Szanowny Panie Andrzeju,

Przeslalismy na:

ftp://sirius.astrouw.edu.pl/pub/submit/noise/

poprawiona prace "Propagation..." wraz z rysunkami.


Prace zmienilismy w znacznym zakresie zgodnie z uwagami Recenzenta.
Przepraszamy za opoznienie.

Odnosząc sie do konkretnych uwag Recenzenta:

Major comments:

1/The paper sufferes from imbalance in construction.  The authors study
in detail statistical performance of ANDRIL reconstruction after
smoothing by averaging of results from different exposures (4 pages).
Then on p.10 they argue convicingly, that this is not the best approach
for dynamically evolving solar flares. They argue in favour of the
artificial background shift, instead. However, in the text scant
attention is payed to the latter algorithm (1 page). This should be
cured by shortenning of Sect. 2 and 3, and accompanied by a paragraph in
Sect. 5 discussing when, if ever, multiple perturbation approach shoud
be useful. 

	Zmieniliśmy organizację tekstu. Skrocilismy i przeredagowalismy  
	rozdzialy. Wprowadzilismy  parametr "q" charakteryzujacy jakosc 
	dopasowania. Dodalismy tabele wartosci tego parametru dla 
	przeprowadzonych symulacji.
	
2/NEF factor for ANDRIL without subsampling amounts 2. It would be
informative to state what actually is gained at this price. At the price
of further rise of NEF by a factor of 3 or so for subsampling one
obtaines spatial resolution improved bya a corresponding factor.
	
	Wyjasnilismy co zyskujemy prowadzac dekonwolucje 1:1 - 
	mianowicie - wierniejsze odtworzenie obrazu względnych jasności 
	w poszczegolnych elementach obrazu. 

3/Not all figures are required in my opinion, one of 3 diagrams in Fig.
3 and 4 would be enough. It is quite obvious that statistics in the
reconstructed image follows Poisson  distribution for sufficiently large
S/N. It would be enough to state in the text clearly at what S/N level
this is attained.

	Usunelismy jeden z rysunkow i odpowiednio zmieniliśmy tekst.


4a/Perturbation procedure is ill defined in text: It should be clearly
stated that averaging of  ANDRIL reconstruction of several exposures is
proposed to decrease noise and reconstruction artefacts.

	Zmieniliśmy tytuły i zawartość rozdzialów.

4b/The name multiple perturbation (p3) is misleading.  True, in
simulations one resorts to pertutrbations of the simulated image.
However, if I understood correctly the discussion on p.10, for the real
data one takes several exposures of the same field. So one should not
speak of perturbations but instead on smoothing by averaging out.

	Zmieniliśmy odpowiednio tytuł i zawartość rozd. 3. W aktualnej 
	wersji nie uzywamy kwestionowanej nazwy.

Minor points

5/The title is lengthy and uninformative, I propose something shorter
like 'Reconstruction of Images with Poisson Noise' or similar.

	Tytul zmieniliśmy wg sugestii.

6/Hasinger at al. (~1988) studied Maximum Likelihood  estimation of
source properties in presence od poissonian noise in ROSAT observations.
Is this reference relevant here?

	Dotarliśmy do publikacji, którą zapewne mial na mysli 
	Recenzent (CRUDDACE, R. G.; HASINGER, G. R.; SCHMITT, J. H., The 
	application of a maximum likelihood analysis to detection of 
	sources in the Rosat data base). Jej cytowanie uznaliśmy za 
	pozyteczne.

7/For a self-standing paper, symbols in Eq. (1) require definitions.

	Powtórzyliśmy opis symboli wg. Pracy I.

8/ S/N symbol on bottom of p.5 is corrupt

	Dokonaliśmy korekty.
	

Pozdrawiamy,

Barbara Sylwester