“TD-3721, Gridded U.S. daily precipitation and snowfall time series” by Pavel Ya. Groisman THE ARCHIVE DESCRIPTION Two gridded data sets were produced for each element (currently for daily precipitation and snowfall for the period from January 1, 1948 to December 31, 1999, or 18,993 days). Directory US_gridded at the flux workstation contains the entire archive at the raid2 drive. The data are located in separate directories (SNOW and PRCP), each of which has two subdirectories (grid1 and grid2) that contain the output of the forth and sixth steps of the gridded procedure. Specifically, grid1 contains the time series only for A-cells (2752 and 2635 for precipitation and snowfall, respectively). Each A-cell that was included in the snowfall data set is also included in the A-cell group of the precipitation data set. The grid2 directory contains A and B cells daily time series (3341 and 3293 for precipitation and snowfall, respectively) that are interpolated (simulated) according to the algorithm of the sixth step and include the time series that are serially complete. UNITS: tenths of mm for precipitation and tenths of cm for snowfall. VOLUME: This data set consists of over 11,000 files. Each uncompressed grid1 file is 0.8 Mb and each uncompressed grid2 file is 0.7 Mb. Keeping in mind a large amount of these files (more than 11 thousand) a compressed format is strongly recommended. Standard UNIX compression reduces the reduces the archive volume to a “manageable” size of less than 1 Gb (~ 835 Mb). Information for each cell is contained in a separate file with a name such as: 116_042.SNOWgrid1, where the element (SNOW/PRCP) and product (grid1/grid2) constitute the last 9 characters of the name, and the first 7 symbols characterize the cell number (the coordinates of its center) according to the rule outlined in the FORTRAN code below. ...... real lon(116),lat(49) do i=1,116 lon(i)=-124.75+0.5*(i-1) end do do j=1,49 lat(j)=25.25 +0.5*(j-1) end do ....... Each "grid1"-file was created with the following write statement: character*2 OB character*6 stnID integer year,mon,day,out(110,12,31,2) real lon(116),lat(49) ..... do 4111 year=1948,1999 iii=year-1890 do 4111 mon=1,12 idaye=idayend(mon,year) do 4111 day=1,idaye write(12,4101) lon(i),lat(j),year,mon,day, * (out(ii,mon,day,kk),kk=1,2),stnID,OB 4101 format(f7.2,1x,f5.2,1x,i4,2i2.2,2(1x,a4),1x,a6,1X,A2) 4111 continue External function idayend(mon,year) delivers the number of days in a given month and year. The output file out() contains the values for each year, month, and day: ? the valid unadjusted value of the "closest" to the grid cell center station with a six digit cooperative station number, stnID, ? the same value but scale adjusted to better represent the cell center, and ? the time of observation, OB (according table 2). If no any station with valid data were found within the A-cell in the given day, missing values (-1) were inserted in these two positions, and the stnID and OB positions are left blank. Examples: File 043_020.PRCPgrid1 -103.75 34.75 19480101 0 0 294175 07 -103.75 34.75 19480102 0 0 294175 07 -103.75 34.75 19480103 0 0 294175 07 -103.75 34.75 19480104 0 0 294175 07 -103.75 34.75 19480105 0 0 294175 07 -103.75 34.75 19480106 0 0 294175 07 -103.75 34.75 19480107 0 0 294175 07 -103.75 34.75 19480108 0 0 294175 07 -103.75 34.75 19480109 0 0 294175 07 -103.75 34.75 19480110 0 0 294175 07 -103.75 34.75 19480111 0 0 294175 07 -103.75 34.75 19480112 0 0 294175 07 -103.75 34.75 19480113 0 0 294175 07 -103.75 34.75 19480114 0 0 294175 07 -103.75 34.75 19480115 0 0 294175 07 -103.75 34.75 19480116 0 0 294175 07 -103.75 34.75 19480117 0 0 294175 07 -103.75 34.75 19480118 0 0 294175 07 -103.75 34.75 19480119 0 0 294175 07 -103.75 34.75 19480120 0 0 294175 07 -103.75 34.75 19480121 0 0 294175 07 -103.75 34.75 19480122 0 0 294175 07 -103.75 34.75 19480123 64 65 294175 07 ......................................... -103.75 34.75 19991214 0 0 297226 18 -103.75 34.75 19991215 0 0 297226 18 -103.75 34.75 19991216 0 0 297226 18 -103.75 34.75 19991217 0 0 297226 18 -103.75 34.75 19991218 0 0 297226 18 -103.75 34.75 19991219 0 0 297226 18 -103.75 34.75 19991220 0 0 297226 18 -103.75 34.75 19991221 0 0 294175 07 -103.75 34.75 19991222 0 0 294175 07 -103.75 34.75 19991223 13 13 297226 18 -103.75 34.75 19991224 0 0 297226 18 -103.75 34.75 19991225 0 0 297226 18 -103.75 34.75 19991226 0 0 297226 18 -103.75 34.75 19991227 0 0 297226 18 -103.75 34.75 19991228 0 0 297226 18 -103.75 34.75 19991229 0 0 297226 18 -103.75 34.75 19991230 0 0 297226 18 -103.75 34.75 19991231 0 0 297226 18 ****************************************** File: 003_031.PRCPgrid1 -123.75 40.25 19480101 -1 -1 -123.75 40.25 19480102 -1 -1 -123.75 40.25 19480103 -1 -1 -123.75 40.25 19480104 -1 -1 -123.75 40.25 19480105 -1 -1 -123.75 40.25 19480106 -1 -1 -123.75 40.25 19480107 -1 -1 ...... -123.75 40.25 19630307 5 4 047404 08 -123.75 40.25 19630308 0 0 047404 08 -123.75 40.25 19630309 30 25 047404 08 -123.75 40.25 19630310 0 0 047404 08 -123.75 40.25 19630311 0 0 047404 08 -123.75 40.25 19630312 0 0 047404 08 -123.75 40.25 19630313 0 0 047404 08 -123.75 40.25 19630314 51 42 047404 08 -123.75 40.25 19630315 61 50 047404 08 -123.75 40.25 19630316 277 229 047404 08 -123.75 40.25 19630317 3 2 047404 08 -123.75 40.25 19630318 0 0 047404 08 ...... -123.75 40.25 19991226 0 0 047404 08 -123.75 40.25 19991227 0 0 047404 08 -123.75 40.25 19991228 0 0 047404 08 -123.75 40.25 19991229 0 0 047404 08 -123.75 40.25 19991230 0 0 047404 08 -123.75 40.25 19991231 0 0 047404 08 ****************************************** Each "grid2" file was created with the following write statement: integer year,mon,day,out(110,12,31,3) real lon(116),lat(49) ............ do 4999 year=1948,1999 ii=year-1890 do 4999 mon=1,12 idaye=idayend(mon,year) do 4999 day=1,idaye write(13,4101) lon(i),lat(j),year,mon,day, * (out(ii,mon,day,kk),kk=1,3) 4101 format(f7.2,1x,f5.2,1x,i4,2i2.2,3(1x,i4)) 4999 continue .......... Each line in a grid2 file contains a year, month, day, and three data values: the first two are the same as in the appropriate grid1 file (when this grid1 file is absent for B- cells missing code values "-1" are in these two columns), and the third value is (a) the same as the second value if it is not missing or (b) interpolated/simulated infilled value. Examples: File 063_017.PRCPgrid2 -93.75 33.25 19480101 -1 -1 0 -93.75 33.25 19480102 -1 -1 0 -93.75 33.25 19480103 -1 -1 0 -93.75 33.25 19480104 -1 -1 0 -93.75 33.25 19480105 -1 -1 152 -93.75 33.25 19480106 -1 -1 0 -93.75 33.25 19480107 -1 -1 0 -93.75 33.25 