Azure SQL Database Data IO Usage frequently spiking to 100% and staying there unpredictably

Question

Our database is having an issue with DTU percentage spiking to 100% during bursts of high traffic to our webapp. There is a stored procedure that runs on almost every page on our site that we are aware needs to be optimized, and we have taken some steps to lessen its impact on the database (creating suggested indexes, moving the database from service level S3 -> S4, etc.).

However, we are seeing an issue with Data IO % going from a constant near 0% usage to spiking to 100% and staying there for a while in some cases. We have a new Azure Automation that scales our database down to S3 at night (10PM CST), and back up to S4 in the morning on weekdays (6AM). This morning when the database was scaled back up to S4 after remaining at S3 all weekend, the data IO percentage stayed at 100% nearly all morning:

Is there some reason why the Data IO percentage would stay near zero when the database is running at S3, which has half the DTUs of S4? The automated scaling happened on Thursday night/Friday morning as well, with the IO remaining near 0 when on S3 overnight and suddenly rising when scaled up again. We usually have less requests made to the site overnight, but its still a pretty constant flow of requests and shouldn't cause IO to drop so suddenly.

Its also worth noting that we have had issues with usage spiking before adding the scaling automation.

Thank you!