README FILE 28 March, 2000 Richard E. Moritz SHEBA Project Office Polar Science Center, APL-UW Data Set: SHEBA Daily Precipitation Amount, Version 3 *************************************************************** THIS DATA SET REPLACES AND SUPERCEDES THE PREVIOUS DATA SET: "SHEBA: Ice Camp Daily Precipitation Amount (ASCII) (Moritz)" that was posted on the JOSS/CODIAC/SHEBA data system. **************************************************************** VARIABLE: Daily Precipitation Amount (Water Equivalent) UNITS: Millimeters of liquid water equivalent INSTRUMENT: Nipher shielded snow gauge system LOCATION: Arctic Ocean, SHEBA Ice Station START DATE: 29 October, 1997 END DATE: 09 October, 1998 SAMPLING RATE: Nominally one measurement per day. MISSING DATA: During the period 11, 12, 13 April, the gauge was out of service as the measurement site was moved away from a pressure ridge that formed at the end of March. After the precipitation observation on 18 August, the gauge was out of service for one day as the measurement site was moved a few tens of feet to a new location. During summer the hole for the gauge support had melted somewhat, producing a tilt (approximately 5 degrees off vertical on 19 August, 1998). Also, due to surface melt the top of the gauge was 67" inches above the nearby surface, 7" more than the nominal 60" height. After freezing in the support at the new site, the tilt was measured at less than 1 degree off vertical, and the height of the top of the gauge was 58". Observations resumed on 20 August. On a few occasions, the gauge could not be visited for the daily observation. In these cases, the measured precipitation represents accumulation over a longer period, extending back to the time of the previous measurement. MEASUREMENT PROCEDURE: Normally once per day, about 10:00 a.m. local time, the snow gauge was inspected visually. Occasionally the measurement was made more than once in a day. If any precipitation is visible in the receiver, the receiver is removed from the shield and a clean, dry receiver put in its place. The receiver is immediately covered to prevent evaporation, and brought into a warm hut. After all precipitation in the receiver has melted (typically a few minutes) the contents of the receiver are poured into a graduated glass cylinder. The observer reads the level of the meniscus in the graduated cylinder, enters this in the log book together with the date and time of observation, and leaves the empty receiver in the warm hut to dry thoroughly. MEASUREMENT SITE: generally about 30 meters from the SHEBA Project Office 10-meter surface met tower #1. Initially (October 97) this tower was located approximately 200 meters off the starboard bow of the ship. After the ridging events of 28 March 1998, the tower had been displaced to a location approximately 300 meters forward of the ship and approximately 30 meters east of a 10-foot pressure ridge. The snow gauge was moved on 13 April, 1998 to the vicinity of the new site of SPO tower #1, approximately 400 meters from the ship, at an angle of approximately 40 degrees to starboard of the bow, and approximately 200 meters east of the pressure ridge. Throughout the entire period, the immediate environment around the snow gauge was multiyear ice of approximately 2 meters thickness with typical undulating snow cover. Except during approximately 29 March - 11 April (the pressure ridge mentioned above) there were no obstructions to air flow near the gauge. The top of the Nipher shield was maintained at approximately 5' above the surrounding terrain. This distance increased gradually to approximately 5' 7" during the melt season. The gauge was repositioned and reset to 58" height, at 2005 GMT on 19 August, 1998. CORRECTIONS TO THE DATA: As noted by many authors (e.g. Goodison and Yang, 1996) it is necessary to apply corrections to precipitation measurements made in high latitudes. In this data set we present both the raw (uncorrected) measurements and a corrected estimate based on Goodison and Yang, 1996. This gives the user of the data flexibility to apply customized corrections to the raw data, if desired. Definitions: TP = True Precipitation (in mm w. eq.) MP = Measured Precipitation (read from the graduated cylinder) CGC = Corrected Gauge Catch (amount that was caught by the receiver during the measurement period) E = Evaporation Loss (amount lost from the receiver between the first precipitation event of the measurement period and the covering of the gauge prior to pouring out the contents). W = Wetting Loss (amount left in the receiver when the contents are poured into the graduated cylinder) DFIR = Amount of precipitation measured by a double-fenced intercomparison reference. WS = Average Wind Speed at gauge height during the precipitation events of a given measurement period (knots). T = 10-meter air temperature averaged over time from the first precip event of the measurement period to the end of the measurement period. The difference between the true precipitation amount TP, and the amount measured in the graduated cylinder MP, depends on the following factors: i. Evaporation During the interval between a meteorological precipitation event and the removal of the receiver from the stand, evaporation