Using an Ultrasonic Snow Sensor for the Remote Collection of Snowfall and Snowdepth Data for Operational and Climatological Purposes

Mark Turner [1]

Extended ABSTRACT:

The National Oceanic and Atmospheric Administration’s (NOAA) National Weather Service (NWS) Weather Forecast Office (WFO) at Caribou, Maine has weather forecast and warning responsibility for Northern Maine. One seasonal aspect of this responsibility is the accurate measurement of snow, both short term (24 hours, snowfall) and long term (24 hours, snowdepth) accumulation.

Historically, NOAA’s NWS has measured both snowfall and snowdepth synoptically (snowfall every 6 hours, snowdepth every 24 hours) at many WFO and Weather Service Office (WSO) locations across the country. This data was used for forecast, warning and climatological purposes.

Two events have complicated the NWS snowfall and snowdepth measuring program over the last decade, the contraction of NWS manned measuring sites associated with the recent NWS modernization, and the deployment of the Automated Surface Observing System (ASOS) data collection platform, which measures neither element.

The NWS office in Caribou, Maine obtained an ultrasonic device designed to measure the distance from the sensor orifice to any surface below that orifice. A test was conducted to determine if the data from the ultrasonic sensor could be used by the WFO to remotely determine snowdepth to the nearest inch and snowfall to the nearest 1/10th of an inch for use in operational and climatological aspects of the WFO mission.

This sensor was deployed at the home of a trained, volunteer, Cooperative (COOP) weather observer in Washburn, Maine. Data was collected from both sources for a period of 48 days, from February 19 to April 6, 2004. Data from the ultrasonic sensor was logged every minute throughout the test and internally converted from distance to depth. Manual measurements were taken every 24 hours at 0700 EST.

The ultrasonic sensor was located 30 meters distant from the manual snow measurement site. This location (Fig. 1) was a small clearing in a stand of hardwood trees. These trees acted as a natural barrier to wind blown snow drifts. For the purposes of this test, AC power was temporarily run to the sensor location to power the sensor and the data logger. The data was collected periodically using a serial port connection to a portable computer.

Figure 1: Ultrasonic Snow Sensor Deployed at Washburn Maine

The raw data from the ultrasonic sensor (Fig. 2) showed an unexpected oscillation. Further analysis of this data revealed the oscillations occurred at approximately the same time each day. This timing seemed to be in conjunction with diurnal temperature increases.

All Data Chart

Figure 2: Plot of Raw Data from the Ultrasonic Sensor

It was determined that the ultrasonic sensor data was corrupted from approximately sunrise to sunset (0700 to 1800 EST). This temperature affected data was removed from the test altogether. The data from the ultrasonic sensor at 0700 EST was then compared to the manual snow measurements taken by the COOP observer at 0700 EST to determine the feasibility of using the ultrasonic sensor to determine snowdepth (Fig. 3).

Snow Depth 7AM

Figure 3: Plot of Manual vs. Ultrasonic Snowdepth Measurements

Snowfall data was derived from the ultrasonic sensor data using the following formula;

If SD^t-24 >SD^t-0 then SF=0, if SD^t-24 < SD^t-0 then SF=SD^t-0-SD^t-24

Where SDt-24 is the snowdepth 24 hours before the current 0700 measurement and SDt-0 is snowdepth at 0700 EST (Fig. 4).

24 Hour Snowfall

Figure 4: Plot of Derived Ultrasonic vs. Manual 24 hour snowfall measurements

The test suggests that snowdepth can be determined using an ultrasonic sensor; the trend of the 24 hour point measurements of snowdepth logged by the ultrasonic sensor correlate closely with the trend of the 24 hour point measurements recorded by the COOP observer.

The derived 24 hour snowfall from the ultrasonic sensor also shows promise. Out of 11 snow events recorded by the COOP observer during the test, 9 events were also recorded by the ultrasonic sensor. Snowfall measurements by the ultrasonic sensor during 6 of the events were within 0.1 inches of that measured by the COOP observer.

Unfortunately, the goal of using the ultrasonic sensor operationally to determine snowfall to the nearest 1/10th of an inch remains elusive. When 24 hour measurements were divided synoptically to more closely match the 6 hourly NWS snowfall reporting criteria (minus the diurnal data oscillations previously mentioned), many false events were recorded (Fig. 5). The ultrasonic sensor recorded 41 events out of 45 days.

Snowfall 2

Figure 5: Plot of derived 24 hour ultrasonic sensor data (summed 6 hourly amounts) vs. manual 24 hour snowfall measurements

Further analysis of this data suggested that the ultrasonic sensor, as deployed, was at the limit of its accuracy/resolution ratio. The accuracy of the sensor is 0.5% of distance measured, and the resolution of the measurement is 0.1 inches. Therefore, the sensor accuracy was 0.4 inches, rather than the 0.1 inches required by the NWS.

This ultrasonic sensor is scheduled to undergo further testing by WFO Caribou during the winter of 2004-2005.

Keywords: snowdepth, snowfall, ultrasonic

Weather Events Home

Maine-ly Weather Home

[1] NOAA/National Weather Service, WFO Caribou, 810 Main Street, Caribou, Maine, 04736