Comparison of Automated and Manual Rain Gauges During Events with High Rainfall Rates and High Winds

Accurately measuring rainfall is critical for many applications including flash flood forecasting, agricultural interests, water resource management, fire weather, forestry, and more. Both rainfall totals and rainfall rates must be measured to effectively provide information for short-term decision-making like crop irrigation and issuing flash flood warnings and long-term planning like drought mitigation, controlled burns, and water resource preservation. Measuring rainfall is a difficult endeavor no matter how you do it. Satellite- and radar-based rainfall products provide a spatial distribution of rainfall, but the rainfall amounts are estimated from remote sensing data using algorithms. These algorithms are developed and validated using rain gauge measurements. Rain gauges measure rainfall directly, but only provide point observations. As a result, rainfall measurements used for forecasting, decision-making, and planning are often are a combination of remote sensing and ground-based observations. There are two ways of quantifying rainfall: amount (or accumulation) and intensity. Rain amount is reported as a depth in units of inches or millimeters and intensity is reported in units of amount per hour. Rainfall intensity is one of the largest contributors to error in measurement of precipitation amount.

High rainfall rates and strong winds are the main factors affecting the accuracy of rain gauge measurements. It has also been shown that rain gauges with narrow orifices perform worse during rainfall events with large droplets (e.g. thunderstorms) because more large droplets miss entering the gauge. Rain gauges perform poorly in high winds because raindrops fall towards the gauge opening at an angle, essentially reducing the size of the gauge opening and, hence, catching less rain. Furthermore, traditional cylindrical-shaped gauges create turbulence and wind shear just above the gauge causing raindrops to miss the gauge opening. With traditional cylindrical rain gauges, wind can reduce the rainfall catch by up to 20 percent. It is well-known that tipping bucket rain gauges under-collect during heavy rainfall intensities, because there is a limit to the speed at which the tipping-bucket mechanism can tip back and forth. During intense rainfall, extra water can accumulate in the bucket before the tipping mechanism tips. In other words, a bucket calibrated to collect 0.01 inch per tip will count only 0.01 inches per tip when in fact more rain fell. This leads to an under-collection. Given their importance, rain gauge measurements are continuously being improved by rain gauge shape design and the development of error correction algorithms based on numerical simulations and laboratory experiments.

To address both issues, Campbell Scientific Inc. offers the rainVUE Series. These tipping bucket rain gauges are equipped with a built-in rainfall rate correction algorithm. The correction algorithm was developed by running known rainfall rates through 10 new rainVUE tipping bucket rain gauges and using regression to find the best fitting mathematical function to relate time between tips to actual rainfall amount in the buckets. An electronic timer in the rainVUE gauges precisely measures the time between tips which is used as the predictor for the actual amount of rain per tip. Additionally, the rainVUE gauges have an aerodynamic hourglass shape to reduce the amount of rain that is lost during high winds.

In this study, the performance of various rain gauge types is compared at the South Alabama Mesonet weather station on the University of South Alabama campus in Mobile, Alabama. Two standard automated tipping bucket rain gauges are located at every South Alabama Mesonet site: a Texas Electronic TE-525 15.24 cm (6-inch) tipping bucket rain gauge and a Hydrological Services TB3 20 cm tipping bucket rain gauge with syphon to regulate water flow for improved performance during high-intensity rainfall events. Three additional gauges are installed at the campus site: a manual 10.16 cm (4-inch) CoCoRaHS gauge, a manual National Weather Service 20.32 cm (8-inch) Standard Rain Gauge, and a 20 cm Campbell Scientific RainVUE20 tipping bucket rain gauge.

Before commencing the experiment, all gauges were leveled and calibrated. The manual gauges are read once a day. All gauges are inspected after each rainfall event and the tipping bucket water content is recorded. The standard tipping bucket gauges record rainfall measurements every minute and the rainVUE gauge records measurements every 5 minutes. Gauges are cleaned regularly. The South Alabama Mesonet stations measure wind speed and direction at 2 and 10 m elevation. These measurements are used to determine if and how the performance of any of the gauges is affected by wind speed during a rain event. Furthermore, the RainVUE20 tipping bucket rain gauge is programmed to report maximum and average rainfall rates, allowing comparison of the gauges during high intensity rainfall rate events.

Preliminary results indicate that the TE-525 and the manual gauges collect more rainfall during higher rainfall total events (1 inch or more). The NWS gauge collects more rainfall than the CoCoRaHS gauge, possibly as a result of its wider orifice. The TE-525 collects more rainfall than the other automated gauges, especially during heavier rainfall rates. During long duration and heavy intensity rainfall events, this can lead to differences in rainfall totals of over half an inch. As expected, the algorithm in the rainVUE rain gauge increases rainfall amounts during high intensity rainfall events. To assess the impact of wind speed, raw (i.e. unadjusted) rainVUE 5-min rainfall totals are compared with TB3 5-min rainfall totals. Both gauges have the same orifice diameter, however not much difference can be seen; the largest differences (absolute values of 0.03 and 0.04 inches) occur at medium wind speeds of 2.5 to 4.8 m/s. At higher windspeeds (between 6.5 and 8.5 m/s) the absolute value of the difference between the 2 gauges is only 0.01 or 0 inches (i.e. 1 or no bucket tips). A sample size of only 622 datapoints was available at this writing, the sample size will be expanded by the time of the conference, hopefully providing more conclusive results.