STANDARDS AND SPECIFICATIONS DOCUMENT DATA COLLECTION AND ACCURACY RANDALL JULANDER 6/98 7.1 Meteor burst system General overview In the meteor burst system, data are measured at remote sites and transmitted via radio telemetry to a master station. The term ‘Meteor Burst’ is derived by the specific way radio transmissions are used by this system. Radio transmissions are directed into the atmosphere and bounced off ionized meteor trails where they are again received on the ground. This system allows the circumvention of conventional line of sight radio transmission. From the master station, data are relayed via ground lines to processing and computing facilities. A standard Meteor Burst SNOTEL remote site consists of measuring devices and sensors, a shelter for the radio telemetry equipment, an antenna tower that supports the antenna and the solar panels used to keep batteries charged and a meteorological tower for various sensors. A standard sensor configuration includes the following measurements: snow water equivalent (snow pillow(s)or other sensor- gamma, strain, etc.), precipitation ( a storage precipitation gage), snow depth (sonic depth gage) and air temperature, maximum, minimum, current and average (thermistor). A standard Soil-Climate Analysis Network (SCAN)sensor configuration includes SWE, precipitation, air temperature, etc. Radio/data logger-processor: the current radio standard is a meteor burst transceiver for SNOTEL and for SCAN it is coupled with a data logger for SCAN. This radio is a transceiver/receiver and it performs the role of data logger and data processor as well. Data are sampled from the various sensors by the radio logger function, calculations performed and results stored until transmission or cycle times have been met. In future versions, the data logging/processing functions will most likely be separate from the receiver/tranceiver portion of the electronic components. In fact, there may be multiple forms of data transmission devices depending on factors such as dependability, vulnerability, economics, versatility, etc. The snow pillows are envelopes of stainless steel or synthetic rubber/hypalon, about 80 square feet, containing a non-freezing solution. As snow accumulates on the pillows, it exerts pressure on the solution forcing the fluid out of the pillow through the connecting lines into the shelter. A pressure transducer is used in the shelter to convert the weight of snow into an electrical reading of the snow's water equivalent--that is, the actual amount of water in a given volume of snow. The storage precipitation gage measures all precipitation in any form that falls during the year. The gage is 12 inches in diameter and between 8 and 35 feet high with either an Alter or Wyoming wind shield. Similar to the pillow sensor, the precipitation gage is plumbed or wired to the instrument shelter and sensed by a pressure transducer. The temperature sensor is a thermistor that senses the difference in electrical conductivity which is directly related to temperature. Algorithms in the data collection/transmission equipment calculate the average, maximum and minimum temperature parameters. Thermistors are used to determine the temperature of many different mediums such as: air, fluid, soil, snow, etc. Depth sensors, typically sonic, measure the distance of the snow surface to the depth sensor. The actual snow depth is then calculated. Additional sensors can be incorporated into a particular site for measuring wind speed and direction, soil moisture and temperature, and a variety of other weather and environmental aspects. The configuration at each site is tailored to meet the purpose of the site as well as any physical conditions, the climate and the specific requirements of the data users. REMOTE SITE PERFORMANCE Standards for remote site performance are set to insure consistency, compatibility and uniformity of data collection across broad geographic areas. These standards should minimize the bias that can be introduced into data sets through human error, subjectivity and technique. By using standard components, installation methods and data collection techniques, data sets become more consistent between areas of jurisdiction. 1. SWE pillow sensor function: snow pillows, both steel and hypalon/butyl are designed as a compression device. They are filled with a non freezing solution such as 50/50 mix of ethanol and water, Propylene Glycol-ethanol and water. The pillow is plumbed to a manometer and a pressure transducer, typically located in the instrument shelter. As weight (snow water equivalent) accumulates on the pillow, this pressure is sensed by a pressure transducer (analog: 0-5 volts) and the data subsequently stored in the data logger portion of the site electronics. Transducers Data collected from snow pillows is first sensed as static head or pressure by the pressure transducer. It is expressed as a linerar analog voltage ranging from 0 to 5 volts and recorded as millivolts. This is subsequently converted to inches of snow water equivalent at the Central Computer Facility (CCF). All transducers used in the system have the same output, 0 to 5 volts but can have different ranges depending on the specific site. The most commonly used transducers are the 50, 100, 150 and 200 inch range. The conversion of a standard 1 millivolt output signal to snow water equivalent is proportional to the range of transducer used. In order to increase data resolution, the correct transducer range should be used at each site. Select the appropriate transducer range that given the maximum snow water equivalent at that site, would not be exceeded. 1. Data units: inches of snow water equivalent 2. Currently, data resolution is 0.1 inch. While measurements can be made at finer scales (0.01 inch), given the uncertainties in snowpack measurement, site installation, climatic affects (barometric and temperature) on sensors and other unquantifiable vagaries such as sensor instability, etc., 0.1 inch is very optimistic. Given all these error potentials, sensor flutter of plus/minus 0.2 inch is acceptable between midnight to midnight. A maximum acceptable flutter of plus/minus 0.5 inches during periods of snow cover. Seasonal variations may also induce sensor instability such as thermal expansion and contraction at various times of the year. 3. The collection frequency or interval standard is one reading per day, at midnight Pacific standard time. Alaska SNOTEL data site collection frequency is one reading per day, at midnight Alaska standard time. Data collection frequency can be as fine as 1 hour depending on need. All stations will record a minimum of 1 snow water equivalent reading per day 3b. The collection frequency for SCAN sites are hourly. 