1 Introduction

Observations show that supernova remnants (SNRs) have anisotropic distributions of surface brightness (Seward 1990; Whiteoak & Green 1996). There are four morphological classes of SNRs: shell-like, Crab-like (plerionic), composite and thermal X-ray composites (TXCs; or mixed-morphology, or centrally-influenced) (Jones et al. 1998; Rho & Petre 1998, hereafter RP98). In the past few years interest in TXCs has risen (e.g., Jones et al. 1998; Cox et al. 1999; Shelton et al. 1999; Sun et al. 1999). TXCs are SNRs with centrally concentrated thermal X-ray and limb brightened radio morphologies. Remnants W44, W28, 3C 400.2, Kes 27, 3C 391, CTB 1, MSH 11-61A and others represent a mixed-morphology class (RP98). Since the publication of RP98, a number of new SNRs have been discovered, others have been observed more precisely or even for the first time in X-rays. Thus, new candidates to the TXC class have appeared in the past two years.

Two physical models have been presented so far to explain TXC (see RP98 for review). One of them is an enhanced interior X-ray emission from the evaporated material of numerous swept-up clouds which increases density in the central region of SNR. This model is frequently used; sometimes its application is intrinsically inconsistent, e.g., as in MSH 11-61A where evaporation timescales exceed the age of SNR 50-100 times (Jones et al. 1998). In the second model, shock temperature is small due to essential cooling; very soft emission of the shell is absorbed by interstellar medium (ISM) and only the interior region remains visible. In this model, thermal conduction may level temperature profiles and increase the central density altering the interior structure (Cox et al. 1999).

Other possibilities are also noted in publications. They are: a) emission from ejecta, b) differential absorption (Long et al. 1991, have concluded that the centrally-peaked X-ray morphology within a radio shell is unlikely the result of absorption alone because distributions of X-ray and radio emitting plasmas have to be different in this case) or c) explosion in a medium with centrally-concentrated density distribution $\rho(r)\propto r^{-m}$, m>0 (such a medium does not give centrally-concentrated morphology (Long et al. 1991) because the Sedov (1959) solutions give a specific internal profile of the flow density for all $m\leq 2$ ( $\gamma=5/3$), other m develop a cavity around the centre).

The mentioned models are used to obtain a centrally-filled morphology within the framework of one-dimensional (1-D) hydrodynamic approaches. When we proceed to 2-D or 3-D models, we note that a simple projection effect may cause the shell-like SNR to fall into another morphology class, namely, centrally-influenced (Hnatyk & Petruk 1999). The main feature of such SNR is the thermal X-rays emitted from swept-up gas and peaked in the internal part of the projection. Therefore, getting beyond one dimension, we obtain a new possibility to explain the nature of TXC. Such a possibility is considered in this paper.

Hnatyk & Petruk (1998) have reproduced the X-ray morphology of IC 443 with the proposed projection model of TXC. Therefore, we restrict ourselves to theoretical consideration of the phenomenon.

SNRs are modelled with an approximate analytic method for hydrodynamic description of the adiabatic phase of an asymmetrical point explosion in an arbitrary large-scale nonuniform medium (Hnatyk & Petruk 1999). Equilibrium thermal X-ray emission is calculated with the use of the Raymond & Smith (1977) model. The model for radio emission of nonspherical SNR is based on Reynolds & Chevalier (1981) and Reynolds (1998) and is described in Sect. 3.2.