grid-nasa-lvis-l2

Grid laser-shot-based LVIS L2 data from NASA into raster data

Contact: Zhan Li, zhanli AT gfz-potsdam dot de, zhanli1986 AT gmail dot com

This is part of the work by Zhan Li at Pacific Forestry Centre of Canadian Forest Service.

Overview

NASA’s Land, Vegetation, and Ice Sensor (aka the Laser Vegetation Imaging Sensor or LVIS) is a full-waveform airborne lidar sensor [1]. The large footprints of LVIS laser shots approximate circles with diameters at tens of meters. These laser shots are geolocated and their return waveforms are processed into the LVIS L2 data products by NASA [2]. For example data, check one of the LVIS L2 products from the Arctic-Boreal Vulnerability Experiment (ABoVE) Airborne Campaign [3].

This repo of scripts grids the shot-based LVIS L2 data into raster data, given a raster grid of user-defined resolution and spatial reference system. Gridded products allow easier synergetic uses of LVIS data with other remote sensing products, particularly remote sensing of land by medium resolution satellites such as Landsat and Sentinel-2.

The gridding procedure in this repo uses weighted average of LVIS L2 variable values from all the shots covering a grid cell. The weight of each shot is the area of this shot covering a grid cell.

Installation

NOTE: the program has been tested only on Linux system.

The required dependencies of this repository are listed in the file environment.yml. Two recommended ways to install these dependencies for using scripts in this repo.

First way: Install dependencies via conda program

1. Install conda program.

2. Create a conda environment using the file environment.yml in the repository.

3. Activate your conda environment.

Alternative way: Unpack dependencies from a pre-zipped file

The dependencies are pre-packaged into a zipped file by conda-pack program. This zipped file, called conda-env-rasterio.tar.gz, comes with a release that you may download from this repository on github.

1. Unzip conda-env-rasterio.tar.gz into a directory you may name your_dir

$ tar -C your_dir -xzf conda-env-rasterio.tar.gz

2. Run the following command to set up environment variables

$ source your_dir/bin/activate
$ conda-unpack

3. Now you are ready to use the scripts.

Quickstart

Use the main script grid_lvis_l2.py to do all the steps in one go that convert an ASCII file of LVIS L2 data into a point vector (preferentially in SQLite format) where each point represents the center of a grid cell that is covered partially or fully by one or more LVIS laser shots. Each point is attached with user-selected LVIS L2 variables to be gridded. Each point is also attached with some ancillary fields that help determine the goodness of coverage of a grid cell by laser shots, such as shot coverage percentage, distance from grid cell centers to covering shot centers, etc.

To see help for this main script

$ python grid_lvis_l2.py -h 

From a point vector of grid cell centers, you can generate a raster file of any field you like using the script rasterize_vector.py. This script rasterizes vector files into raster files based on gdal_rasterize but with some special treatment/improvements of the rasterization, including,

1. when rasterizing points on regular grids, points will be placed in the center of rasterized pixels.

2. when given a template raster, the input vector will be rasterized into a grid that is aligned with the given template raster.

To see help for this script

$ python rasterize_vector.py -h

Examples

• Grid an LVIS L2 product in ASCII format into a 20-m grid in UTM Zone 9 projection. The ASCII file LVIS2_ABoVE2017_0630_R1803_069010.TXT is downloaded from [3].

$ python grid_lvis_l2.py -r 20 \
	--out_srs 'PROJCS["UTM Zone 9, Northern Hemisphere",GEOGCS["WGS 84",DATUM["WGS_1984",SPHEROID["WGS 84",6378137,298.257223563,AUTHORITY["EPSG","7030"]],AUTHORITY["EPSG","6326"]],PRIMEM["Greenwich",0,AUTHORITY["EPSG","8901"]],UNIT["degree",0.0174532925199433,AUTHORITY["EPSG","9122"]],AUTHORITY["EPSG","4326"]],PROJECTION["Transverse_Mercator"],PARAMETER["latitude_of_origin",0],PARAMETER["central_meridian",-129],PARAMETER["scale_factor",0.9996],PARAMETER["false_easting",500000],PARAMETER["false_northing",0],UNIT["Meter",1]]' \
	--column_type lvis_l2_column_types.csvt \
	--column2grid rh10 rh99 -- \
	LVIS2_ABoVE2017_0630_R1803_069010.TXT LVIS2_ABoVE2017_0630_R1803_069010_grid_points.sqlite

The file lvis_l2_column_types.csvt lists the data types of each column of input ASCII file LVIS2_ABoVE2017_0630_R1803_069010.TXT. An example of the file content of lvis_l2_column_types.csvt is as follows,

"Integer","Integer","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Real","Integer","Integer","Integer"

See [4] for more details on this .csvt file type to define data types of columns in ASCII files.

• Grid an LVIS L2 product in ASCII format into a raster grid that aligns with a given template raster in the same spatial reference system. Meanwhlie, after the gridding process, it will keep intermediate files that are generated by gridding.

$ python grid_lvis_l2.py --keep_intermediate \
	-t template_raster.tif \
	--column_type lvis_l2_column_types.csvt \
	--column2grid rh10 rh98 rh99 rh100 complexity -- \
	LVIS2_ABoVE2017_0629_R1803_092329_with_cc.TXT LVIS2_ABoVE2017_0629_R1803_092329_grid_points.sqlite

• Generate a raster of rh10 variable from the point vector of grid cell centers, at 20-m resolution in GeoTiff format.

$ python rasterize_vector.py -r 20 -a rh10_wt_avg -f GTiff --nodata -9999 --init -9999 \
	--ot Float32 \
	LVIS2_ABoVE2017_0630_R1803_069010_grid_points.sqlite LVIS2_ABoVE2017_0630_R1803_069010_grid_points_rh10_wt_avg.tif

• Convert a sqlite file of point vectors of grid cells to a CSV file for easy inspection.

$ ogr2ogr -f CSV -lco GEOMETRY=AS_XY \
	LVIS2_ABoVE2017_0629_R1803_057198_grid_points.csv LVIS2_ABoVE2017_0629_R1803_057198_grid_points.sqlite

References

[1] Blair, J.B., Rabine, D.L., Hofton, M.A., 1999. The Laser Vegetation Imaging Sensor: a medium-altitude, digitisation-only, airborne laser altimeter for mapping vegetation and topography. ISPRS J. Photogramm. Remote Sens. 54, 115–122. https://doi.org/10.1016/S0924-2716(99)00002-7

[2] https://lvis.gsfc.nasa.gov/Data/DataStructure.html

[3] https://nsidc.org/data/ABLVIS2

[4] https://giswiki.hsr.ch/GeoCSV#CSVT_file_format_specification