# Wind rotation question

**URL:** <https://forum.ecmwf.int/t/wind-rotation-question/15203>\
**Category:** C3S - Datasets and Usage\
**Tags:** carra\
**Created:** [29 July 2026 14:53 UTC](https://forum.ecmwf.int/t/wind-rotation-question/15203 "2026-07-29T14:53:19Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Eva\_Hdez](https://forum.ecmwf.int/letter_avatar_proxy/v4/letter/e/58956e/32.png) [@Eva\_Hdez](https://forum.ecmwf.int/u/Eva_Hdez)\
**Post date:** [29 July 2026 14:53 UTC](https://forum.ecmwf.int/t/wind-rotation-question/15203/1 "2026-07-29T14:53:19Z")

</div>

We are interested in converting the grid-relative wind components provided in the CARRA dataset into earth-relative (true north/east) wind components, u\_earth and v\_earth. For the CARRA West domain we are using this code:

!projection parameter for CARRA west  
real, parameter :: LoV = 324.0 ! GRIB\_LoVInDegrees  
real, parameter :: Latin1 = 72.0 ! GRIB\_Latin1InDegrees  
real, parameter :: n\_cone = sin(Latin1 \* pi/180.0)

do j = 1, ny  
do i = 1, nx  
dlon = lon(i,j) - LoV  
if (dlon \> 180.0) dlon = dlon - 360.0  
if (dlon \< -180.0) dlon = dlon + 360.0

```
! local rotation of the model grid
alpha(i,j) = n_cone \* dlon

! rotate grid-relative wind components to earth-relative
angle = -alpha(i,j)\*pi/180.0
u_earth(i,j) = u(i,j)\*cos(angle) - v(i,j)\*sin(angle)
v_earth(i,j) = u(i,j)\*sin(angle) +&nbsp;v(i,j)\*cos(angle)

```

end do  
end do

We assume the wind components archived in CARRA are grid-relative (i.e., aligned with the LCC grid rather than true north/east), so we compute the grid rotation angle at each point and apply the corresponding rotation to recover the earth-relative components.

Is this assumption correct? Does this approach, including the sign convention used in the rotation, correctly convert the CARRA grid-relative winds to earth-relative wind components?

---

<div class="post-metadata">

**Author:** ![Peter\_von\_der\_Gathen](https://forum.ecmwf.int/letter_avatar_proxy/v4/letter/p/c0e974/32.png) [@Peter\_von\_der\_Gathen](https://forum.ecmwf.int/u/Peter_von_der_Gathen)\
**Post date:** [10 August 2026 15:09 UTC](https://forum.ecmwf.int/t/wind-rotation-question/15203/2 "2026-08-10T15:09:13Z")

</div>

Dear Eva,

since no one replied until now, I try an answer. However, I am a user like you. I have done that kind of calculation several years ago and I am not comparing those calculations with yours. In the mean time the CARRA1 and the CARRA2 data set provide wind speed and wind direction (relative to North). I recommend to download these two parameters instead and calculate uv-wind easily from that and avoiding all these projection issues. At least, I have done that for the CARRA2 data, now.

Best wishes

Peter

---

<div class="post-metadata">

**Author:** ![Halldor\_Bjornsson](https://forum.ecmwf.int/letter_avatar_proxy/v4/letter/h/edb3f5/32.png) [@Halldor\_Bjornsson](https://forum.ecmwf.int/u/Halldor_Bjornsson)\
**Post date:** [21 September 2026 12:15 UTC](https://forum.ecmwf.int/t/wind-rotation-question/15203/3 "2026-09-21T12:15:06Z")

</div>

The standard here is to use D ( compass direction) and F (which is a scalar) to calculate U,V. Any curvature or rotation of the grid is then automatically handled. See confluence answer

> **[ERA5: How to calculate wind speed and wind direction from u and v components...](https://confluence.ecmwf.int/spaces/CKB/pages/133262398/ERA5+How+to+calculate+wind+speed+and+wind+direction+from+u+and+v+components+of+the+wind)**