I'm having a problem with the position of the equations and the image.
how come the second and third equations have an indentation and are not aligned? (in particular the second one). I think that by observing them they are not very pleasant to look at
how come the figure is so detached?
thanks.

documentclass[a4paper,10pt]article
usepackage[utf8]inputenc%
usepackage[T1]fontenc
usepackagebabel
usepackage[demo]graphicx
usepackagecaption
usepackagemathtools

%usepackageempheq
usepackageshowlabels %%<----
newcommandde[1]mathrmd#1
newcommandFra[2]displaystyle frac #1 #2
 beginminipage[h].65textwidth
 In un textitP.B.L., in quiete, la forza di gravità bilancia esattamente la forza dovuta al gradiente verticale di pressione, vale pertanto l'equazione idrostatica dell'atmosfera, secondo la quale, la pressione decresce con la quota.
 \
 Questa relazione conserva la sua validità anche in situazioni reali, ossia non statistiche: 
 beginsubequations
 beginalign
 dep = -g,rhodezlabeleq086qquad
 textparbox4cmEquazione Idrostatica\
 mathbfp = rho,R,T = FraR,TVlabeleq086.1\
 rho= FramathbfpR,Tlabeleq086.2qquad
 textparbox4cmEquazione di Stato 
 endalign
 endsubequations
 endminipagehfill
 %
 beginminipage[h].35textheight
 centeringraiseboxdimexpr topskip-height%

 includegraphics[width=0.4textwidth]particella/adiabatica
 captionoffigure\Andamento della densità\con la quota
 labelfig:adiabatica
 endminipage
 vspace1cm

 Combinando le due equazioni, si ottiene : 
 %
 beginequation
 setlengthfboxsep0cm
 setlengthfboxrule0pt
 dep = -g,FramathbfpR,Tdez,Rightarrow, Fradepdez= -g,FratextbfpR,Tlabeleq086.3 \
 endequation

-------------------UPDATE-------------------

I tried to write the code using wrapfigure as suggested by GuM. I hope it is correct or needs a review?

beginminipage[h].95textwidth

 beginwrapfigure[25]r3.5cm
 includegraphics[width=3.5cm]adiabatica
 captionoffigureAndamento della densit`a con la quota
 labelfig:adiabatica
 endwrapfigure%hfill 

 In un textitP.B.L., in quiete, la forza di gravità bilancia esattamente la forza dovuta al gradiente verticale di pressione, vale pertanto l'equazione idrostatica dell'atmosfera, secondo la quale, la pressione decresce con la quota.\
 Questa relazione conserva la sua validità anche in situazioni reali, ossia non statistiche: 

 beginsubequations
 beginalign
 dep &= -g,rhodez
 && textEquaz. Idrostatica 
 labeleq086 \
 mathbfp &= rho RT = FraRTVlabeleq086.1 \
 rho &= FramathbfpRTlabeleq086.2
 && textEquaz. di Stato
 endalign
 endsubequations
endminipagequadhfill
bigskipbigskipbigskip

Combinando le due equazioni, si ottiene: 
%
beginequation
setlengthfboxsep0cm
setlengthfboxrule0pt
dep = -g,FramathbfpR,Tdez,Rightarrow, Fradepdez= -g,FratextbfpR,Tlabeleq086.3 \
endequation

You need &s to mark the alignment points (2n – 1 &s for n alignment points). Here is a possibility, with a recreated preamble. However, note that, with the chosen lengths the figure overflows into the margin.

documentclass[a4paper, italian]article
usepackage[utf8]inputenc%
usepackage[T1]fontenc
usepackagebabel
usepackage[demo]graphicx
usepackagecaption
 usepackagemathtools
newcommandde[1]mathrmd#1
letFra=frac

begindocument

beginminipage[h].65textwidth
In un textitP.B.L., in quiete, la forza di gravità bilancia esattamente la forza dovuta al gradiente verticale di pressione, vale pertanto l'equazione idrostatica dell'atmosfera, secondo la quale, la pressione decresce con la quota.
\
Questa relazione conserva la sua validità anche in situazioni reali, ossia non statistiche:
beginsubequations
 beginalign
 dep & = -g,rhodezlabeleq086 & & textEquazione Idrostatica\
 mathbfp & = rho,R,T= mathrlapFraR,TVlabeleq086.1\
 rho & = FramathbfpR,Tlabeleq086.2 & & textEquazione di Stato
 endalign
endsubequations
endminipagehfill
%
beginminipage[h].35textheight
 centeringraiseboxdimexpr topskip-height%
 includegraphics[width=0.4textwidth]particella/adiabatica
 captionoffigure\Andamento della densità\con la quota
labelfig:adiabatica
endminipage
vspace1cm

Combinando le due equazioni, si ottiene :
%
beginequation
 setlengthfboxsep0cm
 setlengthfboxrule0pt
 dep = -g,FramathbfpR,Tdez,Rightarrow, Fradepdez= -g,FratextbfpR,Tlabeleq086.3 \
endequation

enddocument 

Have a look at this (same remarks as Bernard’s):

