1 Introduction

The importance of the near-IR spectral region centered on the head of the hydrogen Paschen series and CaII triplet has already been acknowledged in literature. It offers valuable spectral classification opportunities as well as useful diagnostic tools in stellar evolution, galactic population studies and chemical abundance analysis (e.g. Andrillat et al. 1995; Jaschek & Andrillat 1998; Montes & Martin 1998; Garcia-Vargas et al. 1998; Munari 1999).

We became interested in this spectral region during an evaluation study of possible spectroscopic performances for the GAIA astrometric mission planned by ESA (Gilmore et al. 1998). The emphasis of our investigation was on the achievable accuracy of radial velocities (to provide the 6$^{\rm th}$ component of the phase-space coordinates) as well as the diagnostic capabilities in terms of spectral classification, chemical analysis, spectral peculiarities, mass loss, rotation, signatures of the interstellar extinction, etc. After a careful evaluation based on the anticipated GAIA target population (mostly F-G-K stars) and the severe technical constraints (e.g. an observable interval $\bigtriangleup \lambda \leq$ 300 Å and from 500 to 1000 pixel budget per spectrum), we went through the published observational atlases and our set of synthetic spectra and eventually selected for our study the region $\lambda\lambda$ 8500-8750 Å. In the near-IR this is the interval less affected by telluric absorptions, which is an useful attribute for compatibility with ground-based preparatory and follow-up observations (cf. Fig. 3).

  

Table 1: A list ordered by resolution of atlases covering all or part of the $\lambda\lambda$ 8500-8750 Å spectral region investigated in the present paper. Only digital atlases with a spectral resolution of 15 Å or better are listed

It soon became apparent that the anticipated goals for the GAIA mission could be best achievable if the $\lambda\lambda$ 8500-8750 Å spectra had a resolution of $\sim$0.4/0.5 Å and cover an interval of 250 Å. No spectral atlas of similar properties exists in the literature as Table 1 illustrates. All atlases have too low a resolution, with the only exception of the Montes & Martin (1998) one, which however samples a limited range of spectral types, its wavelength coverage is not continuous and the explored $\lambda$-range is quite restricted (summing up to 90 Å). It became obvious we had to proceed from scratch with new, extended observations. We performed them with the Echelle+CCD spectrograph of the Astronomical Observatories of Padova & Asiago, adopting a 0.25 Å/pix dispersion and a 0.43 Å resolution in the FWHM of the PSF sense (corresponding to a resolving power 20 000). Such a resolution nicely fills in the gap at the highest resolutions documented by Table 1 (our resolution is 3 times that of Andrillat et al. 1995 and 1/3 the one adopted by Montes & Martin 1998).

The present atlas should be of interest also to observers working with the new generation of high-resolution spectrographs now coming on-line on the largest telescopes, which include the 20 000 among their operative resolving powers (cf. Pilachowski et al. 1995).

In this paper we present an extended mapping of the Yerkes-MKK spectral classification system (Morgan et al. 1943; Morgan & Keenan 1973) through observations of 131 MKK standard stars (mainly selected from the atlas of Yamashita et al. 1977). In successive papers in this series we will deal with peculiar spectra, synthetic Kurucz spectra, chemical abundance analysis, radial and rotational velocities, spectral signatures correlated to the reddening, etc.