This research investigates the critical conditions enabling Unmanned Aerial Vehicle (UAV) photogrammetry to achieve large-scale mapping accuracy without ground control points. Using DJI RTK drone technology as a representative platform, we establish rigorous technical criteria for RTK positioning quality, camera calibration protocols, and flight parameter optimization essential for centimeter-level precision.
Drone technology employing Real-time kinematic (RTK) positioning fundamentally transforms aerial photogrammetry. Four distinct methodologies ensure positioning reliability:
| RTK Fix Solution Methods | Implementation | Use Case |
|---|---|---|
| CORS Network | Direct connection to Continuously Operating Reference Stations | Areas with stable CORS coverage |
| D-RTK2 System | DJI proprietary mobile base station | Remote operations without internet |
| Network 1+1 | Dual data link redundancy via controller | Urban environments with signal interference |
| PPK Processing | Post-processed kinematic correction | Signal-challenged environments |
Positioning quality directly determines mapping accuracy. For reliable photogrammetric outcomes, positional standard deviations must satisfy:
$$\sigma_{\text{lon}}, \sigma_{\text{lat}}, \sigma_{\text{alt}} < 0.1 \text{ m}$$
Non-fixed solutions (float/single-point) introduce significant errors and require exclusion from processing pipelines.
Camera calibration parameters exhibit temporal instability due to mechanical stress and environmental factors. Our calibration protocol requires:
- Dedicated calibration flight over textured terrain with 10-15 ground control points
- RTK-smoothed GCP coordinates with redundant validation
- Bundle adjustment with self-calibration parameters:
$$ \Delta x = -\left( x – x_0 \right) \left[ k_1 r^2 + k_2 r^4 + k_3 r^6 \right] + p_1 \left[ r^2 + 2(x – x_0)^2 \right] + 2p_2 (x – x_0)(y – y_0) $$
$$ \Delta y = -\left( y – y_0 \right) \left[ k_1 r^2 + k_2 r^4 + k_3 r^6 \right] + p_2 \left[ r^2 + 2(y – y_0)^2 \right] + 2p_1 (x – x_0)(y – y_0) $$
where \( r = \sqrt{(x – x_0)^2 + (y – y_0)^2} \)
Optimal flight parameters for ground sampling distance (GSD) ensure mapping accuracy across scales:
| Mapping Scale | Terrain Type | Max Planar Error (m) | Max Elevation Error (m) | Optimal GSD (cm) |
|---|---|---|---|---|
| 1:500 | Flat | 0.13 | 0.11 | ≤5 |
| Hilly | 0.13 | 0.20 | ||
| Mountainous | 0.20 | 0.26 | ||
| Alpine | 0.20 | 0.40 | ||
| 1:1,000 | Flat | 0.30 | 0.20 | 6-8 |
| Hilly | 0.30 | 0.26 | ||
| Mountainous | 0.40 | 0.26 | ||
| Alpine | 0.40 | 0.40 |
Flight planning requires ≥70% frontal overlap and ≥60% lateral overlap, with terrain-following mode activated in high-relief areas. The photogrammetric error equation incorporating GNSS observations is:
$$ v_x = \frac{\partial x}{\partial X}dX + \frac{\partial x}{\partial Y}dY + \frac{\partial x}{\partial Z}dZ + \frac{\partial x}{\partial \phi}d\phi + \frac{\partial x}{\partial \omega}d\omega + \frac{\partial x}{\partial \kappa}d\kappa + \frac{\partial x}{\partial \lambda}d\lambda – l_x $$
$$ v_y = \frac{\partial y}{\partial X}dX + \frac{\partial y}{\partial Y}dY + \frac{\partial y}{\partial Z}dZ + \frac{\partial y}{\partial \phi}d\phi + \frac{\partial y}{\partial \omega}d\omega + \frac{\partial y}{\partial \kappa}d\kappa + \frac{\partial y}{\partial \lambda}d\lambda – l_y $$
where \( (X,Y,Z,\phi,\omega,\kappa) \) represent exterior orientation parameters.
Weighting strategy in bundle adjustment prioritizes observation quality:
$$ P_{\text{GNSS}} = 100 \quad P_{\text{camera}} = 50 \quad P_{\text{angular}} = 10 $$
This weighting schema ensures high-precision GNSS observations dominate the solution while accommodating less reliable IMU attitude data.
Experimental validation covered 116 km² of complex terrain in Fujian Province using DJI M300 RTK drone technology with P1 camera. Critical parameters included:
- GSD: 0.06 cm
- Flight altitude: 120 m AGL
- PPK base station: Central location with 360° sky visibility
- 121 validation points: RTK-surveyed with redundant verification
Accuracy assessment demonstrated consistent sub-decimeter precision:
| Adjustment Method | Control Points | Planar RMSE (m) | Vertical RMSE (m) | Fieldwork Efficiency |
|---|---|---|---|---|
| Image-free GNSS adjustment | 12 (validation only) | 0.101 | 0.093 | 1 field day |
| Traditional GCP adjustment | 121 | 0.091 | 0.074 | 6 field days |
Statistical analysis of checkpoint errors revealed:
- Planar errors: Min 0.020 m, Max 0.213 m, Mean 0.090 m
- Elevation errors: Min 0.010 m, Max 0.280 m, Mean 0.070 m
These results confirm that modern drone technology achieves Category I accuracy for 1:500 mapping in hilly terrain when implementing rigorous technical protocols. The Unmanned Aerial Vehicle workflow reduces field operations by 83% while maintaining centimeter-level accuracy through:
- Strict pre-flight calibration protocols
- RTK/PPK quality assurance thresholds
- Optical system error modeling
- Weight-optimized GNSS-assisted bundle adjustment
The proliferation of drone technology creates compelling economic advantages. For typical 100 km² projects, UAV photogrammetry reduces costs by approximately $4,000 per project through eliminated ground control surveys while maintaining rigorous national mapping standards. This research establishes replicable protocols for integrating Unmanned Aerial Vehicle platforms into large-scale mapping operations without compromising accuracy.