19480108 -1 -1 0 -93.75 33.25 19480109 -1 -1 0 -93.75 33.25 19480110 -1 -1 0 -93.75 33.25 19480111 -1 -1 0 -93.75 33.25 19480112 -1 -1 0 -93.75 33.25 19480113 -1 -1 0 -93.75 33.25 19480114 -1 -1 0 -93.75 33.25 19480115 -1 -1 0 -93.75 33.25 19480116 -1 -1 0 -93.75 33.25 19480117 -1 -1 32 -93.75 33.25 19480118 -1 -1 0 -93.75 33.25 19480119 -1 -1 341 -93.75 33.25 19480120 -1 -1 126 .................................... -93.75 33.25 19991209 -1 -1 0 -93.75 33.25 19991210 -1 -1 42 -93.75 33.25 19991211 -1 -1 0 -93.75 33.25 19991212 -1 -1 918 -93.75 33.25 19991213 -1 -1 0 -93.75 33.25 19991214 -1 -1 0 -93.75 33.25 19991215 -1 -1 0 -93.75 33.25 19991216 -1 -1 0 -93.75 33.25 19991217 -1 -1 0 -93.75 33.25 19991218 -1 -1 0 -93.75 33.25 19991219 -1 -1 0 -93.75 33.25 19991220 -1 -1 0 -93.75 33.25 19991221 -1 -1 0 -93.75 33.25 19991222 -1 -1 0 -93.75 33.25 19991223 -1 -1 0 -93.75 33.25 19991224 -1 -1 0 -93.75 33.25 19991225 -1 -1 0 -93.75 33.25 19991226 -1 -1 0 -93.75 33.25 19991227 -1 -1 0 -93.75 33.25 19991228 -1 -1 12 -93.75 33.25 19991229 -1 -1 0 -93.75 33.25 19991230 -1 -1 0 -93.75 33.25 19991231 -1 -1 0 **************************************** File 021_048.SNOWgrid2 -114.75 48.75 19480101 0 0 0 -114.75 48.75 19480102 0 0 0 -114.75 48.75 19480103 0 0 0 -114.75 48.75 19480104 0 0 0 -114.75 48.75 19480105 0 0 0 -114.75 48.75 19480106 0 0 0 -114.75 48.75 19480107 102 101 101 -114.75 48.75 19480108 0 0 0 -114.75 48.75 19480109 0 0 0 -114.75 48.75 19480110 5 5 5 -114.75 48.75 19480111 0 0 0 -114.75 48.75 19480112 0 0 0 -114.75 48.75 19480113 0 0 0 -114.75 48.75 19480114 0 0 0 -114.75 48.75 19480115 0 0 0 -114.75 48.75 19480116 0 0 0 -114.75 48.75 19480117 0 0 0 -114.75 48.75 19480118 0 0 0 -114.75 48.75 19480119 0 0 0 -114.75 48.75 19480120 0 0 0 -114.75 48.75 19480121 -1 -1 0 ............ -114.75 48.75 19720925 0 0 0 -114.75 48.75 19720926 0 0 0 -114.75 48.75 19720927 25 25 25 -114.75 48.75 19720928 25 25 25 -114.75 48.75 19720929 0 0 0 -114.75 48.75 19720930 0 0 0 -114.75 48.75 19721001 0 0 0 -114.75 48.75 19721002 0 0 0 -114.75 48.75 19721003 0 0 0 -114.75 48.75 19721004 0 0 0 -114.75 48.75 19721005 0 0 0 -114.75 48.75 19721006 0 0 0 -114.75 48.75 19721007 0 0 0 -114.75 48.75 19721008 0 0 0 -114.75 48.75 19721009 0 0 0 -114.75 48.75 19721010 102 128 128 -114.75 48.75 19721011 13 16 16 -114.75 48.75 19721012 0 0 0 ....... -114.75 48.75 19991225 0 0 0 -114.75 48.75 19991226 0 0 0 -114.75 48.75 19991227 0 0 0 -114.75 48.75 19991228 0 0 0 -114.75 48.75 19991229 0 0 0 -114.75 48.75 19991230 0 0 0 -114.75 48.75 19991231 25 25 25 ******************************************** The root directory contains the following two inventory files with the list of the stations used for gridding of precipitation: ? "5886_contig_usa_stn_withPRCP.inv" and snowfall: ? “5361_contig_usa_stn_withSNOW.inv". This directory also contains a file that lists the history of observational times for each station: ? “5886_contig_usa_stn_withPRCP.tobs_history”. The structure of the two inventory files is identical and is described by the following FORTRAN read statement: character*11 stnx character*30 name character*6 stnID character*5 wmo,wban,linf real lat,lon,elev integer i1,j1,gridelev read(10,2100)stnx,lat,lon,elev,name,wmo,wban,linf, * i1,j1,gridelev stnID=stnx(6:11) 2100 