takes place causing a loss of water substance from the receiver. According to Goodison and Yang (1996), in Finland this loss is typically about 0.1 mm per day in winter, and is much larger in summer than in winter. In correcting the SHEBA data, we assume that the time rate of evaporation from the gauge is a linear function of air temperature, and that the total loss by evaporation for one observation is proportional to the elapsed time between the precipitation event and the covering of the gauge when it is removed from the stand. We assume the evaporation rate is 0.1 mm/day at T = -40 C and 0.2 mm/day at T = 0 C, where T is the 10-meter air temperature. The "trace" readings are corrected in a different way by adding a single, constant correction of .1 mm to account for the combined effects of evaporation and wetting. ii. Wetting When the contents of the receiver are poured into the glass cylinder, some of the liquid water remains in the receiver, wetting its sides and bottom. According to Goodison and Yang, "average wetting loss can be up to 0.2 mm per observation." For correcting the SHEBA measurements, we assume a constant wetting loss of 0.1 mm water equivalent each time the receiver is emptied into the graduated cylinder, provided the measured amount is larger than "trace". The "trace" readings are corrected in a different way by adding a single, constant correction of .1 mm to account for the combined effects of evaporation and wetting. iii. Trace amounts According to Goodison and Yang, for the Canadian snow gauge with the Nipher shield, a precipitation event of less then 0.2 mm is treated as a trace. For the SHEBA measurements, we recorded amounts greater than or equal to 0.1mm by reading the level of the meniscus in the graduated glass cylinder. A reading of "trace" was entered when visual inspection showed some precipitation in the receiver, but the amount was insufficient to yield 0.1 mm in the graduated cylinder. Therefore, we shall assign a value of 0.05 mm for the gauge contents for each "trace" observation. When the contents equals "trace", both wetting and evaporation are limited by factors associated with the amount of water substance in the receiver, and so will generally be smaller than for larger contents. We add a single, constant correction of .1 mm to account for the combined effects of evaporation and wetting on trace amounts. Therefore readings recorded as "trace" have a corrected value of TP = 0.15 mm. iv. Wind The corrections in i, ii and iii above aim to produce a more accurate estimate of the amount of precipitation that was actually caught by the receiver during the precipitation events that occurred over the observation period. This amount is called the "corrected gauge catch" CGC. Symbolically, we have (1) CGC = MP + W + E where W and E are the corrections for wetting and evaporation, respectively. It is known from long-term calibration studies that CGC underestimates the true precipitation that falls on a unit area. The correction depends on the wind speed and the type of precipitation during the precipitation events. Generally, the percentage of the true precipitation that is caught by the gauge decreases with increasing wind speed. According to Goodison, the best estimate of true precipitation is obtained with a gauge sited in a large, homogeneous area of bushes cut to gauge height. We identify the precipitation caught in such a "bush gauge" setup with TP. The "bush gauge" setup is not available at most calibration sites, so a reference method has been defined, using a more common setup employing an octagonal, vertical, double fence shield. This setup is called the "DFIR" - Double Fenced Intercomparison Reference. DFIR measurements have been calibrated at the (few) bush gauge sties, and it has been found that after correction for wetting evaporation and wind speed, the DFIR yields 93% of the bush gauge catch. Therefore we will use (2) TP = DFIR/0.93 to correct the SHEBA measurements. It remains to calculate DFIR from CGC. It is known from extensive intercomparison data sets that the Canadian gauge with the Nipher shield, but no fence (this was the setup at SHEBA) understimates the DFIR catch by an amount that depends on the wind speed and the type of precipitation. Averaged over all types of precipitation occuring at Canadian snow measurement stations, and restricting the statistical fitting to cases with at least 3 mm, Goodison and Yang estimate the ratio of corrected gauge catch (CGC) to DFIR as a function of wind speed (at gauge height), for the Nipher (Figure 6 in Goodison and Yang, 1996). Over the range of windspeed from 0 to 9 m/s this correction is well approximated by the function (3) [CGC/DFIR] = R' = exp ( - ( WS/W* )^2 ) where WS is the wind speed at gauge height during the precipitation event and W* is 10.8 m/s = 21 knots. When WS = 0, R' = 1, and when WS = 9 m/s, R' = 0.5. For example, the typical wind speed at gauge level during SHEBA was approximately 4.5 m/s, and in this case the ratio is R' = 0.84. Given CGC and WS, we use this equation in the form (4) DFIR = CGC/R' to calculate the (corrected) DFIR catch. Finally, we use equation (2) to estimate the true precipitation TP from the DFIR. v. Current and Past Weather Types