Let's assume that we make forecasting of another metric partially based on forecasts of the weather forecast, e.g. of temperature, pressure, then we can potentially obtain those forecasts from one of public APIs and have some information about future values of these features and do more precise prediction of another parameter taken as a label.
What is the approach to use in this case if one or more of the features in multivariate forecasting have some forecasted values for the predicted horizon? It looks that in this case not only values from historical data can be used but also predicted values, though not clear how to organize the model in this case, e.g. it can be multivariate LSTM.