% My standard header for TeX.SX answers:
documentclass[a4paper]article % To avoid confusion, let us explicitly 
 % declare the paper format.

usepackage[T1]fontenc % Not always necessary, but recommended.
% End of standard header. What follows pertains to the problem at hand.

usepackage[italian]babel
usepackageamsmath
% usepackagecaption
usepackagecapt-of

usepackagemwe % for this MWE only

newcommand*de[1]mathop!mathrmd#1 % stub
newcommand*Fra[2]frac#1#2 % ???



begindocument

noindentbeginminipage[c].6textwidth
In un textitP.B.L., in quiete, la forza di gravit`a bilancia esattamente
la forza dovuta al gradiente verticale di pressione, vale pertanto l'equazione
idrostatica dell'atmosfera, secondo la quale, la pressione decresce con la
quota.

Questa relazione conserva la sua validit`a anche in situazioni reali, ossia
non statistiche:
 beginsubequations
 beginalign
 dep &= -g,rhodez
 && textEquaz. Idrostatica 
 labeleq086 \
 mathbfp &= rho RT = FraRTVlabeleq086.1 \
 rho &= FramathbfpRTlabeleq086.2
 && textEquaz. di Stato
 endalign
 endsubequations
endminipagequadhfill
%
beginminipage[c].35textwidth
 centering
 includegraphics[width=0.8linewidth]image
 captionoffigureAndamento della densit`a con la quota
 labelfig:adiabatica
endminipage

bigskip

Combinando le due equazioni, si ottiene: 
%
beginequation
 setlengthfboxsep0cm
 setlengthfboxrule0pt
 dep = -g,FramathbfpR,Tdez,Rightarrow, Fradepdez= -g,FratextbfpR,Tlabeleq086.3 \
endequation

enddocument

Here’s the output:

Note the following points, among the others:

• I use the [c] position specifier to align the two minipages
  at their vertical center, without having to tamper with the
  vertical space inside the rightmost one;




• I do not use \ to separate paragraphs inside the minipages;




• I use & and && inside the align environment,
  as Bernard has already taught
  you;




• I replaced textheight with textwidth in the “width”
  argument of the second minipage environment (!!!);




• finally, I added noindent before the first minipage
  (I had previously overlooked this!).

However, you are trying to bypass (La)TeX pagination methods by hand in a way I deem incorrect; are you aware of the wrapfig package?

Update

Here’s an example of how you could use the wrapfigure environment (requires the wrapfig package). Note that, owing to the unusual height of the group of equations, it is advisable to specify explicitly the number of “virtual lines” in the narrower part: this is done via the first optional argument of the wrapfigure environment. In this case, a value of 10 seems adequate. Recall that equations are assumed to take up the equivalent of three lines.

% My standard header for TeX.SX answers:
documentclass[a4paper]article % To avoid confusion, let us explicitly 
 % declare the paper format.

usepackage[T1]fontenc % Not always necessary, but recommended.
% End of standard header. What follows pertains to the problem at hand.

usepackage[italian]babel
usepackageamsmath
% usepackagecaption
usepackagecapt-of
usepackagewrapfig

usepackagemwe % for this MWE only

newcommand*de[1]mathop!mathrmd#1 % stub
newcommand*Fra[2]frac#1#2 % ???



begindocument

Testo che precede la parte nella quale si vuole inserire la figura a margine.
Si assume che la figura debba stare nel margine emphesterno
(( texttto=textitouter )).

beginwrapfigure[10]o.35textwidth
 centering
 includegraphics[width=0.8linewidth]image
 captionoffigureAndamento della densit`a con la quota
 labelfig:adiabatica
endwrapfigure

In un textitP.B.L., in quiete, la forza di gravit`a bilancia esattamente
la forza dovuta al gradiente verticale di pressione, vale pertanto l'equazione
idrostatica dell'atmosfera, secondo la quale, la pressione decresce con la
quota.

Questa relazione conserva la sua validit`a anche in situazioni reali, ossia
non statistiche:
beginsubequations
 beginalign
 dep &= -grhodez
 && textEquaz. Idrostatica 
 labeleq086 \
 mathbfp &= rho RT = FraRTVlabeleq086.1 \
 rho &= FramathbfpRTlabeleq086.2
 && textEquaz. di Stato
 endalign
endsubequations

Combinando le due equazioni, si ottiene
(aggiungo questo commento al solo scopo
di dimostrare che siamo tornati alla giustezza piena):
%
beginequation
% setlengthfboxsep0cm % ???
% setlengthfboxrule0pt
 dep = -gFramathbfpRTdez
 quadLongrightarrowquad
 Fradepdez= -gFramathbfpRT
 labeleq086.3
 % \ % ??? (!!!)
endequation

enddocument

Note also that it is a very bad idea to use labels like eq086.3: always use symbolic labels.

The output for the second example follows: