4.1 Correlation with load
Figure 34 shows the load and wind profiles for in January 2000. As expected, the load waveform has a regular repetitive pattern, which is not the case for wind production. If solar production was shown, it would show a repeating waveform and, at this level of resolution, would seem to closely correlate with load. The correlation of solar power, both and CST, with load demand over a period of a week in the western USA can be seen in Figure 35 . In this figure, the load is equivalent to the sum of all production. It can be seen that power from both solar PV and CST systems correlate well with demand. Individual PV panels display extreme variation, but the variation they exhibit is smoothed out with aggregation. Because of this smoothing effect, the daily production profile of solar should be reasonably consistent. , When this is combined with a repeating daily waveform shown in Figure 35, the profile of solar generation could be considered far more deterministic than wind. The reduced randomness in solar generation should reduce the complexity of scheduling dispatchable generation in comparison with wind.
Looking more closely into correlation with load, Figure 36 and Figure 37 show the penetration of wind and solar respectively at different load levels . Three scenarios depicted for these two graphs; 2006, 2010T and 2010X:
- 2006 (base case) - 2100 MW of wind and 330 MW of solar. It is not specified what these values are as a percentage of overall generation
- 2010T (20% renewable energy) - 7500 MW wind (16%) and 1900 MW of solar (4%)
- 2010X (33% renewable energy) - 12500 MW of wind (27%) and 2600 MW for solar (6%).
Load is lightest at the tenth decile and heaviest at the first. This data from CAISO  again shows solar has better correlation with load, with penetration the highest during the first decile while, in contrast, wind has its highest penetration in the tenth decile. These findings show correlation between solar and load to be greater, but it is important to bear in mind that wind generates throughout the night when there is light load, while solar does not generate. This may be a tipping factor for the correlation to favour solar. It would be instructive to see the analysis repeated with wind generation during the night discarded. Looking specifically at the first decile (peak load) for the 2010T scenario (16% wind and 4% solar as a percentage of total generation capacity) wind can be seen to only provide 4% of generation, which corresponds to a quarter of its full generation capability. For the same scenario, solar is seen to provide 2out of a possible 4% of total generation capability.
Figure 38 shows the for each decile for wind and solar for the 2010T scenario. It can be seen that solar uses more of its capacity than wind during the top three load periods, making it a more efficient resource for reducing peak load.
Another study in  analysed three years worth of load, wind and solar data from three sites - two in Figure 39 shows, there is a greater correlation between load and solar generation for all three sites. This reinforces the conclusion in  that solar generation has a stronger correlation with load demand.and one in - and performed a correlation on the data for each of the four seasons. As