REMOTE SENSING DATA FOR USE IN REGIONAL DIKE INSPECTION SHARON CUNDILL1, ROBERT HACK1, MARK VAN DER MEIJDE1,

JOOST VAN DER SCHRIER2 & DOMINIQUE NGAN-TILLARD3

1UNIVERSITY TWENTE, FACULTY OF GEO-INFORMATION SCIENCE AND EARTH OBSERVATION (ITC), ENSCHEDE, THE NETHERLANDS 2ROYAL HASKONING, NIJMEGEN, THE NETHERLANDS 3DELFT UNIVERSITY OF TECHNOLOGY, FACULTY OF CIVIL ENGINEERING &GEOSCIENCES, GEO-ENGINEERING SECTION, DELFT, THE NETHERLANDS

HOW ARE REGIONAL DIKES CURRENTLY MONITORED?

Inspected visually in person Looking for irregularities About 14 000 km of regional dikes in the Netherlands

http://www.ctv.ca/CTVNews/World/20120106/dutch-villages-rain-dike-120105/

Current method of dike inspection is: slow man-hour intensive subjective limited by accessibility to the areas to be inspected

THE IDEA

Remote sensing has been proposed as a possible tool for dike inspection: To identify possible problematic areas for further investigation Rapid collection and processing of data Large coverage (of specified areas) Relatively low costs Access to flooded and difficult/dangerous to reach areas Objective Regular surveys

BUT has not been fully researched

WHAT IS REMOTE SENSING?

Remote sensing is the acquisition of information about an object without making physical contact with the object

Objects reflect or emit different amounts of energy in different wavelengths

This electromagnetic energy is measured by a sensor Data most often collected in image form Sensor maybe on plane, helicopter, satellite, etc

Diagrams from http://www.univie.ac.at/geographie/fachdidaktik/FD/site/external_htmls/imagers.gsfc.nasa.gov/adventure/rs_lesson.html

THE ELECTROMAGNETIC SPECTRUM

Modified from http://www.kollewin.com/blog/electromagnetic-spectrum/

BACKGROUND

2 key inspection features

Quality of dike covering

Phtotographs from http://www.inspectiewaterkeringen.nl/content.asp?h=4&s=39

Soil moisture of the dike

Weeds

Bare patches

Cracks due to dryness

Wet patches

DIKE COVERING

Most regional dikes – grass covering Remote sensing vegetation studies % vegetation cover Plant health and vigour Species differentiation

SOIL MOISTURE

Thermal inertia of water Wet areas take longer to heat up and longer to cool down than

dry areas

Vegetation response Plant temperature affected by available soil moisture Healthy vs stressed affects spectral response (reflect different

amounts of energy in different wavelengths)

STUDY SITE

STUDY SITE

FIELD CAMPAIGN

15 to 16 July 2010 – over 24 hour period 54 point locations

Remote sensing data: Thermal Visible Multispectral Hyperspectral

Validation data: Soil moisture Grass cover quality

FIELD CAMPAIGN

15 to 16 July 2010 – over 24 hour period 54 point locations

Remote sensing data: Thermal Visible Multispectral Hyperspectral

Validation data: Soil moisture Grass cover quality

Apparent Temperature

Blue, green and red

Few broad bands

Many narrow bands

RESULTS: SOIL MOISTURE

Soil moisture Daytime thermal

RESULTS: SOIL MOISTURE

Soil moisture Near infrared

RESULTS: SOIL MOISTURE

Pearson correlation Soil moisture Significancea

Daytime - Thermal -0.668 0.000

Near infrared - Multispectral -0.614

0.000

a Significant at the 0.01 probability level

RESULTS: THERMAL TIME SERIES

Normalised thermal values for selected locations (normalised to the mean). Apparent solar noon: 13h38. Apparent sunset: 21h30. Apparent sunrise: 05h46. Rain event (in grey) at around 16h00.

RESULTS: COVER QUALITY

RESULTS: COVER QUALITY

Spearman's rho Cover quality Significancea

Green:Red ratio – Visible -0.593 0.000

NIR:Red ratio – Multispectral

-0.583 0.000

a Significant at the 0.01 probability level

RESULTS: COVER QUALITY

CONCLUSIONS

Soil moisture conditions can be observed in daytime thermal and multispectral remote sensing data

Visible and multispectral ratios can also be used to evaluate dike grass cover quality

Hyperspectral data is particularly useful for evaluating the quality of grass cover

Thermal measurements are strongly affected by local conditions. Optimal conditions are dry days with clear skies. Optimal time is apparent noon so as to reduce illumination effects.

THANK YOU

Contact: Sharon Cundill Email: cundill@itc.nl This research is part of the RSDYK project of the Flood Control 2015 programme