Station Criteria

Climate Stations

In 2016, the NDMC wanted to revisit the station selection process again to determine if more stations could be identified which may not have met the initial criteria, but were considered good enough to use going forward. Through this process, stations for both Alaska and Hawaii were identified from which the SPI could be calculated. Stations from the SNOTEL network were also identified as they also had long enough periods of record that a reasonable SPI value could be determined. At that point, the entire COOP network was again analyzed. From these stations, a new set of criteria were determined in which the station had to be open and current, have a record of at least 40 years in length, had data gaps of 2 consecutive years or less, and had no more than 5 data gaps in the last 40 years of record. Stations that met these criteria, along with those stations which were originally selected, make up the inventory of stations used.

The update includes 1558 new stations. Of these new stations, 434 are SNOTEL sites that are operated by the USDA’s Natural Resources Conservation Service.

For all stations (legacy stations plus new stations):

Total Number of Stations: 4183 Number Percent
Stations with 50+ years data 3861 83.6%
Stations with 60+ years data 3218 70.0%
Stations with 70+ years data 2113 45.8%
Stations with 80+ years data 1400 30.3%
Stations with 90+ years data 1060 23.0%
Stations 100+ years data 720 15.6%
Stations 110+ years data 453 9.8%
Stations 120+ years data 229 5.0%
Stations 130+ years data 16 0.3%

Current Stations by State

AL: 54 GA: 82 MA: 34 ND: 62 OR: 195 VA: 60
AK: 40 HI: 49 MD: 23 NE: 129 PA: 91 VI: 4
AR: 114 IA: 124 ME: 31 NH: 17 PR: 29 VT: 12
AZ: 97 ID: 126 MI: 96 NJ: 19 RI: 4 WA: 118
CA: 307 IL: 110 MN: 105 NM: 106 SC: 58 WI: 120
CO: 128 IN: 66 MO: 108 NV: 53 SD: 81 WV: 45
CT: 16 KS: 206 MS: 82 NY: 90 TN: 57 WY: 97
DE: 3 KY: 53 MT: 234 OH: 96 TX: 306  
FL: 60 LA: 71 NC: 103 OK: 73 UT: 152  

Hydrology Stations

The USGS has a network of streamflow stations which are part of their hydro-climatic network (HDCN-2009). These stations are screened to exclude sites where human activities such as artificial diversions, storage, and other activities in the drainage basin or in the stream impact flows. In addition, these sites were selected also based upon the long period of record. These sites were also selected based on the long period of record. This network numbered 1,659 stations in 1992 and has since been reduced to 743 sites.

From the daily data, the streamflow sites were used to calculate a standardized streamflow index (SSI) which is similar to an SPI, but using hydrological data.

Original Station Selection Criteria

To construct a reliable historical drought atlas, we only used data from observation stations that met criteria for consistency.   To establish the best stations, criteria needed to be established for screening the available stations.  Using precipitation records from the National Weather Service Cooperative data (COOP) that is archived by the Regional Climate Centers (RCC) in their Applied Climate Information System (ACIS) as the basis for the data, the following criteria were established for the Drought Atlas:

  • The station had to currently be open and taking regular observations
  • The station had to have a minimum of 40 years of data available, going backwards from 2009
  • The station could not have more than 2 consecutive months of missing data at any time in the period of record
  • A unique period of record (POR) was established for each station based on meeting the stated criteria

Using the 12,000+ stations in the COOP database, two screenings of the data were done.  The first screening looked for stations that had at least 30 years of data in their record and had 80 percent or more of the data available. A report for each station was made showing the months that had missing data for each year.  By looking at the most current data and going back in time, we determined for each station how far back the data met our criteria.  Gaps in the records were investigated to determine if the data existed somewhere else. The National Climatic Data Center (NCDC) provided access to scanned copies of the COOP forms in their archive via the Web Search Store Retrieve Display (WSSRD) system, which has transitioned over to the Environmental Document Access and Display System (EDADS), for this investigation.  By comparing the data that was available on paper with the gaps in the digital archive in ACIS, we identified 288 stations for which the data gaps could be rectified and entered into the digital ACIS archive.  The High Plains Regional Climate Center (HPRCC) staff keyed in the data for the gaps we identified using WSSRD, and these stations were then used in the atlas project.   After consulting with several experts, we determined that a 30 year record is good, but using 40 years of data would be best, so we adjusted the initial station query to include just the stations with 40 or more years of data. 

Initial Review

After this exercise, we presented the information to the American Association of State Climatologists (AASC) at their annual meeting.  We asked for their assistance in determining whether these were indeed good stations and also if they knew of other additional stations that would meet our criteria.  Twenty-two state climatologists and one regional climate center offered their help.  We found that most were satisfied with our stations, but a few did suggest stations we did not have on our list.  After investigating our method, we determined that another screening of the full COOP data set was needed.

Additional Stations

Using the established criteria, we looked at the data going backward and stopped our search the first time a 2-month gap was found.  We found that using the 80 percent of available data criteria excluded a number of stations where the data had large gaps in the data early in the record, but then were complete.  The 80 percent criteria also excluded stations that are primarily “event reporters” in that they only record precipitation when it takes place and do not record zeros very often.  By screening the data a second time and comparing the stations to those we already had, we were able to add 915 more stations to our catalog, bringing the total stations in the Drought Risk Atlas up to 3059.

Initial Stations Chosen

The initial results for stations to be included in the atlas were:

  • 2144 stations with 40 or more years of data meeting all criteria
  • 316 stations with 100+ years of data
  • 482 stations with 90+ years of data
  • 724 stations with 80+ years of data
  • 952 stations with 70+ years of data
  • 1244 stations with  60+ years of data
  • 1775 stations with 50+ years of data

Here is the final breakdown of stations used in the Drought Atlas meeting all our criteria:

  • 3059 stations with 40+ years of data
  • 349 stations with 100+ years of data (11.50%)
  • 537 stations with 90+ years of data (17.68%)
  • 827 stations with 80+ years of data (27.22%)
  • 1170 stations with 70+ years of data (38.51%)
  • 1733 stations with 60+ years of data (57.04%)
  • 2462 stations with 50+ years of data (81.04%)
Atlas Stations

Initial Stations by State

AL: 37 GA: 66 MA: 27 NC: 88 OH: 84 TX: 226
AR: 64 IA: 84 MD: 16 ND: 57 OK: 60 UT: 70
AZ: 79 ID: 53 ME: 23 NE: 106 OR: 94 VA: 56
CA: 187 IL: 75 MI: 66 NH: 15 PA: 70 VT: 11
CO: 67 IN: 47 MN: 74 NJ: 15 RI: 3 WA: 84
CT: 9 KS: 142 MO: 66 NM: 80 SC: 55 WI: 92
DE: 3 KY: 41 MS: 59 NV: 22 SD: 70 WV: 41
FL: 47 LA: 45 MT: 116 NY: 75 TN: 46 WY: 46