what could be the best strategy to horizontally upscale the scylla cluster in the anticipation of high load due to an event
Depends on your topology, but you will typically want to scale the cluster in increments according to your RF. The usual practice is to have RF=3 where each replica is located under a different availability zone.
Thus, you would scale add one node to each AZ to sustain the load you are expecting.
Unless you are planning to scale up your cluster, it also makes sense to remain under the same instance as your existing ones.
And of course, remember to check out the docs on Adding a New Node Into an Existing ScyllaDB Cluster (Out Scale) | ScyllaDB Docs