Starting With The Basics
Water quality monitoring in the field is less about fancy sensors and more about keeping your equipment from lying to you. I spent seven years running stream surveys across the Pacific Northwest before moving into municipal systems. The biggest mistake I see is people buying expensive multiparameter sondes and then getting garbage data because they skipped the simple steps that actually matter. The core problem with field monitoring isn't measurement accuracy. It's preservation. You take a sample at 2 PM in July and it sits in a cooler until 8 AM the next day. By then your VOCs are gone, your ammonia has outgassed, and whatever bacteria you were looking for has either multiplied or died. I learned this the hard way when my team wasted three weeks chasing phantom E. coli positives that turned out to be lab contamination from improperly preserved samples. Sample preservation is the single most important step in field monitoring. Ice water at 4°C slows biological activity but doesn't stop it. For most parameters, you need chemical preservatives and strict chains of custody. The EPA methods give you exact prescriptions. Follow them. The shortcuts cost more in the long run.
Common Methods And What Actually Works
There are basically three approaches: grab sampling, discrete composite sampling, and continuous monitoring. Each has tradeoffs that manufacturers won't tell you about. Grab sampling works fine for compliance checks if you're careful about timing and preservation. The problem is spatial and temporal representativeness. One bottle from one depth at one time tells you almost nothing about what happens over a full hydrograph. I've seen people submit single grab samples as "monthly averages" and wonder why their data got rejected during audits. Composite sampling solves the temporal problem but introduces preservation headaches. If you're collecting a 24-hour composite and the preservative breaks down after 12 hours, your early samples are fine and your late samples are trash. You can't fix this in the lab. I use chilled glass amber bottles with HCl for metals and sodium thiosulfate for residual chlorine. The protocol takes an extra ten minutes per site but saves hours of arguing with regulators later.
Continuous monitoring is the third option and it introduces entirely new failure modes. I installed a YSI sonde in a river in 2018. After six months, the dissolved oxygen membrane had a biofilm that I couldn't see. The readings drifted by 2 mg/L from true values. The data looked beautiful. It was completely wrong. I only caught it when I ran a manual Winkler titration by accident during a site visit.
Get the Full Details
Calibration And Quality Control
Field calibration matters more than laboratory calibration. A sonde calibrated in clean lab water will give different readings in a real river with 200 NTU turbidity and 15 mg/L suspended solids. I recalibrate on site every time I deploy. It takes 15 minutes instead of 5. The 5-minute version produces data I can't defend in court. Quality control samples fall into three categories: field blanks, duplicates, and spike recoveries. Most monitors skip at least one. Field blanks catch contamination from your bottles and preservatives. Duplicates catch precision problems. Spike recoveries catch matrix interference. If you're doing EPA Method 160.1 for metals, you need all three plus a laboratory fortified blank every 20 samples. The blind spot in most quality control programs is the frequency. Sampling one day per month won't catch seasonal variations. I recommend weekly during snowmelt and monthly otherwise. The extra cost is about $200 per site per year in labor and supplies. The alternative is spending $20,000 defending bad data during a regulatory hearing.
Limitations And Where This Breaks Down
Field monitoring has real limitations that textbooks ignore. Optical sensors foul within days in nutrient-rich water. I replace flow cells every two weeks during algal blooms. The manufacturer says monthly. Their data comes from clean standard solutions, not real rivers. Turbidity interferes with UV-Vis absorbance measurements above 50 NTU. I dilute samples and correct mathematically. The dilution introduces pipetting error. I use Class A volumetric flasks and calibrated pipettes. The error is about 3% instead of 15%. The biggest limitation is cost. Proper field monitoring runs about $5,000 per site per year for personnel, supplies, and QA. Cheap programs that cut corners save money upfront and spend it back ten times over during remediation.
For high-frequency monitoring in sensitive watersheds, I recommend combining continuous sondes with weekly grab samples for validation. The sondes catch events. The grabs catch errors. Using both cuts uncertainty from 25% to about 8%, depending on your setup.
