Dilution of precision, in brief
DOP describes how satellite geometry amplifies a normalized observation uncertainty. PDOP concerns three-dimensional position; HDOP and VDOP require the spatial covariance in local east, north and up axes. DOP is not a measured position error or a standalone accuracy guarantee.
Learning objectives
- Connect DOP to the inverse normal matrix.
- Read DOP values from covariance diagonal terms.
- Interpret good and poor geometry numerically.
DOP is geometry amplification
Measurement noise is not the only driver of final position accuracy. Satellite geometry determines how strongly that noise is amplified into the estimated position and receiver clock.
In least squares, that amplification appears in the covariance-like matrix obtained from the design matrix.
Core definition
Horizontal and vertical require a local frame
If the design matrix uses ECEF X, Y, Z, rotate its 3 × 3 position block into local east, north, up before computing HDOP or VDOP. ECEF Z is not the receiver's local vertical. PDOP is unchanged by this orthogonal rotation.
Let E contain local east, north, up unit vectors as rows expressed in ECEF. Then Q_ENU = E Q_XYZ Eᵀ. For a design matrix already in ENU, no additional rotation is needed.
DOP formulas
Step 1
Read a covariance diagonal
For a compact illustrative example, assume the geometry matrix is already expressed in local east, north, up and clock range, with the following diagonal entries in its inverse normal matrix. These numbers are supplied teaching inputs rather than a satellite-derived epoch trace.
Step 2
Compute the DOP values
Use the square root of the relevant diagonal sums.
Interpreting values
| DOP range | Geometry quality | Meaning |
|---|---|---|
| 1 to 2 | Excellent | Satellites are well spread. |
| 2 to 5 | Good | Usable for most routine positioning. |
| 5 to 10 | Moderate | Errors are visibly amplified. |
| 10 to 20 | Poor | Position quality warning. |
| > 20 | Very poor | Positioning is unreliable. |
Common mistake
DOP is dimensionless. If your pseudorange standard deviation is 1.5 m and PDOP is 2.72, the rough 3D position precision estimate is 4.08 m, not 2.72 m.
What you should understand now
- DOP comes from geometry, not from atmospheric modeling.
- The same design matrix used in SPP also gives DOP after convergence.
- Good satellite spread lowers the diagonal terms of the inverse normal matrix.
Related tutorials
Geometry and quality
ECEF to ENU: Coordinates and Covariance
A reproducible ECEF to ENU numerical example: subtract the local origin, construct the rotation, propagate covariance and interpret horizontal and vertical uncertainty.
Positioning algorithms
SPP from Scratch
Compute GPS single point positioning from RINEX observations: transmission time, satellite clocks, broadcast orbits, Sagnac and atmospheric corrections, iterative WLS, residuals and covariance.
Corrections
Klobuchar Ionospheric Correction
A numerical Klobuchar walkthrough from receiver/satellite geometry to L1 delay in meters.
Help improve this tutorial
Point out numerical ambiguity, missing prerequisites, or a step that needs a fuller derivation.