As it was shown by many authors, a slow decrease in X-rays observed during the decay phase of long duration flares (LDE) can be explained only by a magnetic reconnection and energy release ceaselessly ongoing in the coronal part of a flare. Using RHESSI data we try to answer two following questions. How effective are these processes at the LDEs decay phase and how can precisely the energy release rate be calculated based on these data? To answer the questions images of the selected LDEs during their decay phase were reconstructed. Physical parameters of flare coronal sources obtained from image spectral analysis allowed us to study the efficiency of the energy release process. We also examined terms included in the energy equation to find out what is the accuracy of determination of each term.

We analyse the observational characteristics for a set of Long Duration Events (LDE) using Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI). Excellent energy resolution (1 keV) of HXR images reconstructed with RHESSI allowed us to perform imaging spectroscopy of coronal loop top sources (LTS) observed for hours after maximum of the flare. We found that the sources are large structures. The diameter of an LTS is changing with time and energy. High temperature component can be observed for few dozen hours after the maximum. For some time intervals clear non-thermal component was observed in spectra, but there was no significant footpoint emission at that time. All thermal and non-thermal emissions originate from the LTS. The obtained observational characteristics were used in calculating the energy balance within LTS, which are presented in Kolomanski, 2011 (this issue).