format(a11,1x,F6.2,1x,f7.2,f5.0,3x,a30,a5,1x,a5,1x,a5, * 1x,2i3,1x,i4) where wmo and wban are the appropriate WMO and WBAN id station numbers (if present), stnID is the station cooperative ID number (always available), lat, lon, elev, name are the station coordinate, elevation (in meters) and name. The five character string linf characterizes the data availability in five original data sources used to generate the gridded data set in the order of preferences: daily HCN first character; TD3200 second character; TD3205 (Kunkel data set) third character; Dave Robinson data set forth character; and Current data rescue archive fifth character. The presence of the station in the archive is indicated by "1" and the absence by the blank character. For a better visual navigation the linf string preceded by a star character. In an example below the station CLANTON is available from three (of the five) NCDC archives: TD3200, the Robinson data set (Frei et al. 1999), and the current data rescue archive. .... 42500011566 31.53 -88.25 87 CHATOM 4 N …. * 1 74 14 53 42500011620 33.27 -86.33 127 CHILDERSBURG WATER PLANT * 1 78 17 190 42500011690 31.62 -87.55 15 CLAIBORNE L & D * 1 75 14 78 42500011694 32.82 -86.65 186 CLANTON * 1 11 77 16 152 42500011725 31.88 -85.47 152 CLAYTON * 1 80 14 107 .... i1 and j1 indicate the exact 0.5 x 0.5 grid cell number to which the station belongs. gridlev provides the average cell elevation in meters. The two-number order of the cell is defined from the Fortran code below. ...... real lon(116),lat(49) do i=1,116 lon(i)=-124.75+0.5*(i-1) end do do j=1,49 lat(j)=25.25 +0.5*(j-1) end do ....... The file with the history of the times of observations ("5886_contig_usa_stn_withPRCP.tobs_history") can be read with the following FORTRAN statement: integer number, MON(60),YEAR(60) character*2 OT(60),OP(60) read(18,222) stnID,number, (MON(I),YEAR(I),OT(I),OP(I), * I=1,number) 222 FORMAT (a6,1x,i2,60(1X,2I2.2,1x,2A2)) where the number of groups of records (maximum 60 but usually in the first digits) provide the dates when the change of the time of observation took place or the first time was reported. Month MON and Year-1000 (YEAR) are followed by two two-character codes (OT and OP for temperature and precipitation respectively) that characterize the available information about the observation time according to Table 2. Stars that are inserted into some of these groups indicate that the information was taken from the NCD digital database while other groups are the result of manual processing. Example: 010008 3 0151 9924 0754 9907 0399*9907 010140 3 0151 9908 0867 9907 0399*9907 010178 8 0151 9908 0652 9906 0357 9907 0258 0717 0358 0707 0695*0808 0696*0808 0399*0808 010252 6 0151 0808 0160 0707 0660 0808 0380 1717 0783 0707 0399*0707 ................ Known problems. 1. At some stations in early 1948 the absence of snowfall, precipitation, and snow on the ground was coded as missing values. They were simply skipped in the data set and during the digitization phase were interpreted as missing. In the example below this was found at the two Nebraska stations (Fullerton and Genoa 2 W). These stations were first and third in the preference list of four for use to represent the daily snowfall values in this cell in January-February. During the first five months