Once in a while, the weather pattern setting up across the area makes it really difficult to pin down a forecast with any kind of real confidence, and this far out that is how it's going so far for the upcoming weekend, especially as we head into Saturday night and Sunday. The culprit is a combination of fairly flat westerly flow aloft and an approaching cold front that looks as if it will slip into central NC from the northwest Saturday night and then waver north to west, stalling at times but with a variety of solutions from computer models as to where the boundary lies at any given time.
As I write this on Tuesday, we have reasonably good confidence regarding the temperatures and rain chances through Friday and a little less so on Saturday, with a cool, bright Thursday very likely, a few more clouds and somewhat milder temperatures Friday and then a nice warmup to start the weekend (highs in the mid 60s to around 70), but for the holiday on Sunday we have a legitimate chance of seeing high temperatures in the Triangle anywhere from around 50 if the front stalls south of us and we develop a patch or two of rain, to low 70s if the front is a little farther north and we have a few periods of sunshine.
The graph I posted here is output from a a collection many different runs (42 all together) of both the American Global Ensemble Forecast System (GEFS) model and the Canadian Ensemble Prediction system, with each run having slightly different starting conditions or model characteristics. For each 6-hourly time point on the graph, the "boxes and whiskers" show the lower 25% of results as the lower whisker, the middle 50% of results as the yellow box, and the upper 25% of results as the upper whisker, and the dark line inside the box is the median value of all the different runs. The output is shown at 8PM, 2AM, 8AM and 2PM for each day.
If you look at those boxes and whiskers for the first few days of the forecast, you'll see that temperatures, for example, are quite similar among most of the models, but the picture changes considerably through the weekend, and on Sunday the 17th at 2PM even the 50% of model runs in the yellow box cover a spread of possible temperatures for RDU from about 53 to 71 degrees, and the range covered by all 42 models runs, from the bottom of the whiskers to to the top, ranges from a 2PM temperature as cold as 32 degrees up to as warm as 75! When we also check the main operational models (not ensembles), we find the American GFS going for a high in the mid 50s, and the European model now projecting something close to 70 on Sunday after showing a much cooler outlook yesterday.
The result for now is that you can plan with good confidence on a mostly sunny or sunny day for Thursday (notice how in the bottom of the three graphs very few of the model runs show any cloud cover at all that day). If you're making plans for Sunday, though, it would be a good idea to keep checking back as we progress through the week, and be prepared for a variety of possible outcomes and potential changes to the forecast depending on where that front sets up shop.