4. Ground truthing is a manual check to see if the sensor is providing accurate information. The first check is the manometer, does the manometer confirm or reject what the sensor is reading? Second, manual snow samples taken at the sample markers around the pillow (see the "snow survey sampling guide" for manual sampling procedures). Observe the historical correlation between the ground truth samples and the pillow values, does the current value fit? The manually measured value should be within 10% of the pillow value. Third, confirm the snow water equivalent with the precipitation gage. Snow water equivalent and precipitation are certainly different, but related parameters. The data traces from these two properly performing sensors should show symmetry of pattern and amount up to the ablation period. Yet another form of ground truth is observed at meltout. Did the sensor return to the zero level (the same level it was at previous to snowpack accumulation) after the end of the ablation period? As with other ground truth, this is only an index and other factors may play a critical role. The ground surface may move up to an +/- one inch or more due to frost heave and other expansion/contraction phenomena. However, any sensor that does not come back to within +/- 0.5 inch of the initial setting should be checked. All transducers can be checked with a transducer checker or multimeter for their zero setting and span. 5. Sensor performance: the ‘sensor’ for this section will be defined as everything between the pillow and the radio transceiver. Sensor problems can be categorized as 1) hard fail, 2) gross over/under weigh and 3) instability, 4) consistent decrease, 5) consistent over increase. A hard fail indicates a flat pillow or failed transducer. Gross overweigh can mean an ice problem, drifting, snow creep, tree falling on the site, etc. Gross under-weight can mean a leaking pillow, a failing transducer, ice bridging, etc. The most common cause of instability is air bubbles in the SWE (pillow) plumbing. A consistent decrease outside of the ablation period is a sign of a leaking pillow system. A consistent increase over and above the precipitation gage may indicate snow creep, or ice layers on the pillow which leverage more weight that is directly on the pillow itself. In all these cases and potentially others, sensor performance is affected and data quality is impaired. Pillow transducers should theoretically be able to maintain a +/- 25 millivolt tolerance. In field applications there are many more uncontrolled factors that could affect this tolerance. Differential ground surface movement between the transducer and the pillow due to frost heave, soil expansion or contraction due to excess moisture or extremely dry conditions can cause the tolerance to be exceeded without any fault on the part of the sensor. Air bubbles in the plumbing line can compress/expand with variances in pressure, temperature and barometric pressure. Diurnal fluctuations in pillow sensors are common, especially under no snow conditions. These fluctuations are caused by the daily heating and cooling of the system, causing pressure and/or expansion/contraction differences within the system and subsequently measured by the transducer. For comparative purposes, readings of similar circumstances (time, temperature, etc.) should be used to determine overall sensor stability. Most sensor problems will be identified through data analysis and then confirmed by field examination. 2. Snow Depth 1. Sonic Sensor function: Sonic snow depth sensors measure the time between the emission of an ultrasonic pulse and its return to the receiver. The ultrasonic pulse travels at the speed of sound and is dramatically affected by temperature, thus the need for a temperature compensating algorithm in the data processing function. The pulse is emitted from the sensor toward the snowpack where it is reflected back to the sensor receiver. The elapsed time multiplied by the speed of sound, compensated for air temperature and divided by 2 (travel to the snowpack and return) yields the distance from the sensor to the snowpack. This distance, subtracted from the sensor height above the ground surface then gives the depth of snow at that point. Since this sensor measures the fastest return of the ultrasonic pulse, it will consistently measure the highest point of the surface within its cone or target area. It is crucial to have this target area as flat as possible with no obstructions or protrusions. The cone of vision for this sensor is approximately 22 degrees, .38 times the height of the sensor to the target. Since the distance being measured is typically very small and small deviations in temperature can account for substantial errors in the measurement, the temperature used for compensation should be incorporated in the depth sensor itself. Since this technology is susceptible to a myriad of potential errors in measurement, a redundant sampling algorithm is essential. The sensor should sample up to a dozen times, comparing the current sample with the one immediately previous, until 2 samples within 1/2 inch of each other are obtained. This helps to minimize data errors during heavy snowfall when false echoes are returned from falling snow or when echoes are absorbed into very light density snowfall. The initial measurement from this sensor is time and temperature. Distance to the snowpack surface is then calculated via a calculation of the speed of sound at the given temperature. Finally, the snow depth is calculated by taking the current distance measurement and subtracting it from the total height of the sensor above the ground surface. Data units: The units are inches of snow depth. Resolution: The resolution capability of this sensor is 0.1 inch, however given the anomalous surface characteristics of a snowpack over an area of several square yards at this resolution, the effective resolution is 0.5 inch. Collection frequency: The collection frequency or interval standard is one reading per day, at midnight local standard time. Data collection frequency can be as fine as 1 hour depending on need. All stations with a snow depth senor will record a minimum of 1 snow depth reading per day Ground Truth: is a manual check to verify sensor accuracy. The ground truth on a depth sensor can be performed in 2 ways: 1) measure depth of snow at the highest point of the pack on