability in the temperature–mortality relationship on extremely hot days in greater Vancouver, Canada. Environ Health Perspect 125:66–75; http://dx.doi.org/10.1289/EHP224

A Spatial Framework to Map Heat Health Risks at Multiple Scales.

In the last few decades extreme heat events have led to substantial excess mortality, most dramatically in Central Europe in 2003, in Russia in 2010, and even in typically cool locations such as Vancouver, Canada, in 2009. Heat-related morbidity and mortality is expected to increase over the coming centuries as the result of climate-driven global increases in the severity and frequency of extreme heat events. Spatial information on heat exposure and population vulnerability may be combined to map the areas of highest risk and focus mitigation efforts there. However, a mismatch in spatial resolution between heat exposure and vulnerability data can cause spatial scale issues such as the Modifiable Areal Unit Problem (MAUP). We used a raster-based model to integrate heat exposure and vulnerability data in a multi-criteria decision analysis, and compared it to the traditional vector-based model. We then used the Getis-Ord Gi index to generate spatially smoothed heat risk hotspot maps from fine to coarse spatial scales. The raster-based model allowed production of maps at spatial resolution, more description of local-scale heat risk variability, and identification of heat-risk areas not identified with the vector-based approach. Spatial smoothing with the Getis-Ord Gi index produced heat risk hotspots from local to regional spatial scale. The approach is a framework for reducing spatial scale issues in future heat risk mapping, and for identifying heat risk hotspots at spatial scales ranging from the block-level to the municipality level.

Estimation of Continuous Urban Sky View Factor from Landsat Data Using Shadow Detection

Sky View Factor (SVF, a dimensionless value between 0 and 1 representing obstructed and unobstructed sky, respectively) has an important influence on urban energy balance, and is a key contributor to the Urban Heat Island (UHI) effect experienced by heavily built up regions. Continuous urban SVF maps used in modeling the spatial distribution of UHI can be derived analytically using Lidar data; however, Lidar data are costly to obtain and often lack complete coverage of large cities or metropolitan areas. This study develops and validates a method for estimating continuous urban SVF from globally available Landsat TM data, based on the presence of shadows cast by SVF-reducing urban features. SVF and per-pixel shadow proportion (SP) were first calculated for synthetic grid cities to confirm a logarithmic relationship between the two properties; then Lidar data from four US cities were used to determine an empirical regression relating SP to SVF. Spectral Mixture Analysis was then used to estimate per-pixel SP in a Landsat 5 TM image covering the Greater Vancouver Area, Canada, and the empirical regression was used to calculate SVF from per-pixel SP. The accuracy of the resulting SVF map was validated using independent Lidar-derived SVF data (R2 = 0.78; RMSE = 0.056).