Roy Jenne 22 Feb.1991 Rev. 4/91 Snowpack Data -------- ---- Information about snowpack and related hydrometeorologic data is given for the Western US, Alaska and southwest Canada. The data are very important to help forecast warm season river runoff. Manual observations of the water content of the snowpack started at some sites in 1922. By the 1930's, there were many sites. For the whole period, 1930 - present, there have been about 1500-2000 sites with snowpack measurements at any given time. Measurements are taken once a month during the snow season: Jan. 1, Feb. 1, --, May 1 and some stations have a reading at June 1. The readings are as close as possi- ble to the first of each month. There are many stations that just have read- ings Mar. 1, April 1, and May 1. Some stations also have a mid-month read- ing. There are also sites with automated daily measurements, especially daily precipitation, snowpack, and max/min temperature. Automated daily measurements started at about 60 to 70 sites in 1964. This number of sites was rather constant until 1978. These had pillows to sense the weight of snow to determine the water content. At first, only a few stations transmitted data by radio, but many more had chart recorders from which the data was later digitized. The SNOTEL network to automatically collect the data started in 1978, and many more stations were automated. At many of the old automatic sites they kept using all of the previous sensors, but interfaced the hardware with a new data communications system. Automatic SNOTEL sites: There were almost 500 sites in 1980, and about 520 automatic sites in 1985; by early 1991 the number had increased to about 560. The Soil Conservation Service (SCS) has a budget initiative for 1993 to try to achieve about 900 sites total. The cost to add a site (including installation) is about $18,000. The data is communicated to Portland, Oregon in real-time, using meteor-burst technology (the signal bounces off the ion- ized meteor trail). The SCS is responsible for gathering the snow measurements. They did have about 1650 manual snow course sites. In February 1991, there were fewer than 1100. SCS wants to reduce the number of manual measurement sites to about 800, but not fewer. The SCS takes the monthly readings at about half of the manual sites. At the rest, the quantities are measured by cooperating agencies and others. Measurements at an Automatic Site ------------ -- -- --------- ---- All sites measure the weight of the snow on the pillow, and the cumula- tive precipitation (water or frozen) for the water year which starts October first. The gage has a fluid to melt the snow and oil to stop evaporation. All sites measure air temperature. The data are automatically read out. The data base has daily minimum, maximum, average temperature, snow water content, precipitation, and other data if available. Some sites have sensors to support fire weather needs: winds, pressure, etc. A few have solar data. A selection of places have tipping bucket rain gages to help measure summer precipitation. The antenna and sensors (except for the snow pillow) have to be located high enough that they are above the deepest snowpacks that can occur at a given site. Region Covered by Stations and Data ------ ------- -- -------- --- ---- The western US states are covered by snow sites. The region includes Alaska, a few sites in Texas, and Minnesota. The database also includes monthly data from British Columbia and Alberta. Alaska has some long-term stations, but the coverage of stations is rather thin. The reason is that the stations are justified on the basis of benefits for agriculture or power generation. Collection of the Snow Data ---------- -- --- ---- ---- When the automated data arrives each day at the central office in Port- land, Oregon, it is accessed by the five data collection offices in the west (each with 6 people). These offices check the data for problems, and inter- polate for a complete daily record (for daily stations) to fill in small gaps. The data are then ready to be entered in the database at Portland, where three people handle data collection, four provide computer systems sup- ports, and seven prepare forecasts. Others bring the total staff to about twenty. In the central office, the data collection offices, and five water supply offices (each with one person), a total of 55 people are involved. About 97% of the sites are reporting now even though February is one of the more difficult months for meteor-burst communications. Newer sites can remember data for a few days so that if one transmission is missed, the data can be obtained later. SCS is in the fifth year of a seven-year upgrading of hardware at all sites. Maintenance: If the hardware at a site goes down, a person is dispatched to fix it within a very few days. I said that it seems like a nice network. "We are very proud of it". Data Bank ---- ---- There are daily data from the automatic stations and monthly data for both manual and automatic sites. There are 45,000 station-months of daily data and about 270,000 station-months of data in the archive. Only about 10 bytes (or less) of monthly data are needed for each station-month. Publications ------------ All data are available on-line within the Centralized Forecasting System (CFS). Measurements at a Site ------------ -- - ---- For snow measurements, SCS finds a meadow large enough that the measure- ments shouldn't be very affected