Onshore wind is expected to supply much of the growth in renewables, yet its expansion increasingly extends into forests, where wind generation and forest conservation act as competing climate strategies for the same land. Net-zero system assessments typically treat forests homogeneously. Here, we grade forest exclusions by ecological value and couple them with a high-resolution net-zero electricity system model, representing onshore wind on forested and non-forested land as two distinct technologies. Applying a set of ecologically graded scenarios to Norway, we quantify how forest protection by its ecological value reshapes system design, the balance of onshore wind within and outside forests, and the robustness of these outcomes to complementary energy policy levers. At equal cost, the system treats forest as its preferred onshore resource, with forest wind comprising 60 to 90 percent of onshore capacity under permissive scenarios. Excluding forest raises system cost by up to 0.5 percent per GW of foregone forest wind, with diminishing returns as ecological criteria for forest exclusions tighten. Directing wind into forests does not compromise system efficiency, with onshore curt
Secondary forests now dominate tropical landscapes and play a crucial role in achieving COP15 conservation objectives. This study develops a replicable national approach to identifying and characterising forest ecosystems, with a focus on the role of secondary forests. We hypothesised that dominant tree species in the forest canopy serve as reliable indicators for delineating forest ecosystems and untangling biodiversity complexity. Using national inventories, we identified in situ clusters through hierarchical clustering based on dominant species abundance dissimilarity, determined using the Importance Variable Index. These clusters were characterised by analysing species assemblages and their interactions. We then applied object-oriented Random Forest modelling, segmenting the national forest cover using NDVI to identify the forest ecosystems derived from in situ clusters. Freely available spectral (Sentinel-2) and environmental data were used in the model to delineate and characterise key forest ecosystems. We finished with an assessment of distribution of secondary and old-growth forests within ecosystems. In Costa Rica, 495 dominant tree species defined 10 in situ clusters, wi
Developing accurate and reliable models for forest types mapping is critical to support efforts for halting deforestation and for biodiversity conservation (such as European Union Deforestation Regulation (EUDR)). This work introduces ForTy, a benchmark for global-scale FORest TYpes mapping using multi-temporal satellite data1. The benchmark comprises 200,000 time series of image patches, each consisting of Sentinel-2, Sentinel-1, climate, and elevation data. Each time series captures variations at monthly or seasonal cadence. Per-pixel annotations, including forest types and other land use classes, support image segmentation tasks. Unlike most existing land use products that often categorize all forest areas into a single class, our benchmark differentiates between three forest types classes: natural forest, planted forest, and tree crops. By leveraging multiple public data sources, we achieve global coverage with this benchmark. We evaluate the forest types dataset using several baseline models, including convolution neural networks and transformer-based models. Additionally, we propose a novel transformer-based model specifically designed to handle multi-modal, multi-temporal sa
The increasing availability of high-resolution satellite imagery, together with advances in deep learning, creates new opportunities for forest monitoring workflows. Two central challenges in this domain are pixel-level change detection and semantic change interpretation, particularly for complex forest dynamics. While large language models (LLMs) are increasingly adopted for data exploration, their integration with vision-language models (VLMs) for remote sensing image change interpretation (RSICI) remains underexplored, especially beyond urban environments. This paper introduces Forest-Chat, an LLM-driven agent for forest change analysis, enabling natural language querying across multiple RSICI tasks, including change detection and captioning, object counting, deforestation characterisation, and change reasoning. Forest-Chat builds upon a multi-level change interpretation (MCI) vision-language backbone with LLM-based orchestration, incorporating zero-shot change detection via AnyChange and multimodal LLM-based zero-shot change captioning and refinement. To support adaptation and evaluation in forest environments, we introduce the Forest-Change dataset, comprising bi-temporal sate
Regression forests have long delivered state-of-the-art accuracy, often outperforming regression trees and even neural networks, but they suffer from limited interpretability as ensemble methods. In this work, we revisit forest pruning, an approach that aims to have the best of both worlds: the accuracy of regression forests and the interpretability of regression trees. This pursuit, whose foundation lies at the core of random forest theory, has seen vast success in empirical studies. In this paper, we contribute theoretical results that support and qualify those empirical findings; namely, we prove the asymptotic advantage of a Lasso-pruned forest over its unpruned counterpart under weak assumptions, as well as high-probability finite-sample generalization bounds for regression forests pruned according to the main methods, which we then validate by way of simulation. Then, we test the accuracy of pruned regression forests against their unpruned counterparts on 19 different datasets (16 synthetic, 3 real). We find that in the vast majority of scenarios tested, there is at least one forest-pruning method that yields equal or better accuracy than the original full forest (in expectat
This study investigates the application of Nash equilibrium strategies in optimizing forest harvesting decisions, focusing on multiple management objectives in forestry. Through simulation-based analysis, the research explores the evolution of various indicators during the game: 1) the mass of CO2 sequestration, 2) forest stands biodiversity, 3) the harvested wood volume, 4) native species fraction, and 5) protective functions. The results underscore the importance of considering diverse objectives and balancing competing interests in forestry decision processes. The forest stands designated for harvesting in the Toruń Forest District were defined as the initial strategy, and indicators for all objectives were calculated accordingly. A Nash equilibrium was identified through a game involving five players representing individual objectives with partially conflicting aims. The final strategy was obtained by modifying specific forest stands designated for harvesting, thereby maintaining the planned wood volume extraction while simultaneously reducing biodiversity loss by nearly 40%, preserving protective functions across over 600 hectares of forested areas, enhancing decadal carbon se
Raspaud and Wang conjectured that every triangle-free planar graph can be vertex-partitioned into an independent set and a forest. Independently, Kawarabayashi and Thomassen also remarked that this might be true, after providing another proof of a result of Borodin and Glebov, showing this result for planar graphs of girth~5. Subsequently, Dross, Montassier, and Pinlou raised the same question and proved that every triangle-free planar graph can be partitioned into a forest and another forest of maximum degree~5. More recently, Feghali and Šámal improved this bound on the maximum degree to~3. In this note, we further improve the result by showing that every triangle-free planar graph can be partitioned into a forest and a linear forest, that is, a forest of maximum degree~2.
