Surface Temperature Anomalies and Solar Irradiance

Abstract:

This paper shows that enhanced solar irradiance and cold temperature anomalies are correlated. This observation is based on the analysis of 20 years of high-resolution surface irradiance and temperature data in North America. This result may have important energy use implications for northern regions where heating constitutes a large fraction of total energy consumption. Indeed, solar power could be harnessed either directly or indirectly, via PV electrical generation, to displace natural gas peak consumption during cold snaps.

Introduction:

A better understanding of the relationship between temperature anomalies and solar irradiance is important for identifying solar power generation resource management opportunities as temperature is a key driver of utilities’ electrical loads (Hor et al. [1], Guam et al. [2]). Whereas the relationship between irradiance and warm temperature anomalies (heat waves) has been extensively documented and has led to the recognition of PV’s summer peak shaving capacity, little work has focused on the relationship between solar irradiance and cold temperature anomalies. This relationship may be important in northern climates where heating constitutes a large fraction of energy consumption. Wintertime cold temperature anomalies are often associated with gas demand peak hikes from heating demand that impact the price of gas and thus [gas-generated] electricity. A higher solar resource during these episodes could contribute to mitigating electricity demand/price hikes, either directly via solar heating, or more indirectly, via PV electricity generation.

Methods:

• Irradiance Data

For this study we use the high-resolution SolarAnywhere irradiance data as experimental support. SolarAnywhere irradiance data are derived from satellite images using a model developed at the State University of New York at Albany (Perez et al.[3]). The model is implemented commercially by Clean Power Research. Hourly irradiance data are available from 1/1/1998 to the present. Figure 1 illustrates average global horizontal irradiance across the United States.

The underlying model is a semi empirical whereby irradiance is produced by modulating a simple radiative transfer scheme with a cloud signal. The cloud index -- the measure of the cloudiness of a given satellite image pixel -- is calculated from satellite images and combined with the simple radiative transfer model defined by aerosol optical depth (AOD), water vapor, and atmospheric ozone content. The model produces gridded global irradiance (GHI). Clear Sky GHI is the global irradiance under the assumption that no clouds are present. This is the maximum solar irradiance possible at the planet’s surface for a given hour, location and turbidity conditions, discounting phenomena such as enhanced irradiance transients resulting cloud edge effects. The Clear Sky index (Kt*) is calculated as the fraction of hourly GHI over hourly clear sky GHI as shown by equation 1.

For this study we acquired hourly 1-km resolution satellite-derived GHI data spanning a 20-year period from 1/1/1998 to 12/31/2017. Irradiances were subsampled at 1-degree (100 km) grid spacing, amounting to 1114 points across the continental US (figure 2) including portions of Southern Canada and Northern Mexico. Data were screened for questionable/missing values and any data with potential issues, e.g. negative values, were discarded. In addition, data were further screened by removing points with less than 50 Wm^-2 (i.e., chiefly values temporally close to sunrise and sunset). These points were not included because of possible clear sky index ratio distortions at low sun angles.

In addition to gridded satellite-derived irradiances, we also acquired Surface based hourly irradiance data from the Surface Radiation Budget Network (SURFRAD) to validate this study’s results with actual measurements at selected locations. SURFRAD irradiance and temperature measurements are considered to be of the highest quality (Augustine et al. [4]). The SURFRAD station irradiance acquired here span 15 years of data, overlapping with the satellite derived GHI data period.

• Temperature Data

Hourly ground temperature data were also sourced from the SolarAnywhere database. Meteorological data in SolarAnywhere originate from the National Forecast Database (NDFD) and from interpolated Meteorological Terminal Aviation Routine Weather Report (METAR) stations during 1998-2004. Temperature data are treated in the same manner as the irradiance data in terms of quality control. Temperature data were subsampled on the same 1-degree grid as irradiances. Furthermore, temperature data were only considered during daylight hours corresponding to irradiances greater than 50 W/m².

Long term, twenty-year hourly temperature averages were calculated for each point and each day. Hourly temperature anomalies were calculated by subtracting the 20-year average values from each hourly value.

3) Temperature anomaly binning

The hourly temperature anomalies were used to define 1 ⁰C bins. Hourly Kt* values were then averaged for each bin and for each grid point. The considered range of temperature anomaly bins considered for this study extends from -40 ⁰C to +30 ⁰C. Temperature bins outside this range were considered extreme outlier events and were omitted from the analysis.

Results:

Figure 4 shows the result of the Kt* binning for all points and all hours. Increasing trends in mean clear sky index with both warm and cold temperature anomalies are clearly visible. The mean Kt* minimum occurs at approximately -12 C.

This result suggests that higher solar irradiances occur the during coldest temperature anomalies. A possible cause is that cloudy days reduce maximum temperature as irradiance is lower, keeping the daytime temperatures below normal. Higher irradiances are also observed for the warmest temperature anomaly bins. These are associated with hot and mostly clear days, typically occurring in summer.