I want to extract the seasonality from my sarima model.

My dataset is in the format of this:

ds: dates
y: metric

This is weekly data

I have done the following:

ts_data <- ts(for_seasonality$y, frequency = 52, start = c(year(start_date), 1 + for_seasonality$week_num[1]))

sarima_model <- auto.arima(ts_data, seasonal = TRUE)

summary(sarima_model)

Series: ts_data 
ARIMA(0,1,2)(2,0,1)[52] with drift 

Coefficients:
          ma1      ma2    sar1    sar2    sma1   drift
      -0.4227  -0.1011  0.2145  0.1747  0.1393  0.0006
s.e.   0.0510   0.0558  0.5443  0.2208  0.5584  0.0023

sigma^2 = 0.003411:  log likelihood = 533.48
AIC=-1052.95   AICc=-1052.65   BIC=-1025.43

Training set error measures:
                       ME       RMSE        MAE  MPE MAPE      MASE      ACF1
Training set 0.0002524145 0.05785966 0.03492112 -Inf  Inf 0.3350916 0.0122546

How do I extract the seasonality directly from the model instead of using stl or decompose? Is it even possible?