the sensor target area or at the pillow ground truth markers and 2) measure the distance from the snowpack surface up to the sensor. Great care should be taken not to disturb the snow surface under the sensor on the target area as this could affect future readings. Since the standard installation area for these sensors is over the snow pillow and measuring directly over the pillow with a snow tube is forbidden, measuring from the snow surface up to the sensor is preferred. Measuring depth at the ground truth pillow markers is an acceptable location although it could compromise the accuracy of the ground truth measure. Variations of much more than 0.5 inch is common from marker to marker. Sensor Performance: If more than 10% of readings are unresolvable, (return negative indicating that the sensor was unable to return 2 readings within 1/2 inch) it is likely that the sensor needs repair. Possible causes: bad sounding board, bad pulse generator, poor mounting, irregular surface such as weeds growing up, intermitent snow cover. 3. Precipitation sensor 1. Storage gage Sensor Function: this gage catches precipitation in a tube, 12 inches in diameter and of varying heights. It is plumbed or wired to the instrument shelter where there is a manometer and a pressure transducer. The weight of the accumulating precipitation is measured by the pressure transducer. The precipitation gage is recharged with a non-freezing solution. A small amount of lightweight oil is added to create a surface barrier to evaporation. These gages are at least 3 feet taller than the highest expected snowpack and the maximum expected precipitation at each site. Each has an Alter or Wyoming Windshield to mitigate the affects of wind over the catch area of the gage. Transducers Data collected from storage precipitation gages is first sensed as static head or pressure by the pressure transducer. It is expressed as an analog electrical current ranging from 0 to 5 volts and recorded as millivolts. This is subsequently converted to inches of precipitation. All transducers used in the system have the same output, 0 to 5 volts but can have different ranges depending on the specific site. The most commonly used transducers are the 50, 100, 150 and 200 inch range. The conversion of a standard 1 millivolt output signal to snow water equivalent is proportional to the range of transducer used. In order to increase data resolution, the correct transducer range should be used at each site. Select the appropriate range of transducer that given the maximum snow water equivalent at that site, would not be exceeded. Data units: inches of precipitation Resolution: Currently, data resolution is 0.1 inch. Measurements can be made at finer scales (0.01 inch) using temperature compensating algorithms that account for expansion and other variables. These are not currently utilized, but may be in future years. There are other error sources such as: air in the gage system, climatic affects (barometric and temperature) on sensors and other unquantifiable vagaries such as sensor/transducer instability, etc., 0.1 inch is very optimistic. Given all these error potentials, sensor flutter of plus/minus 0.2 inch is acceptable. Seasonal variations may also induce sensor instability such as thermal expansion and contraction at various times of the year. Collection Frequency: The collection frequency or interval standard is one reading per day, at midnight Pacific standard time. Data collection frequency can be as fine as 1 hour depending on need. All stations will record a minimum of 1 precipitation reading per day Ground truthing is a manual check to see if the sensor is providing accurate information. The first check is the manometer, does the manometer confirm or reject what the sensor is reading? Second, a drastic measure to be used only if the manometer is deemed non-functional, is to drain the gage bucket by bucket, weighing each to derive the total gage contents. Sensor performance: the ‘sensor’ for this section will be defined as everything between the precipitation gage and the radio transceiver. Sensor performance can be categorized as 1) hard fail, 2) gross over/under weigh and 3) instability, 4) consistent decrease. A hard fail indicates a drained gage or failed transducer. Gross overweigh can mean an ice problem etc. Gross under-weight can mean a leaking gage, a failing transducer, snow capped gage, lack of oil to prevent evaporation, etc. The most common cause of instability is air bubbles in the gage plumbing. Another cause is birds, rodents and detritus falling in the gage and causing plugs in the plumbing line. The ‘j’ tube prevents some of this (this tube needs a small hole in the top of the ‘j’ to make sure air bubbles do not affect the readings). A consistent decrease is a sign of a leaking gage system. In all these cases and potentially others, sensor performance is affected and data quality is impaired. All transducer systems should be able to maintain a +/- 25 millivolt tolerance. In field applications there are many more uncontrolled factors that could affect this tolerance. Differential ground surface movement between the transducer and the pillow due to frost heave, soil expansion or contraction due to excess moisture or extremely dry conditions can cause the tolerance to be exceeded without any fault on the part of the sensor. Air bubbles in the plumbing line can compress/expand with variances in pressure, temperature and barometric pressure. Diurnal fluctuations in both pillow and storage gage sensors are common. These fluctuations are caused by the daily heating and cooling of the system, causing pressure and/or expansion/contraction differences within the system and subsequently measured by the transducer. For comparative purposes, readings of similar circumstances (time, temperature, etc.) should be used to determine overall sensor stability. 2. Tipping Bucket Rain Gage place holder 3. Weighing Rain Gage place holder 4. Optical Rain Gage place holder 4. Temperature sensor 1. Air Temperature 1. Sensor Function: The current standard is a thermistor temperature probe. Thermistors are formulations of powdered, compressed metal oxides that are then sintered. Discs are calibrated to display nearly identical resistance at equal temperatures. The variable being sensed in this case is electrical resistance. This resistance changes with temperature fluctuations. Temperature can then be calculated using standard equations provided by the thermistor manufacturer. 2. Data units are degrees Celsius. 