CIESIN icon SEDAC icon NASA logo home page icon
Stratospheric Ozone and Human Health Project

UV dose icon

Model Validation

Contents -- there are currently two components to the model validation:

Model Validation and Sensitivity Analysis

(1)   Clear Sky Validation: Modeled vs. Measured Irradiance

The accuracy of the model was tested by comparing model output to clear sky measurements taken with a Biospherical Instruments GUV-511 device. This instrument was installed at the University of Michigan's Department of Atmospheric, Oceanic, and Space Sciences as part of a nationwide UV monitoring network funded by Bausch & Lomb. The GUV-511 is a temperature-controlled unit providing spectral measurements in units of uW cm-1 nm-1 at 305 nm, 320 nm, 340 nm, and 380 nm with FWHM bandwidths ranging from 8 to 10 nm. Erythemal dose and Photosynthetically Active Radiation (PAR) measurements are also provided. A Compaq ProLinea 4/25s PC collects the data using a data acquisition system provided by Global Environmental Engineering Consultants, Inc. that records one-minute averages in daily output files. Clear sky cases only will be considered in this validation study due to lack of data on cloud optical depths over the local area. An inherent difficulty in validating the cloud model is the high spatial and temporal variability of cloud; partly cloudy skies are continually changing over the course of minutes, and reflection from edges of cloud along with diffuse contributions from clear and cloudy areas not directly in the line of sight make this a difficult, if not impossible problem to consider. Even in seemingly invariant overcast conditions, cloud parameters remain fluid as evidenced in data collected from August 2, 1994 (Figure 1). Overcast conditions persisted throughout the day, but considerable variation in the stratus layer is evident in the UV irradiance measured at the surface.

Clear sky conditions were observed on September 11, 1994 in the southeast Michigan region. Total column ozone in the 1°x1.25° grid cell over the Ann Arbor coordinates of 42.15°N, 83.45°W for this day was taken from the on-line database of TOMS Meteor-3 observations. NASA made these data available in near real-time with the understanding that these data are preliminary figures and should not be used for detailed research purposes until validated. With this caveat, the daily column ozone value over this location was 305 Dobson units (DU). This value was coded into the ozone extinction program and the model ozone profile for mid-latitude autumn was normalized to this value. Visibility was relatively high on this day; a correspondingly low surface layer aerosol extinction coefficient of 0.2 km-1 was incorporated into the boundary layer aerosol model.

UV irradiance was calculated every half hour from 8AM to 7PM Eastern Daylight Time (EDT) based on variation in solar zenith angle. Ozone and aerosol levels were assumed constant over the day due to lack of appropriate measurements. A comparison of calculations and instrument output at 305 nm, 320 nm, and 340 nm is shown in Figure 2. Model estimates showed variation of less than 8% at 305 nm and 320 nm, and less than 4% at 340 nm relative to measurements. With uncertainties in boundary layer aerosol and ozone in addition to the non-validated total column ozone value, these estimates give confidence that this model provides an accurate representation of the UV budget at the surface of the Earth.

FIGURE

Figure 1.   UV irradiance under overcast conditions on 8/2/94.

FIGURE

Figure 2.   Model comparison to clear sky measurements on 9/11/94.

(2)   Inter-model comparison

An initial comparison of model output has been conducted with a discrete ordinates radiative transfer model developed at the National Center for Atmospheric Research (NCAR) by Dr. Sasha Madronich. Results from the NCAR model have been published extensively in the literature and used in the United Nations Environment Programme's Environmental Effects of Ozone Depletion report.

This initial validation activity compared model outputs between the two models calculated at varying solar zenith angles from 0° to 80° given the following atmospheric conditions and parameters:


Model Conditions and Parameters: Test Case

  • total column ozone = 300 DU
  • clear sky conditions (no cloud)
    		•
  • aerosol optical depth = 0.38
  • surface albedo = 0.1
    		•

    Erythemal dose and UV Index values were calculated and compared using these initial conditions. Comparison plots are shown in Figures 3 and 4 below. Results are also given in tabular format (in units of Watts per square meter for erythemal dose) along with the percent difference between values computed using each model:

    Figure 3.   Erythema dose calculation comparison between CIESIN and NCAR models

    Figure 3.   Erythema dose calculation comparison between CIESIN and NCAR models.

    Figure 4.   Ultraviolet Index calculation comparison between CIESIN and NCAR models.

    Figure 4.   Ultraviolet Index calculation comparison between CIESIN and NCAR models.



    Erythema Dose UV Index
    Solar Zenith Angle: CIESIN NCAR (% diff.) CIESIN NCAR (% diff.)
    0 3.24e-1 3.13e-1 3.5 12.8 12.5 2.4
    10 3.13e-1 3.01e-1 4.0 12.5 12.0 4.2
    20 2.76e-1 2.66e-1 3.8 11.0 10.6 3.8
    30 2.21e-1 2.14e-1 3.3 8.8 8.6 2.3
    40 1.58e-1 1.55e-1 1.9 6.3 6.2 1.6
    50 9.88e-2 9.83e-2 0.5 3.9 3.9 0.5
    60 5.09e-2 5.20e-2 2.1 2.0 2.1 1.9
    70 2.00e-2 2.12e-2 5.7 0.80 0.85 5.9
    80 4.80e-3 5.87e-3 18.0 0.19 0.23 17.0

    Table 1.   Data calculated from the CIESIN and NCAR models used to create Figures 3,4.


    These results from the CIESIN model show excellent agreement with the NCAR model results, with most values showing less than a 4% difference. The higher variability seen at the higher zenith angles is likely attributed to different spherical geometry schemes that consider the curvature of the Earth and its atmosphere. The NCAR discrete ordinates routine is a plane-parallel model containing a crude Chapman function correction to consider sphericity, while the CIESIN model explicitly calculates the impact of the curvature of the Earth's atmosphere at all solar zenith angles. This spherical geometry component of the CIESIN model is a key feature of the model, as its original purpose was to analyze line spectra emissions from stratospheric and mesospheric oxygen as observed from the High Resolution Doppler Imager (HRDI) instrument on board NASA's Upper Atmosphere Research Satellite (UARS). The close agreement between the two model's output and the inclusion of the spherical geometry component provide additional confidence in the UV dose values generated by the CIESIN model in the UltraViolet Interactive Service (UVIS) and in the daily erythemal and UV Index maps available through the Stratospheric Ozone and Human Health project.

    HOME
    Home     		UV DATA
    UV data     		SEARCH
    Search     		HEALTH
    Health     		LINKS
    Links     		GUIDES
    Guides