Lidar Data in R - Remote Sensing Uncertainty
Welcome to Week 5!
Welcome to week 5 of Earth Analytics! This week, you will explore the concept of uncertainty surrounding lidar raster data (and remote sensing data in general). You will use the same data that you downloaded last week for class. You will also use pipes again this week to work with tabular data.
For your homework you’ll also need to download the data below.
Time | Topic | Speaker |
---|---|---|
9:30 - 9:40 | Review | |
9:40 - 10:30 | Guest speaker - Chris Crosby, UNAVCO / Open Topography | |
10:30 - 12:20 | Coding: Use lidar to characterize vegetation / uncertainty |
1. Readings
- Influence of Vegetation Structure on Lidar-derived Canopy Height and Fractional Cover in Forested Riparian Buffers During Leaf-Off and Leaf-On Conditions
- The characterization and measurement of land cover change through remote sensing: problems in operational applications?
- Learn more about the various uncertainty terms.
2. Complete the Assignment Below (5 points)
Homework Submission
Produce a Report
Create a new R markdown
document. Name it: lastName-firstInitial-week5.Rmd Within your .Rmd
document, include the plots listed below. When you are done with your report, use knitr
to convert it to html
format. Submit both the .Rmd
file and the .html
file. Be sure to name your files as instructed above!
Use knitr Code Chunk Arguments
In your final report, use the following knitr
code chunk arguments to hide messages and warnings and code as you see fit.
message = FALSE
,warning = FALSE
Hide warnings and messages in a code chunkecho = FALSE
Hide code and just show code outputfig.cap = "caption here"
Add a caption to a figure. When you do this, each figure needs to be in it’s own code chunk!
Answer Questions Below in Your Report
- Write at least 2 paragraphs: In this class you learned the relationship between lidar derived canopy height models and measured tree height. Use that plots that you create below, the readings and the course lessons to answer the following questions
- Which lidar tree height metric, (max vs. mean height) more closely relates to human measured tree height?
- What sources of uncertainty (as discussed in class and the readings) may impact relationship between lidar vs human measured tree height?
- Do you notice any differences in the relationship between the lidar vs human measured tree height between SJER vs SOAP field sites? Explain.
- Write at least 1 paragraph: List a minimum of 3 sources of uncertainty associated with the lidar derived tree heights and 3 sources of uncertainty associated with in situ measurements of tree height. For each source of uncertainty, specify whether it is a random or systematic error. Be sure to reference the plots and readings as necessary.
Include the Plots Below
Be sure to describe what each plot shows in your final report. Your plots do not need to be in the order below. I just listed them this way to make it easier to keep track of and grade!
Plots 1 - 2
Overlay the field site point locations on top of the canopy height model for both the SJER and the SOAP field sites.
Plots 3 - 6: Scatterplots
For both the SJER and SOAP field sites, create scatter plots that compare:
- MAXIMUM canopy height model height in meters, extracted within a 20 meter radius, compared to MAXIMUM tree height derived from the in situ field site data.
- AVERAGE canopy height model height in meters, extracted within a 20 meter radius, compared to AVERAGE tree height derived from the in situ field site data.
Plot 7 - 10 Difference Bar Plots
For both the SJER and SOAP field sites, create barplots that show the DIFFERENCE between:
- Extracted lidar max canopy height model height compared to measured max height per plot.
- Extracted lidar average canopy height model compared to measured average height per plot.
Graduate Students Only
Add a regression line to each scatterplot. For both plots write a thoughtful paragraph describing what the regression relationship tells you about the relationship between lidar and measured vegetation height. Does the comparison between lidar and measured average height look stronger? Or Maximum height? Why might one be “better” or a strong relationship than the other.
IMPORTANT: For All Plots
- Label x and y axes appropriately - include units
- Add a title to your plot that describes what the plot shows
- Add a brief, 1-3 sentence caption below each plot that describes what it shows HINT: you can use the
knitr
argumentfig.cap = "Caption here"
if you are knitting to pdf to automatically add captions.
Homework Due: Monday October 9th 2017 @ 8am.
Submit your report in both .Rmd
and .html
format to the D2l dropbox. Once again you are welcome to submit a .pdf
instead of .html
if you wish!
Report Structure, Code Syntax & Knitr Output: 10%
Full Credit | No Credit |
---|---|
.html or .pdf and .Rmd files submitted | |
Code is written using “clean” code practices following the Hadley Wickham style guide | |
Code chunks contain code and code runs | |
All required R packages are listed at the top of the document in a code chunk. | |
Lines of code are broken up at commas to reduce the line width and make the code more readable | |
Code chunk arguments are used to hide warnings and code and just show output | |
.html / .pdf report emphasizes the write up and the code outputs rather than showing each step of the code (note we will still look at and grade your code but it should not appear in your report) |
Report Questions: 40%
Full Credit | No Credit |
---|---|
Student compared the scatter plots of average and max height and determined which relationship is “better” (more comparable 1:1 ) for both field sites | |
Student discusses 2-3 potential sources of uncertainty that may have impacted these relationships | |
Student discusses differences in the relationships observed between the two field sites (SJER vs SOAP) | |
1-2 readings from the homework are referenced in the report. (You can chose whether you’d like to use bookdown or create manual references) | |
3 sources of uncertainty associated with 1) the lidar derived tree heights and 2) insitu tree height measurements are correctly identified as discussed in class and the readings | |
Student identifies uncertainty sources listed above as systematic vs random |
Plots are Worth 50% of the Assignment Grade
Plots 1 - 2 - Basemap - plot locations overlayed on top of the CHM for each field site.
Full Credit | No Credit |
---|---|
Plots have a title that describes plot contents. | |
Plots have a 2-3 sentence caption that clearly describes plot contents. |
Plots 3 - 6 - Scatterplots Insitu vs Lidar for San Joachin (SJER) & Soaproot (SOAP) Saddle sites
Full Credit | No Credit |
---|---|
Scatter plot of maximum measured vs lidar tree height is included | |
Scatter plot of average measured vs lidar tree height is included | |
Plots have a title that describes plot contents. | |
X & Y axes are labeled appropriately. | |
Plots have a 2-3 sentence caption that clearly describes plot contents. |
Plots 7 - 10 - Difference Bar Plot: Insitu vs Lidar
Full Credit | No Credit | |||
---|---|---|---|---|
Bar plot of maximum measured minus lidar tree height is included | ||||
Bar plot of average measured minus lidar tree height is included | ||||
Plots have a title that clearly describes plot contents | ||||
X & Y axes are labeled appropriately | ||||
Plots have a 2-3 sentence caption that clearly describes plot contents |
Graduate Regression Scatter Plot 1
10% of the regression plot grade
Full Credit | No Credit |
---|---|
Bar plot of maximum measured minus lidar tree height is included | |
Bar plot of average measured minus lidar tree height is included | |
Plots have a title that clearly describes plot contents | |
X & Y axes are labeled appropriately | |
Plots have a 2-3 sentence caption that clearly describes plot contents |
90% of the regression plot grade
Full Credit | No Credit |
---|---|
1-2 Paragraphs are included that describe what these plots show in terms of the relationship between lidar and measured tree height and which metrics may or may not be better (average vs maximum height) |
Example Homework Plots
The plots below are examples of what your plot could look like. Feel free to customize or modify plot settings as you see fit! Note that I did not number the plots this week to allow you to place plots where you’d like in your report.