by local snow drifts near trees. About 6 to 20 locations up and down the meadow are established for the manual snow meas- urements (snow-water content). The same locations are used from year to year. The samples are used to prepare one number for snow amount, for each site, at the beginning of each month. When an automatic site is installed in a meadow where manual measurements have been taken, they try to choose the "best" location. There is also an overlap period of manual measurements. The SCS has agreements with the management agencies to prevent changes in any forest near a snow site, but all changes can't be prevented. Trees grow, trees die or fall over, forests burn; but they hope to at least stop nearby logging that is not controlled. Very few sites are above timberline. They are located in areas of maximum snowfall, not much wind, and little snow drifting. The Forest Service helps to provide an extra effort for fire pro- tection of these sites. There has been a tendency to put in automatic stations at many of the best long-term manual sites so that daily data can be obtained quickly. There is a worry at Colorado State University (CSU), that some of our best long-period stations for climate trends can be upset in this way. SCS can make provisions to continue measurements (at least the April 1 measurement) at manual sites of particular scientific value. Forecasts --------- Each month during the period January first to May first, the SCS Water Supply forecasting staff in Portland prepares forecasts of the coming runoff season. These give projected runoff volume for many rivers for April-July and April-September. These are for virgin runoff, not including storage behind dams. The regression equations include the effects of seasonal rain- fall as well as the water from melting of the snow. Fluctuations in warm season rainfall are one source of error in the prediction. Also, a sudden melting of snow can lead to more runoff. All predictions are databased in the Portland facility. Tom McKee, CSU, notes that warm season rainfall in June-Sept. probably doesn't affect the total runoff very much. After the ground dries and it is warm, most of this precipitation probably goes into evaporation instead of runoff. The earlier water precipitation does matter. Who Uses the Snow Data? --- ---- --- ---- ---- People involved with hydropower generation, irrigation districts, river flow questions, water storage, fisheries management, and other water supply questions are all very interested in snowpack data. The climate community has an increasing interest. If climate warmed, there would be changes in the time and amount of runoff. There will be an increasing use of the data by people trying to simulate the snow pack using basic data about atmospheric flow, storms, clouds, radia- tion, temperature, and winds. The high mountain, remote locations of SNOTEL sites make them a desirable platform for air quality and other environmental sensors. Use of Snowpack Data by Colorado State University --- -- -------- ---- -- -------- ----- ---------- CSU obtained snowpack data from the Soil Conservation Service for several western states (Colorado, Wyoming, Utah, Montana, Idaho). They chose sites that had a long record and did some checks: + They wanted data for at least the 1951-85 period. + Looked at the station time series to check for consistency. + They decided not to use stations that were affected by a significant break in the record or a bad station move. + CSU talked with the snow supervisors in each state to obtain first-hand information about the quality of the site. On this basis, they dropped about 25% of the stations. It would be useful to have a few comments about why each station was dropped, and whether the problems of time consistency were considered mild or severe. Such notes were not saved. + They ended up with 275 sites for the five states. + CSU is doing most of their work with April first snowpack data, the last date that the snow is still increasing. These sites have 35 or more consecutive years of data but are not necessarily active in 1990. + The data from SCS looks clean, so if a user wants to work with the 275 sites that CSU used, they can obtain the data from SCS and the station list from CSU. Tom McKee and Nolan Doesken at CSU are the PIs for the CSU project and David Chagnon was the student heavily involved (phone: (303) 491-8545). Their reports include maps of station locations. Continuity of Climate Records ---------- -- ------- ------- For studies of climate trends, it is important to have a number of long period sites where we can be convinced that any measured changes are real and not just caused by changes in a forest or in measurement methods. Certain features of the automatic sites are very attractive; namely, they provide cumulative precipitation and temperature, as well as snow water content. Nolan Doesken, [CSU (303) 491-8545], has argued that one option is to put in automatic gear at a good long-period site, but still continue manual readings for at least the 1 April date. In several states, the 1 April meas- urement is the last monthly reading when the snowpack is still increasing. The snow information was from David Johnson (Program Manager), and Garry Schaefer (Leader, Data Resources team) at SCS, Portland address: Soil Conservation Service Water Supply Forecasting Staff 511 N.W. Broadway, Room 248 Portland, OR 97209 Phone: (503) 326-2843 FAX: (503) 326-6308