of 1948, these were the only two stations available for this particular 0.5 x 0.5 grid cell that sporadically reported snowfall (see example below). As a result from 60 values for these two months in 1948, 9 values are reported as present and 51 others as missing. Among the nine present values only three are equal to zero and six others give non-zero values while the probability of the day without snowfall at this grid cell in these months is close to 0.9 (0.86 for January and 0.91 for February). This means that most probably all non-zero values of snowfall has been already reported for these two months in 1948. However, our infilling process at the sixth stage randomly added 6 more non-zero values instead of missing data. This cell is very well infilled with the daily snowfall data (99.2%) but the above mentioned difficulty has to be kept in mind in analyses of the 1948 snowfall and precipitation data. Example. File 055_033SNOWgrid1 -97.75 41.25 19480101 51 51 253185 17 Stn. Genoa 2 W -97.75 41.25 19480102 -1 -1 -97.75 41.25 19480103 -1 -1 -97.75 41.25 19480104 -1 -1 -97.75 41.25 19480105 -1 -1 -97.75 41.25 19480106 -1 -1 -97.75 41.25 19480107 -1 -1 -97.75 41.25 19480108 -1 -1 -97.75 41.25 19480109 -1 -1 -97.75 41.25 19480110 -1 -1 -97.75 41.25 19480111 -1 -1 -97.75 41.25 19480112 -1 -1 -97.75 41.25 19480113 0 0 253185 17 -97.75 41.25 19480114 -1 -1 -97.75 41.25 19480115 -1 -1 -97.75 41.25 19480116 -1 -1 -97.75 41.25 19480117 -1 -1 -97.75 41.25 19480118 -1 -1 -97.75 41.25 19480119 -1 -1 -97.75 41.25 19480120 -1 -1 -97.75 41.25 19480121 -1 -1 -97.75 41.25 19480122 51 51 253185 17 -97.75 41.25 19480123 -1 -1 -97.75 41.25 19480124 0 0 253185 17 -97.75 41.25 19480125 -1 -1 -97.75 41.25 19480126 -1 -1 -97.75 41.25 19480127 -1 -1 -97.75 41.25 19480128 -1 -1 -97.75 41.25 19480129 -1 -1 -97.75 41.25 19480130 -1 -1 -97.75 41.25 19480131 -1 -1 -97.75 41.25 19480201 -1 -1 -97.75 41.25 19480202 -1 -1 -97.75 41.25 19480203 -1 -1 -97.75 41.25 19480204 -1 -1 -97.75 41.25 19480205 -1 -1 -97.75 41.25 19480206 -1 -1 -97.75 41.25 19480207 -1 -1 -97.75 41.25 19480208 -1 -1 -97.75 41.25 19480209 -1 -1 -97.75 41.25 19480210 76 76 253075 08 /Stn. Fullerton/ -97.75 41.25 19480211 51 51 253075 08 -97.75 41.25 19480212 -1 -1 -97.75 41.25 19480213 51 51 253075 08 -97.75 41.25 19480214 -1 -1 -97.75 41.25 19480215 -1 -1 -97.75 41.25 19480216 -1 -1 -97.75 41.25 19480217 -1 -1 -97.75 41.25 19480218 -1 -1 -97.75 41.25 19480219 -1 -1 -97.75 41.25 19480220 -1 -1 -97.75 41.25 19480221 -1 -1 -97.75 41.25 19480222 -1 -1 -97.75 41.25 19480223 -1 -1 -97.75 41.25 19480224 25 25 253075 08 -97.75 41.25 19480225 0 0 253075 08 -97.75 41.25 19480226 -1 -1 -97.75 41.25 19480227 -1 -1 -97.75 41.25 19480228 -1 -1 -97.75 41.25 19480229 -1 -1 -97.75 41.25 19480301 -1 -1 -97.75 41.25 19480302 -1 -1 …….. 2. The data that come from the Current Data Rescue Data Set (prior to August 1948) are considered preliminary because they still are undergoing extensive quality control procedures. In our datum selection process this Data Set has a lowest priority and contributed only when all other four data sets had no datum at a given station at a given day. Generally, the user will be much ‘better off” using the 1948 data only as a supplement to the 1949 hydrological year (October 1948 to September 1949). 1