3. The resolution of a thermistor should be +/- 0.2 degrees C. The operational range of the thermistor should allow for extremes on both ends of the scale: every thermistor should be stable over the range of -22 to 38 degrees C (-30 to 100 degrees F). Some areas that have the potential for colder temperatures will need a thermistor capable of a much lower bottom range, potentially as low as -50 degrees C. current specifications from YSI, a thermistor manufacturer give the following characteristics: recommended operating range (-100 to 120 degrees C), Stability: +/-0.2 degrees C or better for 10 months at 100 degrees C, Tolerances: +/- 0.2 degrees C 4. Collection frequency for a standard measure is: maximum temperature - once per measurement cycle of 24 hours, midnight to midnight, minimum temperature - once per measurement cycle of 24 hours, midnight to midnight, average - is a calculated value (the sum of all measurements divided by the total number of measurements). The current temperature is that value at the time of the most recent update. Updates may be take hourly. 5. Ground truth must be measured by a calibrated thermometer, placed as close to the thermistor as possible. This measurement must be taken inside the aspirator to avoid the possible affects of direct sunlight on the thermometer. Another way is to install a second (previously bench tested) thermistor inside the same aspirator. In most cases, if a thermistor is suspect, it is easier and far cheaper to simply replace the suspect unit with a new one. Economics is a big factor here, the testing of a thermistor will generally cost more (time, equipment, travel, etc.) than its replacement (about $18, 1998). 6. Sensor performance. The sensor in this case is the thermistor. Since temperature can vary dramatically over short time periods, the thermistor must be capable of responding quickly. Minimum temperatures can not be higher than maximum readings. The maximum must be equal to or greater than any previous reading of the current cycle. Flat line readings, erratic or wildly fluctuating data or other instability indicate a failed sensor. With only extremely rare exceptions, temperatures are cooler in the night and warmer in the day and the thermistor should reflect that pattern. 2. Soil Temperature 1. Sensor Function: The current standard is a thermistor temperature probe and will most likely be combined with some other function such as soil moisture in a single unit. Thermistors are formulations of powdered, compressed metal oxides that are then sintered. Discs are calibrated to display nearly identical resistance at equal temperatures. The variable being sensed in this case is electrical resistance. This resistance changes with temperature fluctuations. Temperature can then be calculated using standard equations provided by the thermistor manufacturer. 2. Data units are degrees Celsius. 3. The resolution of a thermistor should be +/- 0.2 degrees C. The operational range of the thermistor should allow for extremes on both ends of the scale: every thermistor should be stable over the range of -22 to 38 degrees C (-30 to 100 degrees F). some areas that have the potential for colder temperatures will need a thermistor capable of a much lower bottom range, potentially as low as -50 degrees C. current specifications from YSI, a thermistor manufacturer give the following characteristics: recommended operating range (-100 to 120 degrees C), Stability: +/-0.2 degrees C or better for 10 months at 100 degrees C, Tolerances: +/- 0.2 degrees C 4. Collection frequency for a standard measure is: maximum temperature - once per measurement cycle of 24 hours, midnight to midnight, minimum temperature - once per measurement cycle of 24 hours, midnight to midnight, average - is a calculated value (the sum of all measurements divided by the total number of measurements). The current temperature is that value at the time of the most recent poll. Polls may be take hourly. A standard poll is one reading per day. 5. Ground truth must be measured by a calibrated thermometer, placed as close to the thermistor as possible. Since excavation is required, the actual reading of soil temperature must be done quickly as ambient air temperature could affect the soil temperature. Another way is to install a second (previously bench tested) thermistor next to the suspect thermistor. In most cases, if a thermistor is suspect, it is easier and far cheaper to simply replace the suspect unit with a new one. Economics is a big factor here, the testing of a thermistor will generally cost more (time, equipment, travel, etc.) than its replacement (about $18, 1998). Even when the sensor is a malfunctioning unit (soil temperature and moisture), it will be less expensive to replace than to test. Thermistors have been extremely reliable over many years of service and are the least likely meteor burst system component to fail. If a thermistor displays failed symptoms, the recommended procedure is simply replacement without ground truthing the failed component. 6. Sensor performance. The sensor in this case is the thermistor. Since temperature can vary dramatically over short time periods, the thermistor must be capable of responding quickly. Minimum temperatures can not be higher than maximum readings. The maximum must be equal to or greater than any previous reading of the current cycle. Flat line readings, erratic or wildly fluctuating data or other instability indicate a failed sensor. With only extremely rare exceptions, near surface soil temperatures are cooler in the night and warmer in the day and the thermistor should reflect that pattern. The deeper the soil temperature sensor is placed within the soil, the more stable temperatures will become. Diurnal fluctuations will dampen and potentially disappear, showing only seasonal or long term variability. Deep sensors may show only a degree or two of variation over a years time. These sensors must be extremely sensitive and stable. 3. Snow Temperature 1. Sensor Function: The current standard is a thermistor temperature probe. Thermistors are formulations of powdered, compressed metal oxides that are then sintered. Discs are calibrated to display nearly identical resistance at equal temperatures. The variable being sensed in this case is electrical resistance. This resistance changes with temperature fluctuations. Temperature can then be calculated using standard equations provided by the thermistor manufacturer. 2. Data units are degrees Celsius. 