2012 FEMA Topographic Lidar: Hudson-Hoosic and Deerfield Watersheds, Massachusetts

The Light Detection and Ranging (LiDAR) dataset is a survey of the Hudson-Hoosic and Deerfield project area. The entire survey area for Massachusetts is approximately 690 square miles. The LiDAR point cloud was flown at a nominal post spacing of 2.0 meters for unobscured areas. The LiDAR data and derivative products produced are in compliance with the U.S. Geological Survey National Geospatial Program LiDAR Guidelines and Base Specifications, Version 13-ILMF 2010. The flight lines were acquired by Northrop Grumman, Advanced GEOINT Solutions Operating Unit. Derivative products from the aerial acquisition include: high accuracy multiple return LiDAR data, both raw and separated into several classes, along with hydro flattening breaklines, bare earth DEM tiles, control points, and FGDC compliant XML metadata.

The purpose of this project was to produce a high resolution LiDAR data set of approximately 2,895 square miles over the Hudson-Hoosic and Deerfield

Watersheds in New York.

10418

LiDAR points in LAZ format (ASPRS Class 1,2,7,9,10,14)

none

Any conclusions drawn from the analysis of this information are not the responsibility of FEMA, Northrop Grumman, NOAA, the Office for Coastal Management or its partners.

This data can be obtained on-line at the following URL:

https://coast.noaa.gov/dataviewer/#/lidar/search/where:ID=2556

None

Users should be aware that temporal changes may have occurred since this data set was collected and some parts of this data may no longer represent actual surface conditions. Users should not use this data for critical applications without a full awareness of its limitations. These data depict the heights at the time of the survey and are only accurate for that time.

There is not a systematic method of testing when testing horizontal accuracy in LiDAR. However this is tested during calibration of the sensor and is rechecked during the comparing of parallel and perpendicular flight lines. Additionally the horizontal accuracy is checked by collecting building corners during the survey. Lines are then digitized representing the building outline and the differences are measure from each individual survey point to the corner of the building outline. Stats are calculated to ensure horizontal tolerances are met. These measurements resulted in an RMSEr

of .21 meters and equals a RMSE accuracy of .36 meter horizontal accuracy at the 95 % confidence level. Accuracy defined by NSSDA at the 95 % confidence level would be multiplier by 1.7308 times the RMSE.

Quantitative Value: 0.36 meters, Test that produced the value: Value represent horizontal accuracy assessment at the 95% confidence interval. Units in meters.

The accuracy assessment was performed using the NSSDA standard method to compute the root mean square error (RMSE) based on a comparison of ground

control points (GCP) and DEM derived from the LiDAR dataset. Testing was performed prior to gridding of the filtered LiDAR data points and

construction of the 32-bit ESRI float grid format bare earth tiles. The RMSEz figure was used to compute the vertical National Standard for Spatial

Data Accuracy (NSSDA). A spatial proximity analysis was used to select edited LiDAR data points contiguous to the relevant GCPs. A search radius

decision rule is applied with consideration of terrain complexity, cumulative error and adequate sample size. Cumulative error results from the

errors inherent in the various sources of horizontal measurement. These sources include the airborne GPS, GCPs and the uncertainty of the accuracy

of the LiDAR data points. This accuracy is achieved prior to the subsampling that occurs through integration with the inertial measurement unit (IMU)

positions that are recorded. It is unclear at this time whether the initial accuracy is maintained. The horizontal accuracy of the GCPs is estimated

to be in the range of approximately .03 to .04 meters. Finally, sample size was considered. The specification for the National Standard for Spatial

Data Accuracy is a minimum of 20 points to conduct a statistically significant accuracy evaluation which provides a reasonable approximation of a

normal distribution. The intent of the NSSDA is to reflect the geographic area of interest and the distribution of error in the data set (Federal

Geographic Data Committee, 1998, Geospatial National Standard for Spatial Data Accuracy, Federal Geographic Data Committee Secretariat, Reston,

Virginia, p.3-4). Additional steps were taken to ensure the vertical accuracy of the LiDAR data including: Step 1: Precision Bore sighting (Check

Edge-matching) Step 2: Compare the LiDAR data to the Field Survey (The vertical accuracy requirements meet or exceed the required RMSEz of 12.5cm

and the vertical accuracy of 24.5cm at the 95% confidence level). Data collected under this task order exceeds the required National Standards for

Spatial Database Accuracy (NSSDA) accuracy standards. SVA accuracies at the 95th Percentile collected and tested, as target accuracies results as

follows: Tall Weeds Crops =0.22 meters, Fully Forested = 0.40 meters. Consolidated Vertical Accuracies (CVA) at the 95th Percentile =0.25 meters.

Final accuracy statement for this task order is as follows; FVA Tested 0.15 meters vertical accuracy at the 95% confidence level.

; Quantitative Value: 0.15 meters, Test that produced the value:

Value represents the Fundamental Vertical Accuracy (FVA) assessment at the 95% confidence interval. This represents the FVA checkpoints compared

against the derived DEM at the 95% confidence interval. Units in meters.

Ground Truth data was collected of the three major land cover classes representing 10% of the predominate vegetation dispersed within the area of

interest. 30 points were collected in each of the three predominate vegetation classes bare earth, tall weeds crops, and fully forested, points

collected in the trees vegetation class were collected with a Total Station. Pair of points was surveyed using the local NGS CORS network once completed

the total station is used to collect the forested vegetation class. A total station was used to collect all the shots collected in the forested vegetation

class, due to the limited GPS signal when working in and around tree canopy.

The GPS survey was tied into the local NYSNet Realtime Network located in NEW York, and Vermont, and the KEYNet Realtime Network located in the

Northeast covering Massachusetts. These networks are networks of continuously operating GPS reference stations that allows for Realtime Kinematic

(RTK) capabilities within a realtime network (RTN). This allows for corrections to be applied to the points as they are being collected, eliminating

the need for an adjustment. As a quality control measure several check-in points consisting of NSRS published horizontal and vertical control points

were used as checks within the real-time networks used. The survey crew checked into these published points daily to validate the consistency of the

network. The NSRS published points also confirm that the project will meet the 5cm local network accuracy at the 95% confidence level. Data analysis

was accomplished by comparing ground truth checkpoints with LiDAR points from the derived DEM and reported three ways 1. FVA 2. SVA 3. CVA. Additionally

the FVA points were assessed against the TIN derived from the LAS LiDAR point cloud controlled and calibrated swath data to ensure they met the

required accuracy of 12.5cm RMSEz and 24.5cm at the 95% confidence interval.