How could we interpret a SI Scatter Index and RMSE?

1. SI is RMSE divided by the average value of the observed values (or the predicted values? am confused)?

Your understanding is right.

Scatter index is calculated by dividing the root-mean-square deviation (RMSD) or root-mean-square error (RMSE) with mean of the observations at each grid point and multiplying it with 100. It presents the percentage of RMS difference with respect to mean observation or it gives the percentage of expected error for the parameter.

Mathematically,

\begin{align*} \text{Scatter index (SI)} = \frac{\text{RMSE}}{\overline{X}} \end{align*}

is SI = 25% acceptable? (is the model good enough? )

The Scatter index (SI) is a normalized measure of error, often reported as a percent. Lower values of the SI are an indication of better model performance. Like the RMSE, ambiguities exist in the definition of the scatter index, with authors either defining it as the standard deviation of the errors (i.e.,demeaned RMSE) divided by the mean of the observations (Mentaschi et al. 2013), (Ris etal. 1999; Rogers et al. 2012; Akpinar et al. 2012).

More on Evaluation Statistics Computed for the Wave Information Studies (WIS)