Back

ALGR: A multi-purpose agricultural landscape generator in R

Goldstein, E.; Deutscher, A.; O'Keeffe, E.; Wiegand, K.

2025-04-07 ecology
10.1101/2025.03.31.646291 bioRxiv
Show abstract

O_LIAgricultural and ecological modelers commonly use maps as input for spatially explicit simulations. While real world maps are often used, they are limited by being static objects, therefore making it difficult to assess how patterns within the landscape contribute to ecological processes. Agricultural landscape generators (ALG) are a useful tool for simulating maps in a more flexible way. They can increase robustness of models that rely on landscape maps as input, they allow modelers to give spatial representation to non-spatial models, and they are a useful tool for recreating spatial patterns in agricultural-dominated landscapes. A limitation of previous ALGs is that they have rarely been designed for general use (non-open source software, not written in R, and designed for specific projects). Furthermore, they are typically either extremely general and thus oversimplified or have a high specificity for particular use cases. C_LIO_LIALGR bridges this gap by providing a general-purpose, dynamic landscape generator that balances structural realism with adaptability. ALGR generates agricultural landscapes with a three-step approach: first, outlining potential space, second, field placement inside of that space, and third, enrichment of the landscape with information. This stepwise approach ensures that ALGR generates landscapes with realistic spatial patterns while remaining adaptable to diverse regions and applications. C_LIO_LIIt is the first ALG that is specifically designed to allow a simple integration within the R programming environment and the r-spatial package environment. ALGR is designed as a general-purpose generator, which is simple to use and facilitates an easy integration in modelling workflow. C_LIO_LIWe present several examples of workflows using ALGR, to demonstrate its usefulness. Our examples include: 1) simulating different land use shares, 2) parameter tuning of ALGR to recreate real world landscape patterns 3) spatially distributing crop portfolios, and 4) using real world maps as a basis for field placement C_LI O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=101 SRC="FIGDIR/small/646291v1_ufig1.gif" ALT="Figure 1"> View larger version (46K): org.highwire.dtl.DTLVardef@19c02c3org.highwire.dtl.DTLVardef@17975f6org.highwire.dtl.DTLVardef@12f7aa5org.highwire.dtl.DTLVardef@1cfdd0_HPS_FORMAT_FIGEXP M_FIG Visual Abstract: The three-step conceptual framework of ALGR for landscape generation C_FIG

Published in PLOS ONE (predicted rank #6) · training set

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.