3. The resolution of a thermistor should be +/- 0.2 degrees C. The operational range of the thermistor should allow for extremes on both ends of the scale: every thermistor should be stable over the range of -22 to 38 degrees C (-30 to 100 degrees F). some areas that have the potential for colder temperatures will need a thermistor capable of a much lower bottom range, potentially as low as -50 degrees C. current specifications from YSI, a thermistor manufacturer give the following characteristics: recommended operating range (-100 to 120 degrees C), Stability: +/-0.2 degrees C or better for 10 months at 100 degrees C, Tolerances: +/- 0.2 degrees C 4. Collection frequency for a standard measure is: maximum temperature - once per measurement cycle of 24 hours, midnight to midnight, minimum temperature - once per measurement cycle of 24 hours, midnight to midnight, average - is a calculated value (the sum of all measurements divided by the total number of measurements). The current temperature is that value at the time of the most recent poll. Polls may be take hourly. A standard poll is one reading per day. 5. Ground truth must be measured by a calibrated thermometer, placed as close to the thermistor as possible. Another way is to install a second (previously bench tested) thermistor close to the primary sensor. In most cases, if a thermistor is suspect, it is easier and far cheaper to simply replace the suspect unit with a new one. Economics is a big factor here, the testing of a thermistor will generally cost more (time, equipment, travel, etc.) than its replacement (about $18, 1998). Thermistors have been extremely reliable over many years of service and are the least likely meteor burst system component to fail. If a thermistor displays failed symptoms, the recommended procedure is simply replacement without ground truthing the failed component. 6. Sensor performance. The sensor in this case is the thermistor. In this application, as with soil temperature, very little fluctuation will be observed once a substantial snowpack has accumulated. The thermistor must be extremely stable for long periods where temperatures may change only a few degrees. Minimum temperatures can not be higher than maximum readings. The maximum must be equal to or greater than any previous reading of the current cycle. Flat line readings, erratic or wildly fluctuating data or other instability indicate a failed sensor. 4. Fluid Temperature 1. Sensor Function: The current standard is a thermistor temperature probe. Thermistors are formulations of powdered, compressed metal oxides that are then sintered. Discs are calibrated to display nearly identical resistance at equal temperatures. The variable being sensed in this case is electrical resistance. This resistance changes with temperature fluctuations. Temperature can then be calculated using standard equations provided by the thermistor manufacturer. 2. Data units are degrees Celsius. 3. The resolution of a thermistor should be +/- 0.2 degrees C. The operational range of the thermistor should allow for extremes on both ends of the scale: every thermistor should be stable over the range of -22 to 38 degrees C (-30 to 100 degrees F). some areas that have the potential for colder temperatures will need a thermistor capable of a much lower bottom range, potentially as low as -50 degrees C. current specifications from YSI, a thermistor manufacturer give the following characteristics: recommended operating range (-100 to 120 degrees C), Stability: +/-0.2 degrees C or better for 10 months at 100 degrees C, Tolerances: +/- 0.2 degrees C 4. Collection frequency for a standard measure is: maximum temperature - once per measurement cycle of 24 hours, midnight to midnight, minimum temperature - once per measurement cycle of 24 hours, midnight to midnight, average - is a calculated value (the sum of all measurements divided by the total number of measurements). The current temperature is that value at the time of the most recent poll. Polls may be take hourly. A standard poll is one reading per day. 5. Ground truth must be measured by a calibrated thermometer, placed as close to the thermistor as possible. Another way is to install a second (previously bench tested) thermistor close to the primary sensor. In most cases, if a thermistor is suspect, it is easier and far cheaper to simply replace the suspect unit with a new one. Economics is a big factor here, the testing of a thermistor will generally cost more (time, equipment, travel, etc.) than its replacement (about $18, 1998). Thermistors have been extremely reliable over many years of service and are the least likely meteor burst system component to fail. If a thermistor displays failed symptoms, the recommended procedure is simply replacement without ground truthing the failed component. 6. Sensor performance. The sensor in this case is the thermistor. In this application, there is greater fluctuation than with snow or soil, but considerably less than with air temperature. Minimum temperatures can not be higher than maximum readings. The maximum must be equal to or greater than any previous reading of the current cycle. Flat line readings, erratic or wildly fluctuating data or other instability indicate a failed sensor. 5. Wind sensors 1. Direction 1. sensor function: this sensor reports the direction from which the current wind is blowing. Over time these measurements form a rosette, typically of both direction and magnitude. A typical rosette has 8 directions (n, ne, e, se, s, sw, w, nw). Distance from the center indicates velocity, radial width indicates the relative proportion of time spent at a specific velocity band. Thus most appear as telescopes with the small end far from the center and the large part at the center. In a data base, this information is stored as the amount of time spent at a specific direction. 2. data units: compass degrees, 0 to 360 3. resolution: +/- 3 degrees 4. collection frequency: wind direction is essentially an instantaneous measure or a measurement over a specified period of time ( an average of instantaneous readings). Thus, frequency is more a matter of this summarization than of knowing what direction the wind is blowing at any current moment. Wind samples are taken every 1 minute is a sufficient indicator of wind direction. Data reported once per day. 5. ground truth: This sensor can be checked on site to the nearest 10 degrees with a hand compass, (offset by the appropriate declination). More sophisticated directional/engineering equipment can be utilized to ascertain direction to the nearest degree such as: Global positioning systems, etc. 6. sensor performance: This sensor can be affected by ice and snow buildup on the sensor, blocking the impeller and the rotational capability of the sensor. 