def interpolate( self, in_lat, in_lon ):
'''Interpolation method of wind at specified position.
Uses bilinear aproximation between the vectors.
Proportion is linear scaling of angle and linear scaling of length
'''
in_lat = norm_lat(in_lat)
in_lon = norm_lon(in_lon)
lat_ind = in_lat // self.step_la
if in_lat == 90:
lat_ind -= 1
lon_ind = in_lon // self.step_lo
if in_lon == 360:
lon_ind = 0
left = lon_ind * self.step_lo
right = ( lon_ind + 1 ) * self.step_lo
bottom = lat_ind * self.step_la
top = ( lat_ind + 1 ) * self.step_la
LB = self.wind_grid[bottom][left]
LT = self.wind_grid[top][left]
RB = self.wind_grid[bottom][norm_lon(right)]
RT = self.wind_grid[top][norm_lon(right)]
top_vec = self.proport( LT, RT, (in_lon - left) / (right - left) )
bot_vec = self.proport( LB, RB, (in_lon - left) / (right - left) )
return self.proport( bot_vec, top_vec, \
(in_lat - bottom) / (top - bottom) )
def phi_theta( x, y, z ):
'''Converts x, y, z tripplet into phi, theta normalizing output on the fly
'''
r = sqrt( x*x + y*y + z*z )
phi = norm_lon( atan2( y, x )*ratioPI )
theta = norm_lat( acos( z / r )*ratioPI - 90.0 )
return ( phi, theta )
def interpolate_naive( self, in_lat, in_lon ):
'''A bit stupid methon
Resulting vector is proportional to how it`s close to \
corner base grid vectors
'''
in_lat = norm_lat(in_lat)
in_lon = norm_lon(in_lon)
lat_ind = in_lat // self.step_la
if in_lat == 90:
lat_ind -= 1
lon_ind = in_lon // self.step_lo
if in_lon == 360:
lon_ind = 0
left = lon_ind * self.step_lo
right = ( lon_ind + 1 ) * self.step_lo
bottom = lat_ind * self.step_la
top = ( lat_ind + 1 ) * self.step_la
LB = self.wind_grid[bottom][left]
LT = self.wind_grid[top][left]
RB = self.wind_grid[bottom][norm_lon(right)]
RT = self.wind_grid[top][norm_lon(right)]
ltx = math.sqrt( \
(in_lat-top)*(in_lat-top) + (in_lon-left)*(in_lon-left) )
lbx = math.sqrt( \
(in_lat-bottom)*(in_lat-bottom) + (in_lon-left)*(in_lon-left) )
rtx = math.sqrt( \
(in_lat-top)*(in_lat-top) + (in_lon-right)*(in_lon-right) )
rbx = math.sqrt( \
(in_lat-bottom)*(in_lat-bottom) + (in_lon-right)*(in_lon-right) )
norma = ltx + lbx + rtx + rbx
return ( ((norma-ltx)*LT[0]/norma + (norma-lbx)*LB[0]/norma + \
(norma-rtx)*RT[0]/norma + (norma-rbx)*RB[0]/norma)/3 , \
((norma-ltx)*LT[1]/norma + (norma-lbx)*LB[1]/norma + \
(norma-rtx)*RT[1]/norma + (norma-rbx)*RB[1]/norma)/3 )
def points_gen( self, psi, units=DEGREES, verbose = True ):
'''Genertion of the points randomly distributed with *average*
resolution of psi.
Arguments:
psi (float) -- resolution
units (DEGREES/RADIANS) -- how psi is set (default: DEGREES)
'''
k = 0.5 / pi
p0 = k * self.domain.phi[0]
p1 = k * self.domain.phi[1]
t0 = 0.5* ( 1 + cos( self.domain.theta[1] ) )
t1 = 0.5* ( 1 + cos( self.domain.theta[0] ) )
N = self.domain.point_count( psi, units )
if verbose:
print("points amount to generate " + str(N))
self.points_clear()
# arccos method:
for j in range( N ):
U = random.uniform( p0, p1 )
V = random.uniform( t0, t1 )
phi = 2*pi* U
theta = acos( 2*V - 1 )
x,y,z = geocoords.xyz_spherical( theta, phi )
self.points.append( (x,y,z) )
self.points_angular.append( ( norm_lon( phi*ratioPI), \
norm_lat( theta*ratioPI) ) )
return