Inferring Injection Returns from Chloride Monitoring Data