2. Speed 1. sensor function: determine both instantaneous wind speed and cumulative wind run. The propeller type units produce an AC sine wave voltage or pulse count which is directly proportional to wind velocity. 2. data units: wind speed in miles per hour or meters per second. 3. resolution: +/- 1.1 mph 4. collection frequency: wind speed is always instantaneous, wind run is the cumulative distance over a specified period of time, in this case, 24 hours. typically instantaneous values samples once per minute. Maximum, minimum and average wind speeds can be obtained from the cumulative daily record. 5. ground truth: gross errors may be detected by hand held wind devices. these can be very inaccurate and their use should be restricted to only the quantification of a near terminal sensor. the only other way is to mount a second, bench tested sensor next to the suspect one. 6. Solar Radiation 1. Sensor Function: to measure various forms of solar radiation. there are various terms used to describe solar energy such as long wave, short wave, total, diffuse, reflected, or specific wavebands designations such as photosynthetically active, etc. The solar sensor typically uses a silicon photovoltaic detector. For use with the meteor burst system of 0-5 volts, typically a amplifier module is required to boost the voltage output from the sensor. 2. data units: watts per square meter 3. resolution: +/- 5%, 80 micro amps per 1000 watts/meter square. 4. collection frequency: solar radiation is typically collected as a daily summation with a minimum, maximum and average value. 5. ground truth: secondary calibrated unit comparison or calibrate with a Eppley Precision Spectral Pyranometer. A calibrated unit will replace the existing unit every 3 years as a minimum. 6. sensor performance: not a clue - if it be broke denn fixit. 7. Relative Humidity 1. sensor function: to determine relative humidity via several possible techniques such as a capacitor with dielectric hygroscopic film, hygroscopic membranes with porous electrodes, etc. Each needs a temperature sensor, typically a thermistor. As humidity changes, the permitivity of the membrane, film, etc. changes and this is converted to a voltage output. In the case of a capacitance type sensor, as humidity increases, so does dielectric constant which is directly related to the capacitance. 2. data units: voltage converted to % relative humidity 3. resolution: +/- 2% over the entire range 4. collection frequency: as needed, sampled frequently to determine maximum, minimum and average for the daily record. 5. ground truth: horsehair hygrometer or an electronic device used to measure RH. 6. sensor performance: 8. Barometric Pressure 1. place holder 9. Soil Moisture 1. Capacitance type sensor 1. Sensor Function: this type sensor measures electrical capacitance and the conductive properties of a soil medium which are directly related to soil moisture and salinity. Additionally, soil temperature is measured as well. 2. Data units: % of water by volume and degrees Celsius 3. resolution: 0.03 (fraction water by volume) +/- 0.6 degrees C 4. Frequency: the collection frequency or interval standard is one reading per day, at midnight local standard time. Data collection frequency can be as fine as 1 hour depending on need. All stations will record a minimum of 1 soil moisture/soil temperature reading per day. 5. Ground truth: these units may be verified by doing gravimetric, Portable neutron probe samples (must have certified/trained personnel), or secondary sensors. 6. Sensor Performance: Because of the difficulty of installation and ground truth, these sensors must be capable of years of consistent performance. Precipitation events recorded at the ground surface should, within reasonable time, be sensed by the soil moisture sensor. During the spring, most shallow soils should be saturated by snowmelt infiltration, showing a drying cycle later in the spring and summer. Remote Site Operational Performance Standards/Procedures In this section, standards are set to insure prompt repair and maintenance of impaired or deficient sites. there are many circumstances which will impact the ability to service a given site with a problem. Any site problem that cannot be addressed within the given times and standards should have the extenuating circumstances and rational documented. 1. Hard Fail - this condition is where the site no longer responds. a. The site shall be repaired within 14 days of the failed condition. 2. Pillow sensor failure a. flat pillow - sensor should be repaired as soon as practical. Sensor replacement in winter should be avoided. 3. Precipitation gage failure a. Gage should be repaired/replaced as soon as practical. Sensor replacement in winter should be avoided. 4. Transducer failure a. transducers should be replaced within 14 days of the time the transducer is determined to be faulty. The overall safety of the individual doing the repairs is the first and formost consideration. 5. Temperature Sensor failure a. temperature sensors should be replaced within 14 days of the time the sensor is determined to be faulty. The overall safety of the individual doing the repairs is the first and formost consideration. 6. Depth Sensor failure a. Depth sensors should be replaced within 14 days of the time the sensor is determined to be faulty. The overall safety of the individual doing the repairs is the first and formost consideration. 7. Battery Failure / power supply a. the battery shall be replaced before data quality is affected. This could be a lengthy time depending on the site, number of sensors and transmission frequency. 