VARIABILITY OF SOLAR RADIATION DATA OVER SHORT TIME INTERVALS Frank Vignola Department of Physics 1274-University of Oregon Eugene, OR 97403-1274 e-mail: fev@darkwing.uoregon.edu ABSTRACT In order to evaluate satellite and cloud cover models, it is useful to understand the short-term variability of solar ra- diation. This article examines at the variability of beam and global solar radiation over short time intervals. First, 5-minute data are compare with data collected 5 and 15 minutes later. Second, data collected during the middle 5- minute period of an hour are compared to hourly average data. These data comparisons show a ‘Nugget Effect’ simi- lar to that observed when verifying satellite modeled solar radiation values with ground-based data. 1. INTRODUCTION It is difficult to evaluate solar radiation values obtained from models utilizing satellite or cloud cover observation data with solar radiation data measured at a specific loca- tion [1]. The satellite modeled data measures solar radia- tion over a large area, typically from 100 to 10,000 km 2 . Ground-based cloud cover observations happen at best once an hour. Ground-based solar radiation measurements are made continuously but from just one location. The mean bias error between satellite-derived and measured solar radiation data can be small, but the standard devia- tion between satellite-derived and measured data is on the order of 20%. This is called the ‘Nugget Effect’ and repre- sents a limit to the comparison between satellite-derived values and measured data [2]. Even if the satellite-derived values exactly estimate the incident solar radiation for an area, the ground-based measurement in the middle of this area is not expected to match the satellite-derived values. Even over an hour, the ground-based station will not experience all the variations in the cloud cover observed by the satellite. In addition, the satellite modelers typically use one snapshot to repre- sent each hour’s worth of data. The temporal and spatial variation of the solar resource are somewhat related. Cloud patterns move over a solar moni- toring site and produce a variability in the measured solar radiation. While this cloud pattern doesn’t all pass over the ground-based observer and the pattern does change over time, on average it does have characteristics that are in- dicative of the variability of the solar radiation over the area. In other words, the variation of the solar radiation over area should have similar properties to the variation of solar radiation over time. Therefore, to better evaluate the com- parisons between satellite-derived and measured values, it is worthwhile to examine the short term variability of solar radiation. Two aspects of the variability of solar radiation over short time intervals are studied in this article. First, the variabil- ity of solar radiation from one time interval to another is examined. Second, solar radiation data obtained over a short interval in the middle of the hour are compared to the data obtained over the whole hour. The applicability of these results to satellite modeled and cloud cover modeled solar radiation data is then discussed. 2. SHORT TIME INTERVAL COMPARISON Before looking at the variability of the data in this study, it is necessary to say a few words about the data. The data come from the solar radiation monitoring station in Eugene, Oregon. The beam data are measured with an Ep- pley NIP pyrheliometer and the global data are measured with an Eppley PSP pyranometer. Data are sampled at 2-