Looks like you've used the "stacked=true" setting for a Metric Widget, hence you are plotting all metrics on the graph in a "stacked area" visualization. Only use this for metrics that add up logically.
In your case CPU is being added to the Memory below it, giving the 120 value. Remove the stacked setting, or set it to false (default). Also consider moving the Memory metric to the right Y-axis so the scale of each metric does not "drown out" the detail of the other.
Actually, CPUUtilization of ECS service can go over 100%. It's because this metric is not for CPU utilization of host machine, but rather for CPU utilization of CPU units allocated to your tasks. There are more details in AWS doc. Though for the chart itself - indeed there seems to be an issue somewhere as those figures do not match, but it's not that obvious which one is wrong.
CloudWatch CPUUsage does not match topasked 2 months ago
Scale errors on simple CloudWatch DashboardAccepted Answerasked 10 months ago
CloudWatch dashboards suddenly show no dataasked 2 years ago
CPU + memory usage missing from SM Studio notebook toolbarAccepted Answerasked 4 months ago
Kafka Monitoring Dashboard question - CPU Usage by Brokerasked 3 years ago
Filters for CloudWatch Dashboard widgetsasked 24 days ago
Elastic Beanstalk Memory usage monitoring with CloudWatch agentasked 2 months ago
CloudWatch DashboardAccepted Answerasked 2 years ago
HTTP API metrics not showing in CloudWatch automatic dashboardasked 8 months ago
tomcat8 jvm memory usage metricasked 6 months ago