8. General a. all above ground sensors mounted on a tower as well as any type of failure within the shelter should be repaired within 14 days of failure determination. Any surface or below surface sensor will be replaced/repaired as soon as practical after the snowpack melts and reasonable access is available. Every effort shall be made to insure the entire system is 100% going into the snow accumulation season. The overall safety of the individual doing the repairs is the first and formost consideration. MANUAL DATA COLLECTION Philosophy: Manual data collection has been done for centuries and has provided reliable long term records. It tends to be labor intensive and provides the lowest amount of data per unit resource expended. With advent of relatively low cost automation, the amount and types of data collected can be enormous such that the cost per datum is extremely low. To be as efficient a program as possible, it is to our advantage to convert every manual site possible to full automation. This applies to snow course, rain gages, aerial markers, etc. If data from the site are important, at some point, automation should be considered. Standards: Standards for manual data collection are set to insure data consistency, compatibility and uniformity in space and time. In manual data collection, the vagaries of human technique can introduce significant bias into data sets. Standards and periodic personnel training help minimize those biases. For example, there has been much research done on the Federal Snow Sampler. Through various means, it has been shown that this sampler "over samples" snowpack from 3% to 12% when compared to "ground truth" taken at a much larger scale. This is a "consistent" bias over the entire sampling regime, both in time and space and hence of little consequence when data are used as point index or if related both through modeling or statistically to other variables as typically is the case. Arguably, it may not be a bias at all, simply a scale phenomenon. However, if a station has 60 years of record taken with a federal snow sampler and then switches to a McCall or other type sampler, that data set is no longer consistent and a bias has been interjected. The same sampler that has historically been used to measure a site should continue to be used. In fact, it is preferable to have the same individuals measure a site for a long as possible as this introduces less error and potentially a "more consistent and more recognizable" error than having multiple individuals at random times. Human bias is introduced by using non standardized or poor sampling techniques. Such common errors include: not having the sampler vertical, not obtaining a proper core, walking on the sample points, discarding the earth plug too close to the sample point, contaminating subsequent readings and a whole host of other possibilities. These biases can be random or systematic and may be impossible to eradicate from the dataset. It is of utmost importance for each surveyor to strictly adhere to the measurement standards and methods. This insures the best possible dataset in space and time. 1. Snow Water Equivalent 1. SWE: how to sample Refer to the "Snow Survey Sampling Guide", Soil Conservation Service (currently the Natural Resources Conservation Service) Agriculture Handbook Number 169. 2. Sensor Function: The sensor in this case, is a federal snow sampler, or in the case of extremely deep snowpacks, the McCall sampler. It is essentially a tube with which an individual may take s snowpack sample at a specific location and point in time. The sampling technique is destructive and non-replaceable, once sampled, that same point can’t be sampled again. A sample very close to the original may be taken. Bias can be introduced into the samples of this sensor by the user. It is important that each sampler be rigorously trained in the science of snow measurement so that each snow course is measured consistently in time and space, regardless of the human component of the sensor. 3. Data units: inches of snow water equivalent 4. Resolution: 0.5 inches 5. collection frequency: The collection frequency is determined by data need. The most typical situation is to manually sample snow courses as near the beginning of the months of: March, April and May as possible. April is typically the peak snowpack month and sampling one month prior and post April covers the potential of both early and late snowpack anomalies. Historically snowpacks have been measured beginning in January and ending in June. With the advent of telemetered data, this intensive measurement scheme was unnecessary. For specific cases where telemetered data are not available and data needs dictate, this type of measurement schedule might be applicable. Special or unusual circumstances may dictate the need for non standard sampling intervals such as extremely high or late melting snowpacks. 6. Ground truth: Manual sampling is traditionally the ground truth for other sensors as it physically takes a core of snow to weigh. There is no independent corroboration for this sample because it is a destructive, non-replicatable sample, however there are related ways that indicate whether a sample is correct. All sample densities along a course being measured should be within a 5% range unless there are extenuating circumstances. Having a range greater than 5% indicates that one or more samples need to be retaken. Another check is the core measurement compared to the overall depth measurement. The core measurement should be at least 90% of the depth measurement except in very low or very high density snowpacks. A low core measure may indicate that part of the sample was lost, or that the sample was taken in or near an existing sample hole. The core measure may not exceed the depth sample, as this would indicate there was snow remaining in the sampler from a previous sample when the current one was taken. 7. Snow note checking: all field notes must be first checked in the field at the time of collection. Then the notes are rechecked and calculated in the office to prevent potential errors migrating into the database. Refer to the "Snow Survey Sampling Guide", Agriculture Handbook 169. 8. Data entry: the final step in the manual data collection procedure is data entry. The collected, processed and rechecked data are entered into the Snow Survey computer data base. As soon as this is done, a data listing must be printed and the values checked against those processed values from the office. any mistakes should be immediately corrected. 9. Sensor performance: performance can be affected in several ways: 1) cutters - these must be kept sharp, with the proper angle. Cutter teeth must be filed 90 degrees to the center of the tube in order to keep bias free (not over or under sampling by directing snow in/out) The precise diameter is the inside edge of the cutter. 2) bent of otherwise damaged tubes can affect the sampler. 3) tubes that have not been sufficiently siliconed or waxed may plug in cold circumstances resulting in underweight. 4) scales that are not properly maintained and calibrated can be a source of error. Properly maintained equipment minimizes bias. 2. Snow depth 1. Snow tube samples "how to sample" 1. Refer to the "Snow Survey Sampling Guide", Soil Conservation Service (currently the Natural Resources Conservation Service) Agriculture Handbook Number 169. 