LIDAR Bathymetry in very shallow waters. Shachak Pe eri CCOM, UNH William Philpot Cornell University

Transcription

1 LIDAR Bathymetry in very shallow waters Shachak Pe eri CCOM, UNH William Philpot Cornell University

2 Nd:YAG laser generates pulses in the infrared (164 nm) and green (532 nm) simultaneously IR radiation is reflected at the water surface Green light travels through the water and is reflected by the water and the bottom Green light generates Raman scattering from water and fluorescence from bottom vegetation or phytoplankton Bathymetric Lidar laser detectors

3 Water Raman Scattering Virtual state Energy Rayleigh & Mie scattering Raman scattering Inelastic: emission at a longer wavelength than excitation Fixed frequency shift For excitation at 532 nm emission centered at 647 nm

4 Fluorescence Inelastic: emission at a longer wavelength than excitation For excitation at 532 nm emission over a broad wavelength range, including the red channel

5 Lidar return signal (lidar waveform) Received 532nm signal Total return amplitude Surface (interface) return Detection points Bottom return Raman backscatter Volume backscatter Δt time

6 SHOALS: data for single pulse

7 SHOALS: data for single pulse

8 Introduction

9 Introduction: Raman Channel Waveform There are three main components building the Raman channel waveform: "Surface" return (found in all depths). Vegetation (found in shallow waters (Philpot and Wang, 2) Water depth (found in shallow waters). Note: There is no return from the air-water interface. All of the "surface" Raman return is from the water volume. The vegetation contribution "rides" on the surface waveform. The water column contribution decreases in power in the bins after the peak.

10 Lake Tahoe Date of acquisition: July 15 19, Maximum water depth noticed: m Vegetation under water: No Geology: lake deposits, and till. Bathymetry: moderate slope

11 Line 5-1 mt717b Line 12-1 mt717f Color bathymetry: LIDAR (1) Multibeam (Gardner et al., 1998) Line 48-5 mt718b Line 57-1 mt715a

12 5 Basic Red Waveform: depth > 2. m Peak Tail Head Rise dependent on water type depth independent

13 Shallow-water water Red Waveform: depth m 5 Peak Head Rise Head, Rise and Peak are the same as the basic waveform slope of the Tail increases as depth decreases. Falloff begins earlier as the depth decreases Tail

14 The basic waveform varies with location (water type) but is invariant locally

15 Lake Michigan, Dahlia Shoals Date of acquisition: Aug. 27 Sept. 1, 1 Maximum water depth noticed: m Vegetation under water: Yes Geology: Lake deposits. Bathymetry: moderate slope

16 Comparison with depths of shallower waters ( m) m

17 Comparison with depths of shallower waters ( m) m 3 3.1

18 Comparison with depths of shallower waters ( m) m

19 Comparison with depths of shallower waters ( m) m

20 Comparison with depths of shallower waters ( m) m

21 Comparison with depths of shallower waters ( m) m

22 Comparison with depths of shallower waters ( m) m

23 Comparison with depths of shallower waters ( m) m

24 Comparison with depths of shallower waters ( m) m

25 Comparison with depths of shallower waters ( m) m....

26 Comparison with depths of shallower waters ( m) m

27 Comparison with depths of shallower waters ( m) m * interface is the water depth between -.5 m

28 Comparison with depths of shallower waters ( m) m

29 Comparison with depths of shallower waters ( m) m

30 Comparison with depths of shallower waters ( m) m

31 Comparison with depths of shallower waters ( m) m 1.6

32 Comparison with depths of shallower waters ( m) 1 8 Bottom vegetation fluorescence generates a second peak in the waveform The 2 nd peak occurs earlier as the depth decreases Waveform energy drops quickly after the peak 12m 1.5

33 Comparison with depths of shallower waters ( m) m 1.4

34 Comparison with depths of shallower waters ( m) m 1.3

35 Comparison with depths of shallower waters ( m) m 1.2

36 Comparison with depths of shallower waters ( m) 1 8 Tail falls below the "dark current" level implies that the air return is not zero (not likely) ringing due to impedance mismatch between the detector and the digitizer 12m 1.1

37 Comparison with depths of shallower waters ( m) m.9

38 Comparison with depths of shallower waters ( m) m.8

39 Comparison with depths of shallower waters ( m) m.7

40 Comparison with depths of shallower waters ( m) 1 8 At depths < 1. m Early rise Overshoot baseline 12m.6

41 Comparison with depths of shallower waters ( m)* 1 12m 8 interface * interface is the water depth between -.5 m

42 1 8 Comparison with depths of shallower waters ( m)* Fluorescence-induced peak would be detectable at even greater depths. interface m * interface is the water depth between -.5 m

43 Line 5-1, Session mt717b King's Beach

44 Kings Beach, Lake Tahoe Total number of points available: 368 Number of points with depth value: 17 (46% of the total points)

45 Calculating the depths Objective: calculating the shallow water depths using the Raman channel waveforms. Method: Decision rules and classification. Results: preliminary (demonstrating capability)!

46 Problem At King's Beach, of the 368 available LIDAR observations, a water depth was only assigned to 17 (46%) of the observations. Approach 1. Land/water discrimination 2. Separate between water and shallow water 3. Differentiate between shallow and extremely shallow water 4. Assign depths within each range

47 Land/water discrimination Decision rules (water) - Low IR signal - High Raman signal

48 Land/water discrimination Decision rules (land) - High IR signal - Low Raman signal

49 Land/water discrimination Decision rules (water?) - High IR signal - High Raman signal

50 Land/water discrimination Decision rules (vegetation) - High IR signal - Complex Green channel signal - Low Raman signal

51

52 NDI land/water /water Normalized Difference Index NDI land / water bin = bin bin + bin If NDI land/water >.2 water

53 Water soundings Predicted Water Water? Land Actual Water Water? Land Correspondence (Accuracy)=98%

54 NDI (bin11 & bin27) Water soundings Manual

55 Decision rules (deep waters)

56 Decision rules (deep waters) 5 3 Avergage std 1

57 Decision rules (shallow waters) 7 5 NDI shallow = bin bin bin + bin Average std deep 1.2 1

58 Decision rules (shallow waters) sapmle1 sapmle2 sapmle3 sapmle4 sapmle5 land deep

59 Results Total number of available points in the dataset: 368. Number of points with depth value using the Raman waveforms: 36 (97% of the total points). Number of points with depth value using the APD waveforms : 17 (46% of the total points). A water depth comparison between the two waveform datasets (145 points) shows an average difference of.2 m with a std of.14 m.

60 Algorithm Mapping the depth via classification: Raman waveforms Land Water Deep Waters Shallow Waters Shallow Water (depth > 1m) Water Depth Extremely Shallow Water Water Depth

61 Summary The objective of this study is to increase the reliability of shallow water LIDAR surveying using the Raman channel waveforms. There is a direct relationship between the Raman channel waveforms and depths in shallow waters (lowering of the tail). In water depths shallower than 1 m the waveform s peak and fall are pushed towards the head of the waveforms. These results are preliminary, but show the capability of mapping shallow waters to resolution in depth of about a decimeter.

62 DTM comparison (5m resolution) APD Raman