FRE 490: Spatial Data Analysis and Remote Sensing

Semester: Winter 2024 (Term 2: Jan 08, 2024 to Apr 12, 2024)

Lecture: Monday, Wednesday 10:00-11:30 in Ponderosa Commons North - Oak/Cedar House Room 1003

Instructor: Jonathan Proctor (jon.proctor@ubc.ca)

Teaching Assistant: Dennis Engist (engistd@student.ubc.ca)

Instructor Office Hours: Monday 11:30-12:30 in MacMilan 235, or by appointment

Teaching Assistant Office Hours: 15:00-16:00 in MacMillan 192A, or by appointment

Note: currently listed as FRE 490 002: Current Issues in Food and Resource Economics. 

This course introduces students to spatial data analysis and remote sensing using the R programming language with a focus on social science applications. The first half of the course teaches students to create, store, manipulate, and analyze spatial data including points, lines, polygons and rasters. The second half of the course introduces students to color- and texture-based approaches for remote sensing, as well as more advanced spatial data analysis methods. Lectures will teach course concepts, and the lab assignments will give students further experience implementing concepts in code. The course has a midterm exam and final project. 

Grades are based on: lab assignments (45%), midterm (15%), final project proposal (5%), final project presentation (5%), final project writeup (20%), and class participation (10%). 

While the labs are designed to teach R coding skills, and no prior coding experience is required, some prior coding experience will reduce the time it takes to complete the labs. A basic background in probability and statistics (e.g., a basic understanding of linear regression) is also recommended. 

Optional textbook: "Spatial data analysis" by Christopher Lloyd, Oxford University Press. Because spatial data analysis is a relatively young field there is not a perfect textbook for this course; some may find this text helpful. 

Lab assignments are where much of the learning of the course will occur. Labs will be submitted online via canvas. While the initial labs are relatively basic, they increase in difficulty as the topics studied increase in complexity. Use of ChatGPT on the lab assignments is discouraged, as learning to analyze spatial data requires practice (i.e. typing!). Use of such resources for the final project is acceptable. Students are encouraged to talk through approaches to lab assignments with each other; however the labs are individual assignments and code should never be directly shared. Assignments that share duplicated code will all receive zero credit. Late assignments are accepted up to one week after they are due and receive a one point penalty out of three points (i.e. they can receive a maximum score of 2/3). Lab solutions are posted a week after labs are due; late assignments are not accepted after solutions have been posted. Labs are due before the start of class (10:00 am) on the day stated for each assignment. 

Class policies: 

• Computers are not allowed to be open during lecture to promote a respectful and attentive class environment, unless explicitly required for a class activity. 
• Please post course questions on canvas so all can see replies. Personal questions can be asked through email. Instructors will aim to respond to questions within two working days, but it should not be expected that they respond sooner than that time-frame. Please plan ahead accordingly. 
• Course feedback is welcome and encouraged! And if you find a mistake in the course materials, please let me know as soon as possible so that it can be corrected.
• This is a new course, so topics and structures may be adapted based on feedback. 

Course Schedule (subject to change): 

Week 1: 1/8 and 1/10: Introduction

• Course objectives, structure and schedule
• A brief history of spatial data
• Notation
• Coordinate systems

Lab 0 assigned, Lab 1 assigned.

Week 2: 1/15 and 1/17: Space: measurement and representation

• Measurement, error propagation
• Distance
• Triangulation 
• Latitude and Longitude
• Projection systems

Reading: TBD 

Lab 0, Lab 1 due.

Lab 2 assigned.

Week 3: 1/22 and 1/24: Point processes

• Poisson point processes 
• Density estimation
• Clustering and dispersion

Reading: TBD 

Lab 2 due

Lab 3 assigned

Week 4: 1/29 and  1/31: Lines, Polygons, and Networks

• Definition and manipulation of lines, polygons and networks. 
• Buffers
• Centroids

Reading: TBD 

Lab 3 due

Lab 4 assigned

Week 5: 2/5 and  2/7: Fields, interpolation, evaluating model performance

• Fields and rasters
• Masking
• Moving window summary statistics
• Interpolation: polynomial regression, nearest neighbor, inverse distance weighting
• Model evaluation: cross validation and spatial cross-validation
• Bias-variance tradeoff

Reading: TBD 

Lab 4 due

Lab 5 assigned

Week 6: 2/12 and 2/14: High dimensional fields

• vector fields (rasters), dynamic fields (time series), dynamic vector fields (raster block over time). 
• Moving window regression
• Point-wise (time-series) regression
• Spatial averaging
• Linearity and scale-invariance of aggregation (e.g. climate and agriculture)

Midterm Exam (2/14) during class. 

Lab 5 due

Final Project Proposal assigned

No class: Mid-term break 2/19 - 2/23

Week 7: 2/26 and  2/28: 3D Data, Policies in Space, and Causal Inference

• 3-D data: points, surface, 3-D fields
• Principal components analysis
• Optimizing policies in space
• Dijkstra’s algorithm
• Prediction and causal inference

Reading: TBD 

Final project proposal due.

Lab 6 assigned (two weeks to complete)

Week 8: 3/4 and  3/6: Spectral-based remote sensing methods

• Light and the electromagnetic spectrum
• Color, spectral bands
• Spectral indices
• Overview of existing satellite data and data products

Reading: TBD

Week 9: 3/11 and  3/13: Texture-based remote sensing methods

• Sharpening (unsharp masking)
• Convolutional filters
• Ridge regression
• Evaluation of performance (cross-validation part 2)

Reading: TBD

Lab 6 due

Lab 7 assigned (two weeks to complete)

Week 10: 3/18 and  3/20: Texture-based remote sensing methods continued

• Convolutional neural networks
• Transfer-learning
• Scale-invariance of unsupervised featurizations.  

Reading: TBD

Week 11: 3/25 and  3/27: Use of remotely sensed measurements in inference

• measurement error and correction
• implications for parameter bias and variance.

Reading: TBD

Lab 7 due

Week 12: 4/1 and 4/3: Spatial tools for estimating climate change impacts:

• Interpretation of cross-sectional regression, panel regression, long differences regression
• Spatial regression models (spatial lags)
• Spatial autocorrelation, clustering standard errors. 

Reading: TBD

Week 13: 4/8 and 4/10: Final project presentations 

Final project write-up due April 16th, 2024. 

Note: Special thanks to Solomon Hsiang and previous graduate instructors for Spatial Data Analysis for sharing many course materials that this course builds off of.