2. Sensor Function: The sensor in this case, is a federal snow sampler, or in the case of extremely deep snowpacks, the McCall sampler. It is essentially a tube with which an individual may take a snowpack sample at a specific location and point in time. Bias can be introduced into the samples of this sensor by the user. It is important that each sampler be trained in the science of snow measurement so that each snow course is measured consistently in time and space, regardless of the human component of the sensor. 3. Data units: inches of snow 4. Resolution: 0.5 inches 5. collection frequency: The collection frequency is determined by data need. The most typical situation is to manually sample snow courses as near the beginning of the months of: March, April and May as possible. April is typically the peak snowpack month and sampling one month prior and post April covers the potential of both early and late snowpack anomalies. Historically snowpacks have been measured beginning in January and ending in June. With the advent of telemetered data, this intensive measurement scheme was unnecessary. For specific cases where telemetered data are not available and data needs dictate, this type of measurement schedule might be applicable. 6. Ground truth: Manual sampling is traditionally the ground truth for other sensors as it physically takes a core of snow to weigh. There is no independent corroboration for this sample, however there are related ways that indicate whether a sample is correct. All sample densities along a course being measured should be within a 5% range unless there are extenuating circumstances. Having a range greater than 5% indicates that one or more samples need to be retaken. Another check is the core measurement compared to the overall depth measurement. The core measurement should be at least 90% of the depth measurement except in very low or very high density snowpacks. A low core measure may indicate that part of the sample was lost, or that the sample was taken in or near an existing sample hole. The core measure may not exceed the depth sample, as this would indicate there was snow remaining in the sampler from a previous sample when the current one was taken. 7. Snow note checking: all field notes much be rechecked and calculated in the office to prevent potential errors migrating into the database. Refer to the "Snow Survey Sampling Guide", Agriculture Handbook 169. 8. Data entry: the final step in the manual data collection procedure is data entry. The collected, processed and rechecked data are entered into the Snow Survey computer data base. As soon as this is done, a data listing must be printed and the values checked against those processed values from the office. any mistakes should be immediately corrected. 2. Aerial Markers: "How to sample" refer to NEH 22, section 2, page 19-22 1. Sensor Function: an aerial marker is simply a pole with cross members at specified intervals. The marker is "read" by flying over the marker and counting the number of cross members still visible above the snowpack along with an estimate of where the snow elevation is between the last visible cross member and the one below that is covered by snowpack. By knowing the total number of cross members, and their spacing interval, the total depth of snow can be calculated. An estimate of snow water equivalent may be calculated by using density from nearby locations. Aerial markers are used to supplement information from data sparse areas, or areas that are difficult, dangerous or expensive to access. Historically, there were 3 types of aerial marker locations: 1) a marker located directly on or adjacent to a snow course sample point, 2) a marker located near a snow course and 3) a marker as a stand alone data collection point. Through time, most of the type 1 and 2 markers were phased out leaving those aerial markers that are not associated with a snow course. Data from aerial markers are relatively inexpensive to obtain, they are relatively low maintenance sites. However, they represent only a single point of data collection and are only an estimate of snow water equivalent. 2. Data Units: this sensor measures snow depth in inches. 3. Data Resolution: this sensor can be read to the nearest 6 inches. 4. Collection frequency: the data collection frequency is dictated by need, most often read during the normal snow survey cycle (March, April, May) for 3 readings per year. 5. Ground Truth: this sensor may be corroborated from nearby sensor or by physically going to the site and taking a manual sample for both depth, density and snow water equivalent. In practice, ground truth is rarely performed at these sites because if they were easy, safe or economical to service, they would have already been converted to a snow course or SNOTEL in order to obtain better quality snowpack data. 3. Snow Stakes 1. sensor function: this sensor measures snow depth. It is simply a stake/pole/post on which is a tape or other graduated marker. The depth of snow is read from the marker, typically to the nearest 1/2 inch. These sensors have been typically associated with schools or with precipitation gages. 2. Data units: inches 3. resolution: 1/2 inch 4. collection frequency: data from snow stakes may be collected as needed. If in conjunction with a school or an individual, daily records may be taken. Monthly data are more common, in line with normal snow survey schedules and precipitation gage measurements. 5. Ground truth: the depth may be independently verified with a federal snow sampler. In common practice, it is rarely done because the measurement is so simple. 4. Storage precipitation gage 1. sensor function: this gage measures accumulated precipitation. Precipitation falls into a cylindrical tube and is stored till manually measured. In the past, this manual measure may have been done by manometer, dipstick or weighing the gage contents bucket by bucket. The current standard is an 8 inch diameter gage with an alter shield and uses a manometer to measure the contents. 2. data units: inches 3. resolution: 0.05 inch 4. collection frequency: as needed and practical, typically associated with a snow course and measured conjunctively. 5. sensor performance: check for leaks, an oil film to prevent evaporation, objects that may have fallen or been thrown into the gage, vandalism. It is hard to get more precipitation than has actually fallen, but extremely easy to get less due to leaks, gage capping, vandalism, etc. 6. ground truth: snowpack SWE increase from month to month is an index, surrounding station increases, etc.