Data Fusion from Airborne Hyperspectral Data, Airborne LiDAR Data and Aerial photographs at Aramo, Spain

Metadata information

Full Title

Data Fusion from Airborne Hyperspectral Data, Airborne LiDAR Data and Aerial photographs at Aramo, Spain

Fusion of different airborne remote sensed and already processed data gathered from color aerial photography, LiDAR and hyperspectral data acquisition over the Aramo site in Spain.

Abstract

This dataset comprises results from the S34I Project, derived from processing of airborne hyperspectral data, airborne LiDAR data and color aerial imagery acquired at the Aramo pilot site in Spain.
This document describes processing of color imagery, production of color orthophoto, processing of LiDAR data, and fusion of these data with processed and classified thematic hyperspectral data.

Eurosense conducted complex airborne data acquisition in two consecutive days 30.09.2023 and 01.10.2023 using Riegl LM7800-9184 LiDAR sensor and IGI Digicam H4D-50 medium format RGB camera. 1,645 high resolution RGB images were collected over 24 flight lines. Eurosense produced LiDAR point cloud and color orthophoto mosaic.

LiDAR data processing:

Description of the software’s used

AeroOffice and GrafNav  – software used for direct georeferencing of mobile and aerial mapping sensors using GNSS and inertial technology.

SDCimport applies the so-called ONLINE Full Waveform Analysis to the digitized echo signals provided by the laser scanner and additionally transforms the geometry data (i.e., range and scan angle) into Cartesian coordinates. The output is a point cloud in the well-defined Scanner's Own Coordinate System (SOCS) with additional descriptors for every point, e.g., a precise time stamp, the echo signal intensity, the echo pulse width, a classification according to first, second, up to last target.

RiWorld  transforms the scan data into the coordinate system of the position and orientation data set, usually ETRS89 of WGS84 geocentric. It thus provides the acquired laser data of the object's surfaces within a geocentric coordinate system for further processing. In that case the final coordinate system was WGS84 UTM30N – GRS80.

TerraMatch fixes systematic orientation errors in airborne laser data. It measures the differences between laser surfaces from overlapping flight lines or differences between laser surfaces and known points. These observed differences are translated into correction values for the system orientation - easting, northing, elevation, heading, roll and/or pitch.

TerraScan  is the main application in the Terrasolid Software family for managing and processing all types of point clouds. It offers import and project structuring tools for handling the massive number of points of a laser scanning campaign as well as the corresponding trajectory information. Various classification routines enable the automatic filtering of the point cloud.

Geometric corrections

Its content mainly concerns the geometry of the point cloud and quality control.

Initial setting

At the start of treatment, data was calculated by applying the sensor alignment settings corresponding to the last scanner calibration (boresight angles).

Roll: -0.22300

Pitch: -0.04320

Yaw: 0.00170

Determination of connecting lines

The first operation is the extraction of the tie lines used for the adjustment. They are determined by automatic analysis of the data of the different bands, classified as ground (2) and building (6).

They are extracted after the expedited automatic classification described in the previous paragraph.

Absolute control of altimetry

Absolute control of the altimetry is carried out using field measurements of the reference and control fields.

Elevation reference fields

A set of 6 altimetric reference fields were measured in the field by a surveyor.

Result of the absolute adjustment.

Average dz -0.001

Minimum dz: -0.091

Maximum dz: 0.089

Average magnitude: 0.026

Root mean square: 0.034

Std deviation: 0.034

Classification

The delivered classification contains class “Ground” (2), “Vegetation” (4), “Building” (6), “Water” (9) and class 1 “Unclassified”, based on the ASPRS standard.

Evaluation of LiDAR processing results

Absolute height

Both the connection fields and the independent control fields fit within the height tolerances. Global average difference on control fields it is less than -0.001 cm.

Point density and data coverage.

The covered area meets the point density requirement of 10 pts/sqrm.

All checks show that the data meets the accuracy specifications of an accurate LiDAR project.

Orthoprocessing:
Triangulation is needed for precise positioning of aerial photographs. The full camera calibration performed because the practice shows that it is necessary for medium format cameras. The control points were collected from point cloud on such objects which were well recognizable in point cloud and also on aerial photographs. For the full area 43 control points are defined and measured in both datasets. The control points coordinate mean residuals are the following in the result of aerial triangulation adjustment:  rmsx =0.18 m; rmsy =0.17 m; rmsz =0.26 m.
Because of double flights (opposite directions on same flight lines) gave the possibility to produce dsm based ortho-mosaic in 25cm ground resolution.

Data fusion of different sensors data (Postprocessing)
The generated raster data are delivered as georeferenced TIFF files. These raster data are covering 116 km² from LiDAR data and 114.6 km² from aerial photographs with a spatial resolution of 1.2 m per pixel. The no-data value is set to -9999, representing areas which are outside of photo and LiDAR coverage. The projected coordinate system is UTM Zone 30 Northern Hemisphere WGS 1984, EPSG 4326.

 
Generated LiDAR raster data and aerial ortho-mosaic image down-sampled to hyperspectral band ratio mosaics resolution (which has the following pixel size x: ~1.2m   y: ~1.09m).
Generated raster from point cloud are the following: Intensity, Digital Terrain Model, Digital Surface Model.
Intensity band had been interpolated with average method while DTM (from class 2) and DSM (from class 2,4,6,9) with IDW methods. RGB true color composite ortho-mosaic resampled to 1.2m. The ortho-mosaic R, G, B bands are separated to 3 single bands and reformatted to float pixel type and no-data value set to -9999

All bands of three sensors, merged into one composite image with following bands and with the following short names:
BRn Band1 – 9 Band ratio of hyperspectral data according to former document (https://zenodo.org/uploads/14193286) BR1 - BR9

LDint Band10 LiDAR intensity raster

LDdtm Band11 DTM layer generated from LiDAR data class 2

LDdsm Band12 DSM layer generated from LiDAR data class 2,4,6,9

OmosR, OmosG, OmosB Band13,14,15 are R G B channels of true color ortho-mosaic of aerial images

Keywords

Earth Observation, Remote Sensing, Hyperspectral Imaging, Automated Processing, Hyperspectral Data Processing, Mineral Exploration, Critical Raw Materials

Pilot area

Aramo

Language

English

URL Zenodo

https://zenodo.org/uploads/xxxxxxxxx

Temporal reference

Acquisition date (dd.mm.yyyy)

30.09.2023; 01.10.2023

Upload date (dd.mm.yyyy)

04.02.2025

Quality and validity

Format

GeoTiff

Spatial resolution

1.2m 

Positional accuracy

0.5m 

Coordinate system

EPGS 4326

Access and use constrains

Use limitation

None

Access constraint

None

Public/Private

Public

Responsible organisation

Responsible Party

EUROSENSE - Esri Belux

Responsible Contact

Victoria Jadot

Metadata on metadata

Contact

victoria.jadot@eurosense.com

Metadata language

English