Airborne LIDAR Technology for Sustainable Watershed Management: Ecological Health of Riparian Zones

Michez et al. (2013) examine the application of LIDAR data to monitor the ecological riparian zones, which are of concern in most parts of Europe due to the European Water Framework Directive. As a transition zone between dry and wet environments riparian areas are typically high in ecosystem productivity and biodiversity. The authors use LIDAR point cloud data to derive a Digital Terrain Model (DTM) and canopy height model which allows for establishing general riparian zone attributes and ecological integrity. The approach presented also disaggregates/re-aggregates the analysis in 50 m long channel axis reaches to provide important insights for natural resource managers as it provides information to prioritize areas in need of intervention.

LIDAR data was acquired for the Houille River trans-boundary watershed starting in Belgium and crossing into France. The approach enabled a cost effective assessment of the riparian zone's ecological integrity made possible by the availability of remote sensing imagery, increases in computing capacity and advancements in geomatic methods. The approach was applied to a 24 km section of the Houille River with a floodplain width varying from 55 to 540 metres and average river width of 8 m. Using object-based image analysis (OBIA) and ArcInfo toolboxes the author was able to extract information on the following parameters: Wetted Channel Extent, Floodplain Extent, Riparian Forest Patches, Riparian Forest Longitudinal Continuity, Overhanging Vegetation, Evergreen Coniferous Stand Detection, Mean and Relative Standard Deviation of Tree Height and Riparian Forest Relative Water Level at the time of survey.

The first step in the analysis was to establish a DTM accomplished by separating the data into 50 m transects along an axis derived by smoothing the centerline of the riverbed. Along with an altimetric reference plan, the DTM data was used to derive a relative DEM in relation to the channel elevation. The data was then re-aggregated at 1 km segments and analyzed for establishing the characteristics of the riparian forest (patch detection, overhanging character, longitudinal continuity, evergreen coniferous stands, heights of riparian forest stands and relative water levels). A canopy height model and leaf off intensity vegetation layer were used to identify the forest patches using a height threshold of 1.5 metres so it can be distinguished from tall herbaceous plants during leaf off period. The overhanging character was derived by examining both the length and total area of the canopy over the river relative to the corresponding bank. The longitudinal continuity or presence of forest along the stream network were identified by classifying all non-riparian forest patches as gaps in the longitudinal continuity when 1) mean height was less than 1.5 metres, 2) width was greater than 5 metres and a minimal area of 50 m2 was observed. A continuity gap index was then derived by subtracting 1 from the riparian forest gap area in the project area divided by total project area. The evergreen coniferous stands were identified using a canopy height model and leaf off “intensity vegetation” rasters as well as photosynthetic activity in March 2011 which enabled a differentiation between broadleaved trees. The height of riparian forest stands were derived using basic statistical analysis.

A process of validation was undertaken based on existing data (altimetric information and ortho-images) and field visits to verify the wetted channel and floodplain extent and evergreen coniferous stands. The validation exercise resulted in an acceptable level of error when compared to other authors. The DTM derived had a root-mean-square-error (RMSE) of 0.14 and a mean DTM error of 0.11 +/- 0.08 m (mean +/- Standard Deviation). The mean distance between the LIDAR wetted channel extent and field measured wetted channel extent was 0.71 and 0.89 m with an RMSE of 1.29 m. The classification accuracy was 76.8%. There was a consistent overestimation of the floodplain extent of 5.52 and 6.02 m. For the riparian forest height values there was a slight overestimation with a RMSE of 2.18 m. The classification accuracy (Kappa Index) was 87.3% when examining all tree types (Evergreen and Deciduous stands and Isolated Trees).

The authors then report on the riparian zone attributes that the analysis yielded (channel width pattern, floodplain width pattern, continuity index, riparian tree height, pattern of overhanging, flooding frequencies along the riparian zone derived with the DEM). The Channel Width Pattern generally increases from 6 m to 10 m with 2 local width peaks of 15 m. The floodplain width ranges from 50 to 400 m. The continuity index shows that the riparian forest patches have a higher continuity close to the wetted channel but that there is a marked decrease when crossing built-up areas. The riparian tree height consistently decreased when crossing built-up areas but there was no upstream downstream differences. With respect to the pattern of overhanging no differences were observed. The area of coniferous stands was not linked to the built up areas but a coniferous plantation was identified and since they are prohibited within the areas of the wetted channel (+1 and +8 m buffers) this information is useful for land and water managers. The DEM provided the basis to examine the flooding extents under various river flow scenarios. It was observed that the areas with riparian patches were less likely to be flooded.

To conclude the authors provide a systematic way of assessing the ecological integrity of riparian zones using LIDAR point cloud data and applying the OBIA approach along with classic geo-processing. The approach allowed for measuring and deriving important parameters of the wetted channel and floodplain scales. At the wetted channel scale, the wetted channel extent and overhanging riparian forest were derived from the data. At the floodplain scale, the ecological integrity of the riparian zone can be assessed by examining the longitudinal continuity, structure (height), composition (coniferous stands) and relative water levels (flooding frequencies and groundwater availability). The resulting information can be very useful for land and water managers to develop and prioritize intervention strategies to assist with enhancing the ecological health of riparian zones which have been identified in the European Water Framework Directive as primordial. The researchers aim to replicate this work in 2014 as additional areas of Belgium are slated to receive LIDAR coverage in the near future.