Investigating the usability of LIDAR-derived intensity image in shoreline extraction studies

Images acquired from varied sensors are frequently used in shoreline extraction. Aerial photographs, orthophotos and satellite images are the prime of these. In recent years, the data provided by Light Detection and Ranging (LIDAR) technique have also included in this group for feature extraction. Many LIDAR sensors transmit beam of 1064 nm to gather information from earth surface. This value is very close to the range in which many multispectral (MS) images record in the Near-Infrared (NIR) portion of the electromagnetic spectrum. One of the most important features of LIDAR technique that is different from others is the intensity data which is defined as the return power recorded by the sensor. LIDAR-derived raster image called as intensity image can be created at a certain spectral range by utilizing the intensity values. In this thesis, it was investigated whether or not the LIDAR-Derived Intensity Image (LDII) may be an alternative to the NIR band in a MS image. Orthophoto, Pléiades Satellite Image (PSI), Spot Satellite Image (SSI) and LDII data including Kestel Dam in Bergama district of Izmir province were processed in the shoreline extractions and comparative analyses. PSI, SSI and two Fused Images (FI) consist of PSI R, G ,B bands and LDIIs were used to extract shoreline of dam lake. The difference between two FIs is the realization of geometric correction applied on the raw LDII. Mean Shift Segmentation (MSS) process was applied on the LDIIs to remove noise while maintaining edge details before fusion with MS bands of PSI. Shorelines were extracted by means of object - oriented classification based on Normalized Difference Water Index (NDWI) values derived from imageries. Comparative analyses were carried out by comparing classification and extraction results with each other. Change detection difference maps were created in order to determine the amount of shrinkage in the dam lake between the acquisition dates of the data. According to these maps produced, there are changes of 5.54 ha including increment of 0.66 ha and decrement of and 5.12 ha including increment of 0.72 ha and decrement of 4.40 hectares in the water of dam lake based on the PSI and FIs. The difference of 0.48 ha caused by geometric correction and increments of approximately 0.7 ha are at negligible level considering the dam occupancy rates depending on the maximum and minimum operation volumes and total lake area. In addition to this, change detection based on distance differences were performed. Distance differences between digitized vectors and extracted shorelines were calculated and statistics were obtained. According to the results obtained, the differences between mean values of the distance calculations from digitized raw LDII to other shorelines and shoreline extracted from FI including uncorrected LDII to other shorelines took a sub-pixel values. These values are 0.1 m, 0.96 m and 1.23 m. Also, the differences between mean values of the distance differences from two shorelines extracted from both FIs to the shorelines obtained from PSI, SSI and orthophoto are quite small. These values are 0.5 m, 0.75 m and 1.16 m and indicate the geometric correction effect. Non-water objects such as soil, vegetation or buildings took values greater than 0 in the NDWI images derived from FIs. Water has low intensity values due to higher absorption of laser beam, resulting in low spectral reflectance for LDII as in the MS NIR band. Additionally, beam of 1064 nm used in LIDAR is closer to the middle infrared portion for many MS imageries. Therefore, MSS-applied LDII provides better distinction with the help of this characteristics. Briefly, LDII can be used as NIR band in feature extraction performed by RBOOC after being made meaningful and utilizable.