Current state and changes in the composition and configuration of landscapes worldwide

class: inverse, left, nonum, clear
background-image: url("figs/rayshader.png")
background-size: cover

.titlestyle[Current state and changes] <br>
.titlestyle[in the] <br>
.titlestyle[composition] <br>
.titlestyle[and] <br>
.titlestyle[configuration] <br>
.titlestyle[of landscapes worldwide]

.captionstyle[Jakub Nowosad and Tomasz F. Stepinski]
.pull-right2[.captionstyle[2020-05-13, the IALE-North America 2020 Annual Meeting]]

---
# Introduction

.lc[

- Quantifying composition and configuration of landscapes is often done on a local scale
- The systematic study of landscape compositions and configurations over the entire landmass is lacking

**Goals:**

- Create a global inventory
- Show current state and changes in spatial composition and configuration of landscapes worldwide

]

.rc[
<img src="figs/lc_map2016.png" width="3333" style="display: block; margin: auto;" />
]

.rc[
- Land cover data for years 1992-2016 from the CCI-LC project
- [Goode homolosine projection](https://en.wikipedia.org/wiki/Goode_homolosine_projection) and nine main categories
- Partitioned into 30 x 30 kilometers square blocks 
- 150,000 mesoscale landscapes
- For each of these landscapes, for each year, a large set of variables was extracted, including **marginal entropy** and **relative mutual information**
]

---
# IT metrics

.pull-left[
- Five information theory metrics based on a co-occurrence matrix exist [(Nowosad and Stepinski, 2019)](https://link.springer.com/article/10.1007/s10980-019-00830-x)

- **Marginal entropy [H(x)]** - diversity (*composition*) of spatial categories - **from monothematic patterns to multithematic patterns**
- **Relative mutual information [U]** - clumpiness (*configuration*) of spatial categories **from fragmented patterns to consolidated patterns**)
- **H(x) and U** are uncorrelated
]

.pull-right[
<img src="figs/it_metrics_maps2.png" width="100%" style="display: block; margin: auto;" />

<img src="figs/it_metrics_maps3.png" width="100%" style="display: block; margin: auto;" />
]

---
# Current state: composition

---
# Current state: composition

<div class="figure" style="text-align: center">
<img src="figs/entropy_map2016.png" alt="Spatial distribution of marginal entropy values" width="2400" />
<p class="caption">Spatial distribution of marginal entropy values</p>
</div>

---
# Current state: configuration

<div class="figure" style="text-align: center">
<img src="figs/relmutinf_map2016.png" alt="Spatial distribution of relative mutual information values" width="2400" />
<p class="caption">Spatial distribution of relative mutual information values</p>
</div>

---
# Global changes: composition

---
# Global changes: composition

.lc[
<img src="figs/entropy_changes2.png" width="1600" style="display: block; margin: auto;" />

]

.rc[

- Global changes in entropy between 1992 and 2016 are spatially autocorrelated
- **Increases in entropy:** Amazon region, east China, as well as most of Europe
- **Decreases in entropy:** Northern Canada, and north parts of Eurasia, as well as central Asia and west Africa
- More areas had an increase in entropy than the decrease

<img src="figs/entropy_changes_map2.png" width="2400" style="display: block; margin: auto;" />
]

---
# Local changes: composition

.pull-left[
**Major decreases in entropy:**
<img src="figs/lc_change_africa3.png" width="100%" style="display: block; margin: auto;" />
]

.pull-right[
**Major increases in entropy:**
<img src="figs/lc_change_amazon3.png" width="100%" style="display: block; margin: auto;" />
]

---
# Global changes: configuration

.lc[
<img src="figs/relmutinf_changes2.png" width="1600" style="display: block; margin: auto;" />

]

.rc[

- **Increases in relative mutual information:** northern Argentina
- **Decreases in relative mutual information:** Amazon region, east China, as well as most of South Sudan
- More areas had an increase in relative mutual information than the decrease

<img src="figs/relmutinf_changes_map2.png" width="2400" style="display: block; margin: auto;" />
]

---
# Local changes: configuration

.pull-left[
**Major decreases in relative mutual information:**
<img src="figs/lc_change_china3.png" width="100%" style="display: block; margin: auto;" />
]

.pull-right[
**Major increases in relative mutual information:**
<img src="figs/lc_change_argentina3.png" width="120%" style="display: block; margin: auto;" />
]

---
class: left, top, clear

.lc2[
## Summary

- **Marginal entropy** and **relative mutual information** are universal indicators of landscapes configuration and composition
- These metrics are **not dependent on a certain category of land cover** (but this information can be incorporated into analyses)
- Globally, there is a trend of increasing entropy **(more landscapes with multithematic patterns)**
- There is also a trend of increasing relative mutual information **(more landscapes with consolidated patterns)**

## Next steps

- Confirming the results on different scales and levels of details
- Relating the state and trends of landscape changes to human and environmental factors
]

.rc2[

## Contact:

<i class="ion  ion-social-twitter "></i> jakub_nowosad

https://nowosad.github.io

## Resources:

- **Slides:** [nowosad.github.io/iale-na_20](https://nowosad.github.io/iale-na_20)
- [**iPoster**](https://2020toronto-ialena.ipostersessions.com/default.aspx?s=40-15-E7-38-9A-EE-C6-76-09-30-1D-88-C9-A5-81-B5)
- **Software:** R packages [motif](https://nowosad.github.io/motif/) and [landscapemetrics](https://r-spatialecology.github.io/landscapemetrics/index.html)
- **Blog post:** [Information theory provides a consistent framework for the analysis of spatial patterns](https://nowosad.github.io/post/ent-bp1/)

]