For each non-zero MP (including "trace" entries) we define a time interval extending back to the previous observation. For each synoptic reporting time (00, 06, 12, 18 GMT) in this interval, we extract the wind speed, air temperature, present weather and past weather variables from the SHEBA Surface Ship Report data set. We categorize "events" for each synoptic interval according to the present and past weather as follows: **Precipitation Events** Present Weather Categories 50 to 99 Various types of precipitation at time of observation 29 Thunderstorm in past hour 20-27 Other precipitation types in past hour Past Weather Categories 5-9 Various types of precipitation in the past 6 hours If there is a non-zero MP, but no Precipitation Event, we look for "Other Events", defined here to be events that can cause a non-zero measurement of gauge contents, but do not necessarily imply that any precipitation fell on the surface (top of the snow surface or sea ice surface). **Other Events** Present Weather Categories 10 Mist 36-39 Blowing Snow 40-49 Fog Past Weather Categories 4 Fog 3 Blowing snow At the conclusion of this step, the following mutually exclusive outcomes are possible for each non-zero value of MP: (A) We have identified MP with one or more Precipitation Events, OR (B) We have identified MP with one or more Other Events, OR (C) We have identified MP with no events. Case (A) We assign a time to the first and last precipitation events in the interval. Then we compute the average surface air temperature for the period from the first event to the time of the observation of precipitation amount. We also compute the wind speed at gauge height averaged over all the precip events (but not necessarily over all the time since the first event in the interval). The Surface Ship Reports are used to compute these averages. The wind speed in the Ship Reports is measured at 10 meter height, and the precipitation gauge is nominally at 5 foot height. We adjust the reported wind speed to gauge height by multiplying by the constant factor 0.70. The 10-meter air temperature is used directly as an estimate of air temperature at the gauge height. The evaporation rate is computed from this temperature, and this is multiplied by the elapsed time between the first precipitation event in the interval and the time of precip observation, to obtain E. CGC is then computed by adding E and W to MP. Then the average wind speed at gauge height is used to compute the ratio R', and equations (2) and (3) are used to calculate TP. In the event the measured contents were entered as "trace", we proceed in the same way using CGC = 0.15 mm. In case (B), we have a non-zero value of measured contents, no "precipitation event" and at least one "other event" in the time interval. In this case, we assign the constant value TP = 0. This corresponds to the assumption that the water substance in the gauge got there through a process that did not add water substance to the surface of the snow and sea ice, averaged over a wide area. These other events may alter the horizontal distribution of snow, e.g. blowing snow. In case (C), we have a non-zero value of measured contents, no "precipitation event" and no "other event" in the time interval. In this case, we define an event time midway between the observation times of the current and preceding measurements. Using the air temperature and adjusted wind speed at this time, we compute CGC, R' and TP as in case (A). The resulting value of TP is then divided by two to produce the estimate of true precipitation for case (C). This corresponds to the assumption that it is equally probable that the non-zero MP resulted from precipitation processes, or non-precipitation processes, since the current weather code provides no information about this. ADDITIONAL DETAILS The times of "precipitation events" and "other events" are defined to be the time of the corresponding surface ship reports when using current weather variables. When using past weather variables, we subtract 3 hours from the time of the corresponding surface ship reports to get the event time. FORMAT The final SHEBA Project Office precipitation data file has the following information: Column 1: Date and Time of the current Precip Obs: yymmddhhmm (GMT) Column 2: Date and Time of the previous Precip Obs: yymmddhhmm (GMT) Column 3: Date and Time of the First "Event" in the interval: yymmddhhmm (GMT) Column 4: Weather code for the First "Event" (0 - 99) Column 5: 1 => Col 4 refers to Current Weather; 2 => Col 4 refers to Past Weather Column 6: Date and Time of the Last Event in the interval: yymmddhhmm (GMT) Column 7: Type of Last Event: 00-99 Column 8: Average 10 meter wind speed during the Events: VV.V (knots) Column 9: Average 10 meter air temperature since the first event: TTT.T (Celsius) Column 10: Measured Precip (MP): CCC.CC (water equivalent millimeters) Column 11: Estimated True Precip (TP): PPP.PP (water equivalent millimeters) REFERENCES ----------- Goodison, B.E. and D. Yang, 1996: In-situ measurement of solid precipitation in high latitudes: the need for correction. Proc. of the Workshop on the ACSYS Solid Precipitation Climatology Project, Reston, VA, USA, 12-15 Sept. 1995, WCRP-93, WMO/TD No. 739, pages 3-17.