Based on wavelength, ultraviolet radiation (UVR) is subdivided into UVA (315–400 nm), UVB (290–315 nm), and UVC (200–290 nm). The UVC band is the most energetic radiation but does not penetrate the atmosphere. UVB and UVA that penetrate the atmosphere have the highest biological impact to human health.
Specifically, skin cancer is one of the most significant public health problems. More people are diagnosed with skin cancer each year in the U.S. than all other cancers combined [
1]. UVR has been associated with three major kinds of skin cancer: basal cell carcinoma, squamous cell carcinoma, and cutaneous malignant melanoma [
2]. According to the American Cancer Society, over 100,000 new cases of melanoma skin cancer and more than 7000 death of melanoma will occur in 2020 [
3].
Quantifying the distribution of UV radiation in a given landscape and mapping solar radiation on humans can help elucidate the relationship between UV exposure and children’s health and guide the design of sun protection programs. Current methods for estimating individual UVR exposure include field measurements using personal dosimeters, simulations based on the human body and solar position modeling, and satellite-based approaches. However, these methods have their limitations:
This study thus provides a new approach of integral calculation using six-directional (up, down, south, north, east, and west) field-measured UVR data and an estimated body exposure ratio (ER) to calculate both children’s and adults’ UVR exposure in a given landscape. Results are analyzed to (1) identify children’s and adults’ whole body ERs in different seasons in consideration of clothing conditions and solar positions, (2) construct a model to predict individual UVR exposure for children and adults, and (3) assess the model’s performance and limits.
Literature Review
Increasing epidemiological studies have supported the relationship between sunlight UVR exposure and enhancement of skin damage. Continuous exposure to sun or artificial UVR intentionally will increase the risk of getting skin cancer. However, the public awareness of the risk is not optimal, skin cancer rates continues to grow each year in all age groups, including in the younger population [
2].
The variability of solar UV radiation at the Earth’s surface depends on the combined effect of several factors. For instance, atmospheric gases and aerosols, ozone column, cloud cover, solar zenith angle and distance from the sun (season and time), latitude, longitude and altitude, surface albedo, trees, and other objects in a given landscape [
14]. An individual’s UV exposure is related to the direct, diffuse and reflected UVR irradiance, body posture and activity, duration of exposure in the open, skin type, and protective behavior [
13,
15].
UVR exposure during childhood is a critical period for the increase in skin cancer risk later in life [
4,
16]. An epidemiological study revealed that people who migrated from low UVR areas to high UVR areas at an older age had a lower risk of getting skin cancer compared to those who arrived at younger ages or as children [
17,
18]. Among US children, the incidence of sunburn is high: 29% to 83% over the entire summer season [
19,
20]. In 2010, approximately one third of US teens (14 to 17 years old) reported having had a sunburn during the previous 12 months [
20]. Due to the differences in body characteristics of children, their UVR exposure calculations were separated from those of adults in this study.
Electronic or personal UV dosimeters have been tested and used in various studies for the measurement of personal UV exposure [
10,
15,
21,
22]. Wrists and shoulders were found to be valid sites on the body for mounting personal dosimeters in previous studies (e.g., Vanos et al. [
15]). However, a measurement made by personal UV dosimeters facing upwards was found to underestimate full UVR exposure [
15]. Due to direct, diffuse, and reflected UV irradiances, inclined surfaces hold a substantial amount of radiation. Measuring only horizontal surfaces is thus an inadequate indicator of environmental irradiance [
23,
24].
Moreover, personal UVR dosimeter measurements, such as photosensitive dosimeters and polysulphone film band dosimeters, are strongly related to an individual’s specific position and environmental shade conditions, are costly, and are prone to behavioral biases [
8,
13,
21,
22]. For example, 196 children in Stockholm, Sweden were asked to carry dosimeters and pedometers in different outdoor environments from May–June 2004 to test their daily UVR exposure ratio (ER) and UVR exposure. The range of ER in environments with trees and little vegetation were as wide as 4–60% [
22]. Pagels et al. [
21] conducted a similar experiment in Sweden among 2nd, 5th, and 8th graders to determine their daily UVR exposure. ER and sky view factors were found to be uncorrelated among the 8th graders but positive for the 2nd graders and negative for the 5th. Previous studies like these show that it is difficult to identify a consistent and quantitative relationship between individual UVR exposure and ER in a specific environment due to the complexity of the environment’s shade conditions and daily behaviors.
Numerous efforts have been made to calculate solar irradiance in the directions of typically oriented surfaces of the human body using three-dimensional computer graphics techniques and calculations. For instance, a study by Pope and Godar [
25] provided a geometric conversion factor to convert horizontal erythemal UV irradiance to a cylinder model to represent human body UVR exposure. However, this study didn’t consider human seasonal clothing condition, and wasn’t validated. Streicher et al. [
26] produced a sun-tracing and irradiance algorithm to calculate UVR exposure of each anatomical area of human body. Hoeppe et al. [
27] also developed a virtual three-dimensional measuring system, ASCARATIS, to calculate UVR exposure of human skin, which contained about 20,000 triangles of the human body. Vernez et al. [
28] used a 3D numeric model SimUVEx to compute daily UVR dose and ER for various body sites and body postures. A recent modification for this model was later made by Religi et al. [
13] using a manikin with two resolutions (high: 13,476 vertices; low: 837 vertices) for 45 anatomical zones. These sophisticated models have other useful applied areas, such as identifying the risk of skin cancer at some specific body part, although reflecting whole body UVR exposure in a real outdoor environment is difficult.
A human UVR exposure ratio (ER) refers to the ratio between the UVR amount received by a specific body site and the corresponding UVR amount received by a flat horizontal surface at ground level [
29]. It has been used in previous studies to estimate individuals’ UVR exposure by considering the ambient UVR [
10,
30]. For example, Downs and Parisi [
30] developed an approach to estimate ER of each body part for different latitudes under solar zenith angles ranges (0°–30°, 30°–50°, and 50°–80°), although this approach didn’t consider human body posture and activities. Weihs et al. [
10] determined ER for four different outdoor activities by measuring 10 different positions of the body. However, the result from the study is hard to be applied to different location under different solar positions. Vernez et al. [
29] developed a general regression model for predicting ER for different anatomical body parts and validated it by comparison with the SimUVEx model. However, the estimation or measurements of the ER in these studies were made in an open environment, which can’t be used for complicated environmental shade conditions.
To collect more relevant environmental information on UV exposure of the whole human body, a new model calculating individual UVR exposure has been developed based on the principle of UVR transmission in the environment [
31,
32] and its interaction with human skin. Due to differences in body surface area between children and adults, two versions of UV exposure models were separately calculated for this study. The accuracy of the models is presented through a comparison of the results from Vernez et al.’s [
29] regression model using locally measured UVR data in College Station, Texas. This model was compared to the SimUVEx model and showed a high agreement (R
2 = 0.988). As such, it can be used to accurately predict ER and UVR amount based on available data, e.g., global UV erythemal irradiance measured at ground surface stations or inferred from satellite information.