search

nansencenter / openwind

A python package for estimating high resolution wind from SAR images
41 stars 25 forks source link

Use Mouche2005 polarisation ratio instead of Thompson1998 #15

Open knutfrode opened 9 years ago

knutfrode commented 9 years ago

Making wind climatology from many VV and HH images, I find that HH-wind is clearly too high, by almost 2 m/s in average(!)

This is probably due to limitation of the simple Thompson1998 polarisation ratio (though cannot ignore the possibility of a bug somewhere). Note that that polarisation has been developed for Radarsat data, and pol ratio is probably sensor specific in practice (though it should not in theory).

The polarisation ratio function of Mouche2005 (below) should be used/tested instead, as this is newer, and is developed for ASAR data. This one depends on wind direction also, so it can only be implemented after wind direction information has been added to the object.

  theta=inc_angle
   phi=wind_dir
   A0 = 0.00650704
   B0 = 0.128983
   C0 = 0.992839
   Api2 = 0.00782194
   Bpi2 = 0.121405
   Cpi2 = 0.992839
   Api = 0.00598416
   Bpi = 0.140952
   Cpi = 0.992885

   P0_theta = A0*exp(B0*theta)+C0
   Ppi2_theta = Api2*exp(Bpi2*theta)+Cpi2
   Ppi_theta = Api*exp(Bpi*theta)+Cpi

   C0_theta = (P0_theta+Ppi_theta+2*Ppi2_theta)/4
   C1_theta = (P0_theta-Ppi_theta)/2
   C2_theta = (P0_theta+Ppi_theta-2*Ppi2_theta)/4

   pr = C0_theta + C1_theta*cos(phi*!dtor) + $
            C2_theta*cos(2*phi*!dtor)