Actually, it is very simple to generate scenarios. I think for wind PDF we use the Weigh-bull probability density function, and it is also provided in matlab. So, you can give it some information, which are related to your historical data, and then get your samples by rnd function.
There are other solutions for that too. For example you could use the LHS or time series for that.
After that you may want to reduce your scenarios that there are different ways for that.
These references might help:
1-Generating Short-Term Probabilistic Wind Power Scenarios via Non-Parametric Forecast Error Density Estimators
2-A scenario-based analytical method for probabilistic load flow analysis
Article A scenario-based analytical method for probabilistic load fl...
Article A scenario-based analytical method for probabilistic load fl...
3-Probabilistic load flow computation using Copula and Latin hypercube sampling