Atmospheric Water Vapor Transport and Precipitation

Revisiting forest impact on

atmospheric water vapor transport and precipitation

Anastassia M. Makarieva1*, Victor G. Gorshkov1 and Bai-Lian Li2

1Theoretical Physics Division, Petersburg Nuclear Physics Institute, 188300, Gatchina, St.Petersburg, Russia

2XIEG-UCR International Center for Arid Land Ecology, University of California, Riverside, CA, 92521, USA

*Corresponding author: Anastassia M. Makarieva, e-mail: , fax: +7813713-19-63; tel: +7813714-60-96

Online Resource

1. Data

1.1 Precipitation data

Precipitation data set of Legates and Wilmott (1990) (LW90) was downloaded from http://disc.sci.gsfc.nasa.gov/precipitation/documentation/readme_html/legates_gauge_precip_readme.shtml

CCML precipitation data set (McGuire et al. 2001) was downloaded from http://eos-webster.sr.unh.edu (To quickly do so, the reader is advised to apply the following search criteria: Collection: Carbon Cycle Model Linkage-CCMLP CCMLP-Input; Spatial: Category: Continents; Location: Africa, Asia, Australia and Oceania, Europe, North America, South America; Temporal: not specified; Keyword: not specified; Freetext: precipitation.)

LW90 data are arranged as a 361´721 element array, with each 0.5´0.5 degree cell centered at the nodes of the grid. E.g., the data at 60.5°N are for grid cells with coordinates (60.5,-180), (60.5,-179.5),…,(60.5,179.5),(60.5,180), where negative values are for western longitude and positive for eastern longitude, such that the first and last points coincide. Data for, say, 30°E, include the point (-90,30), i.e. the South pole. The same point is included for all other longitudes (-90,31),(-90,32) etc.

The CCML data are conceptually arranged as a 360´720 array, with each 0.5´0.5 degree cell centered at the nodes of the grid. Longitude is counted from -179.75 to 179.75, latitude is counted from 89.75 to -89.75, e.g. (60.25,-179.75), (60.25,-179.25), …,(60.25,179.75).

To make the data comparable, we removed one copy (easternmost and southernmost) of the coinciding data points from LW90 to turn it into a 360´720 array. In the resulting array, LW90 cell (60.5,-180) corresponds to CCML cell (60.25,-179.75), with the general formula (LW90 Latitude) = (CCML Latitude) + 0.25°; (LW90 Longitude) = (CCML Longitude) - 0.25°. Everywhere in the paper we give LW90 coordinates of the cells. Since we analyze large-scale precipitation patterns (over a distance of 103 km or more) this minor discrepancy in the centering of grid cells may not have introduced any noticeable bias.

CCML data set contains monthly precipitation for years from 1950 to 1980 (41 year in total), while LW90 data set contains average monthly precipitation for the period 1920-1990. CCML data were averaged over 41 years for each month to be comparable to LW90 data.

1.2 Land cover data

Data for the land cover (Friedl et al. 2010) were downloaded from the The Oak Ridge National Laboratory Distributed Active Archive Center at http://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=968 which is the IGBP Land Cover Data for the 2000-2001 time period. We used the half degree data set, which is a 360´720 element array with a total of 259200 cells. The original land cover data are arranged in 17 classes, which we grouped into four (Fig. 1):

1.3 Mean elevation data

Data on mean elevation in half degree grid cells (Verdin 2001), also a 360´720 array, was downloaded from http://daac.ornl.gov/ISLSCP_II/guides/hydro1k_elevation_xdeg.html

1.4 Runoff data

Runoff data were taken from the UNH, GRDC, Global Composite Runoff Data Set (v1.0). Data were downloaded from “A Global Change Master Directory Portal for the Global Terrestrial Observing System” using the following link:

http://gcmd.nasa.gov/KeywordSearch/Metadata.do?Portal=gtos&KeywordPath=%5BFreetext%3D%27DIF%2FIDN_Node%3AGOSIC%2FGTOS+AND+DIF%2FKeyword%3AWater+discharge%27%5D&OrigMetadataNode=GCMD&EntryId=UNH_GRDC_GCRDS&MetadataView=Full&MetadataType=0&lbnode=mdlb2

The runoff data page is here: http://www.grdc.sr.unh.edu/html/Data/index.html

Data represent a 0.5´0.5 ° latitude/longitude grid with missing values in quite a lot of cells. In our analysis, Fig. 5, we only used observations (i.e., files obs_ro##.grd from runoff.zip at http://www.grdc.sr.unh.edu/html/Data/index.html) and not model reconstructions.

2. Information on precipitation transects

Table 2. Studied transects

Short names denote transects shown by white arrows in Fig. 1. Lat1, Lon1, Lat2 and Lon2 are the latitude and longitude of the beginning and the end of the transect, respectively. N is the number of 0.5°´0.5° cells in the transect. Length (thousand km) is the length of the transect in thousand km. “Tundra” and “forest” refer to the tundra and forest parts of the meridional boreal transects.

1While the data of LW90 cover the whole planet, CCML data are for land only. For this reason, in some transects studied by A11 comparison of LW90 data with CCML was made for a smaller part of the transect. In particular, CCML data do not exist for 0°S 9°E, which is the first cell in the Congo transect of A11. Therefore, this cell was not included into statistical analysis: the transect was considered from 9.5°E to 30°E. Similarly, in the Florida Panhandle transect of A11 there is no CCML data for the first cell 30°N 85°W, so the transect was considered from 30.5°N to 38°N, rather than from 30°N as in A11.

2North American transect was chosen by MGL to run along the northern border of the large forested area located in U.S.A. south to 40°N. According to the 2000-2001 land cover data of the IGBP (Friedl et al. 2010), Fig. 1, the northern border of this area lies at 37.5°N rather than at 40°N. All data in this paper are shown for the transect going along the 37.5°N parallel.

In latitudinal transects and strictly meridional (Co, Eu, No, Le, Ob), distance from the coast was counted along the parallel or the meridian, respectively, by multiplying the number of cells by a unit distance corresponding to half a degree at a given latitude or by 55.5 km along the meridian. For “slant” transects, i.e., those where both latitude and longitude changed with distance from the ocean (Am, Am2, Ma, Ye1, Ye2) for each cell distance from the ocean was calculated as the shortest distance between the beginning of the transect and the current cell. The following cells were considered for these transects:

Table 3. Grid cells considered for Am1, Am2, Ma, Ye1 and Ye2 transects

Some neighboring cells originally present in Yenisey transect of A11 (Ye1) were located very closely to each other (their distance to the ocean differed by less than 12 km), see Fig. 9 in A11. One in each pair of such twin cells was removed from the list, with the goal the cells taken into consideration in the statistical analysis to be relatively uniformly spread in space. Taking closely situated cells would improperly increase the statistical weight of precipitation values observed in those particular cells.