We employ an ensemble of 24 hydrodynamic cluster simulations to create spatially and spectrally resolved images of quality comparable to Chandra ’ s expected performance .
Emission from simulation mass elements is represented using the XSPEC m ekal program assuming 0.3 solar metallicity and the resulting spectra are fit with a single-temperature model .
Despite significant departures from isothermality in the cluster gas , single-temperature models produce acceptable fits to 20,000 source photon spectra .
The spectral fit temperature T _ { s } is generally lower than the mass weighted average temperature T _ { m } due to the influence of soft line emission from cooler gas being accreted as part of the hierarchical clustering process .

The nature of this deviation depends on the bandpass used for spectral fitting .
In a Chandra -like bandpass of 0.5 to 9.5 keV we find a nearly uniform fractional bias of ( T _ { m } - T _ { s } ) / T _ { s } \simeq 20 \% , although smaller clusters sometimes demonstrate much greater deviations .
If the minimum energy threshold is raised to 2 keV , however , the effect of line emission on the spectrum is greatly decreased and T _ { s } becomes a nearly unbiased estimator of T _ { m } for smaller clusters .
The fractional deviation in T _ { s } relative to T _ { m } is scale-dependent in this bandpass and follows the approximate relation ( T _ { m } - T _ { s } ) / T _ { s } = 0.2 \log _ { 10 } T _ { m } .
This results in an observed M _ { ICM } – T _ { s } relationship for the simulations with slope of about 1.6 , intermediate between the virial relation M \propto T _ { m } ^ { 3 / 2 } and the observed relation M _ { ICM } \propto T ^ { 2 } .

Tracking each cluster in the ensemble at 16 epochs in its evolutionary history , we catalogue merger events with mass ratios exceeding 10 % in order to investigate the relationship between spectral temperature and proximity to a major merger event .
Clusters that are very cool relative to the mean mass-temperature relationship lie preferentially close to a major merger , suggesting a viable observational method to cull a subset of dynamically young clusters from the general population .