For studying of air temperature sets and precipitations instead of spars data from meteorological stations often use more accessible climatic archives, where primary observations are processed and girded. They are especially convenient in areas, where a meteorological stations network are rare and non-uniform. Various methods of spatial generalization, from interpolation to complex reanalysis are used for construction of net archives. Results have specific features and describe a climate of separate regions with different accuracy. Next four air temperature and precipitations archives gained universal currency.

The School of Environmental Sciences, University of East Anglia, UK Climatic Research Unit data set (CRU TS 2.1) is composed of monthly grids of observed climatic characteristics for the period 1901–2002 and covers the global land surface at 0.5 degree resolution. Nine climate variables are available: mean, minimum, and maximum temperature, diurnal temperature range, precipitation, wet day frequency, frost day frequency, vapor pressure, and cloud cover. The data set was produced by interpolating station anomalies of climatic parameters relative to the 1960–1991 normals as a function of latitude, longitude, and elevation using thin-plate splines. The anomaly grids were combined with the gridded 1961–1990 normals (CRU CL 1.0) to obtain absolute values. The CRU TS 2.1 data set is publicly available.



The Willmott and Matsuura Arctic climate data set (W&M) was developed in the Department of Geography, University of Delaware. Monthly mean air temperatures and precipitation calculated for 4517 land surface weather stations located north of 43°N were used to produce a gridded archive with 0.5° latitude/longitude resolution. Traditional interpolation was accomplished with the spherical version of Shepard’s algorithm, which employs an enhanced distance-weighting method. To increase the accuracy of spatial air temperature fields Digital Elevation Model (DEM)-assisted interpolation of air temperature was employed. Station air temperature was first translated to sea level using an average environmental lapse rate (6.0°C km-1). Traditional interpolation then was performed on the adjusted-to-sea-level station air temperatures. Finally, the gridded sea level air temperatures were translated back to DEM-grid height, again at the average environmental lapse rate. The accuracy of this interpolation exercise was assessed by station-bystation cross validation. The resulting data set provides monthly means of air temperature and precipitation for the 1930–2004 period.



The original ERA-40 2 m air temperature output contains temperature at 6-hour intervals in a reduced Gaussian grid (N80) with approximate resolution of 125 km and covers the period September 1957 through August 2002. A broad overview of ERA-40 can be found in the extended abstracts from the 2001 ERA-40 workshop and the ECMWF. The 6-hour air temperatures are averaged arithmetically into daily and monthly values. ERA-40 air temperature from forecasts for the 2 m level are postprocessed, interpolated between the lowest model level and the surface, and assimilated with ground-based measurements. In this sense, ERA-40 air temperature is an analyzed field produced by the atmospheric reanalysis. However, the assimilated air temperature is not used as an initial condition for the forecast in the next time step. The primary ERA-40 archives also provide 6-hour accumulated precipitation fields in the N80 grid. ERA-40 precipitation data were processed into monthly precipitation sums by year. We used total precipitation, which is the sum of the large-scale and convective components. ERA-40 precipitation is a modeled (or forecast) field not influenced by ground-based measurements. The ERA-40 reanalysis data are available from the ECMWF .



The NCEP-1 reanalysis starts in 1948 and is updated continuously. The NCEP-1 system, based on a T62 Gaussian grid with a spatial resolution of about 1.875°x1.875° and 28 vertical sigma levels. The NCEP-1 2 m air temperature is a standard modeled field, produced by linear interpolation between the surface skin temperature and the temperature at the lowest model sigma level (0.995). It is therefore influenced strongly by the modeled surface energy budget. Air temperature at the lowest sigma (0.995) level is not influenced as strongly by the modeled surface energy balance and should provide for relatively realistic temperature variability. Unlike the ERA-40 2 m air temperature, the NCEP-1 2 m air temperature is not assimilated with ground-based measurements. NCEP-1 precipitation is also a standard modeled field of the NCEP reanalysis.



The archives received by interpolation of measurements

The archives based on modelling reanalysis

It is possible to estimate quality of these archives, having compared them to measurements on 455 Russian and contiguous states meteorological stations. The Greatest interest is represented by(with) an estimation of accuracy of several temperature characteristics (mean annual air temperature, annual temperature amplitude and thawing degree-days) and precipitations as the permafrost condition to the greatest degree depends on their size

The difference of mid-annual air temperature calculated according to archives and observation, are distributed almost inversive (on a sign) at CRU and W&M archives in the European territory of Russia and in Western Siberia. As differences from observation in these areas are great enough, at many stations exceeding 2ºC, archive CRU mainly overestimates (positive differences), and W&M - underestimates (negative differences) temperature. On all east part of Russia both archives give close estimations which it is appreciable below the given observation. As a whole the standard error σT on 455 stations at archive W&M is much less, than at CRU archive (1.36 and 2.14, accordingly).

The archives based on reanalysis, yield closer results and are well agreed with observation on considerable part of Russian European territory, and is slightly worse in Western Siberia. In east areas of Russia both archives overestimate temperature at the average more than by 2 ºC. Furthermore, ERA-40 gives the smaller standard error, than NCEP-1 (1.48 and 1.90, accordingly) though concedes on this indicator to W&M archive. If to take into consideration that ERA-40 gives as well shifted air temperature estimate on all territory of Russia (-0.04 ºC), this archive should accept as the most admissible for solution of the problem connected with the analysis of temperature characteristics.

The archives based on reanalysis, overestimate Russian annual precipitations in whole, moreover archive NCEP-1 on rather considerable size (188 mm/year). Thereby on the most part of considered territory annual norms of precipitations are 300 – 600 mm range, the standard error of these archives are, accordingly, 151 and 237 mm/year. It is indicative that both average shift (bias), and the standard error slightly exceed predicted values of changes of precipitations on the middle 21 centuries. Against this background, two archives received by interpolation of observations have quite comprehensible accuracy though also they in some regions give considerable bias of precipitations.

webarticletitleimg: