class: middle, title-slide <!-- top logo (comment to remove or edit on `conf/css/style.css:23`) --> <div class="lab-logo"></div> <!-- <div class="uni-logo"></div> --> # Who is Willian Vieira? <hr width="65%" align="left" size="0.3" color="#FFC800"></hr> ## and what has he been doing? ### Willian Vieira, .small[Analyst] <br><br><br><br><br> [<i class="fa fa-github fa-lg" style="color:#e7e8e2"></i> WV-Habitat/me](https://github.com/WV-Habitat/me) [<i class="fa fa-twitter fa-lg" style="color:#e7e8e2"></i> @WillVieira90](https://twitter.com/willvieira90) --- # Outline - My thesis project - The ECCC project - Side projects - Future projects --- class: middle, center, inverse # My PhD thesis <hr width="100%" align="left" size="0.3" color="#FFC800"></hr> --- .center[.font140[**How can we better predict tree species distribution?**]] <br> - Species Distribution Models (SDM) are not appropriated for trees - But we can use *theory* to create **mechanistic models** - What are the drivers of forest dynamics shaping tree distribution? - Mathematical and statistical models to assess the effect of climate, competition, and forest management --- # Chapter 1: State Transition Model .center[![:scale 55%](https://raw.githubusercontent.com/willvieira/PhD/master/chapter1/img/fig1.png)] - Derived from the metapopulation theory - **B**oreal, **T**emperate, **M**ixed, and **R**egeneration ~ MAT + MAP - Extension to include forest management practices - Analytical equations, Matrix algebra, Jacobian approximation - [STM R package](https://willvieira.github.io/STManaged/index.html); [Shiny App](https://github.com/willvieira/shiny_STM-managed) .cite[Model from Vissault et al. [2020](https://onlinelibrary.wiley.com/doi/abs/10.1111/jbi.13978)] --- # Chapter 2 & 3: Integral Projection Model <br> .center[![](images\ipm3.png)] - Demographic models for 31 tree species from eastern North America - `\(\lambda\)` ~ [Growh + mortality + recruit] ~ climate + competition - Non-linear hierarchical Bayesian models - Stan: statistical programming language derived from C++ - High-performance computing (HPC) - C.2: Random forest for sensitivity analysis - C.3: Scale integration - from individual performance to the metapopulation --- class: middle, center, inverse # The ECCC project <hr width="100%" align="left" size="0.3" color="#FFC800"></hr> ## Sampling design for monitoring boreal birds in Quebec --- # BMS project <br> .pull-right[![](https://willvieira.github.io/sampling_BMS/studyArea_files/figure-html/fig-ecoregion-1.png)] - National effort to develop a protocol to sample boreal birds - Adapt and implement national design for the Quebec region - Spatially stratified sampling approach weighted by: - Habitat - Cost - Legacy sites - [Report](https://willvieira.github.io/sampling_BMS/) --- # BMS project .font130[R&D: A new approach to account for legacy sites] .center[![:scale 60%](https://willvieira.github.io/sampling_BMS/legacy_files/figure-html/fig-sampleSize-example2-1.png)] --- # BMS project .font130[R&D: A new approach to account for legacy sites] .center[![:scale 60%](https://willvieira.github.io/sampling_BMS/legacy_files/figure-html/fig-ecosExample-1.png)] --- class: middle, center, inverse # Side projects <hr width="100%" align="left" size="0.3" color="#FFC800"></hr> .font140[*From learning useless programming languages to automating unnecessary tasks*] --- # Side projects <br> .font120[Reporting and templates] - [Manuscript template](https://github.com/willvieira/ms_STM-managed) (using markdown, LaTeX, Pandoc, Lua) - [Presentation template](https://github.com/willvieira/talkTemplate) (this presentation, some CSS and JavaScript) - [Lab notebook](https://willvieira.github.io/book_forest-demography-IPM/) .font120[Task automation] - Make (reproducibility of manuscripts) - Crontab ([Kijiji scraper R package](https://github.com/willvieira/KijijiScraper)) - GitHub Actions to automatically: - Test and build R packages - Reproduce analysis - Deploy reports and websites --- class: middle, center, inverse # Future projects <hr width="100%" align="left" size="0.3" color="#FFC800"></hr> .font140[*Or what I would like to explore*] --- # Future projects <br> .font120[Modeling] - Statistical models (spatiotemporal autocorrelation, time series) - Machine learning beyond random forest .font120[Development tools] - Modularization (Packages, unity test, deployment, versioning, code review) - Containerization (Docker) - Data Engineering - Cloud computing --- class: middle, center, inverse # Cloud-based geospatial technologies <hr width="100%" align="left" size="0.3" color="#FFC800"></hr> --- # Cloud-based geospatial technologies <br> .font120[**Cloud Optimized GeoTIFF (COG)**] It is based on two complementary frameworks: 1. Optimal GeoTIFF storage (store and organize each pixel information) - Tiling - Overviews 2. HTTP GET range requests - extract only a portion of the GeoTIFF file - Not mandatory but already built into cloud services (Google, Azure) - Ranges are determined by external metadata --- # Cloud-based geospatial technologies <br> .font120[**Spatial-Temporal Asset Catalog (STAC)**] - A common language to describe geospatial information - metadata standard (JSON) - Structured metadata repository describing - **What** it is - **Where** data is located - **How** it can be used - Hierarchically structured into items, collections, and catalogs Example with the Landsat images: - RED band of an image is an asset - All color bands of an image is an item - All images together are a collection --- # Cloud-based geospatial technologies <br> .font120[**Why should we move towards STAC?**] - Once linked to STAC, we have access to any [new] open data effortless - Catalog is centralized, providing easy searchability of new data sets - Inclusion of new data is easy - Community-based metadata extensions for specific problems -- .font120[**Constrants**] - It comes with high implementation costs (especially if we become data providers) - Less useful if we are interested in only a few spatial datasets ---