Several key states in various regions of the U.S. have experienced recent sawtimber as well as pulp and paper mill closures, which raises an important policy question: how have and will key macroeconomic and industry specific indicators within the U.S. forest sector likely to change over time? This study provides empirical evidence to support forest-sector policy design by using a vector error correction (VEC) model to forecast economic trends in three major industries - forestry and logging, wood manufacturing, and paper manufacturing - across six of the most forest-dependent states found by the location quotient (LQ) measure: Alabama, Arkansas, Maine, Mississippi, Oregon, and Wisconsin. Overall, the results suggest a general decline in employment and the number of firms in the forestry and logging industry as well as the paper manufacturing industry, while wood manufacturing is projected to see modest employment gains. These results also offer key insights for regional policymakers, industry leaders, and local economic development officials: communities dependent on timber-based manufacturing may be more resilient than other forestry-based industries in the face of economic disru
Understanding historical forest dynamics, specifically changes in forest biomass and carbon stocks, has become critical for assessing current forest climate benefits and projecting future benefits under various policy, regulatory, and stewardship scenarios. Carbon accounting frameworks based exclusively on national forest inventories are limited to broad-scale estimates, but model-based approaches that combine these inventories with remotely sensed data can yield contiguous fine-resolution maps of forest biomass and carbon stocks across landscapes over time. Here we describe a fundamental step in building a map-based stock-change framework: mapping historical forest biomass at fine temporal and spatial resolution (annual, 30m) across all of New York State (USA) from 1990 to 2019, using freely available data and open-source tools. Using Landsat imagery, US Forest Service Forest Inventory and Analysis (FIA) data, and off-the-shelf LiDAR collections we developed three modeling approaches for mapping historical forest aboveground biomass (AGB): training on FIA plot-level AGB estimates (direct), training on LiDAR-derived AGB maps (indirect), and an ensemble averaging predictions from th
Let $G$ be a graph on $n$ vertices. For $i\in \{0,1\}$ and a connected graph $G$, a spanning forest $F$ of $G$ is called an $i$-perfect forest if every tree in $F$ is an induced subgraph of $G$ and exactly $i$ vertices of $F$ have even degree (including zero). A $i$-perfect forest of $G$ is proper if it has no vertices of degree zero. Scott (2001) showed that every connected graph with even number of vertices contains a (proper) 0-perfect forest. We prove that one can find a 0-perfect forest with minimum number of edges in polynomial time, but it is NP-hard to obtain a 0-perfect forest with maximum number of edges. Moreover, we show that to decide whether $G$ has a 0-perfect forest with at least $|V(G)|/2+k$ edges, where $k$ is the parameter, is W[1]-hard. We also prove that for a prescribed edge $e$ of $G,$ it is NP-hard to obtain a 0-perfect forest containing $e,$ but one can decide if there existsa 0-perfect forest not containing $e$ in polynomial time. It is easy to see that every graph with odd number of vertices has a 1-perfect forest. It is not the case for proper 1-perfect forests. We give a characterization of when a connected graph has a proper 1-perfect forest.
In this paper we propose using the principle of boosting to reduce the bias of a random forest prediction in the regression setting. From the original random forest fit we extract the residuals and then fit another random forest to these residuals. We call the sum of these two random forests a \textit{one-step boosted forest}. We show with simulated and real data that the one-step boosted forest has a reduced bias compared to the original random forest. The paper also provides a variance estimate of the one-step boosted forest by an extension of the infinitesimal Jackknife estimator. Using this variance estimate we can construct prediction intervals for the boosted forest and we show that they have good coverage probabilities. Combining the bias reduction and the variance estimate we show that the one-step boosted forest has a significant reduction in predictive mean squared error and thus an improvement in predictive performance. When applied on datasets from the UCI database, one-step boosted forest performs better than random forest and gradient boosting machine algorithms. Theoretically we can also extend such a boosting process to more than one step and the same principles out
Ensembles of randomized decision trees, usually referred to as random forests, are widely used for classification and regression tasks in machine learning and statistics. Random forests achieve competitive predictive performance and are computationally efficient to train and test, making them excellent candidates for real-world prediction tasks. The most popular random forest variants (such as Breiman's random forest and extremely randomized trees) operate on batches of training data. Online methods are now in greater demand. Existing online random forests, however, require more training data than their batch counterpart to achieve comparable predictive performance. In this work, we use Mondrian processes (Roy and Teh, 2009) to construct ensembles of random decision trees we call Mondrian forests. Mondrian forests can be grown in an incremental/online fashion and remarkably, the distribution of online Mondrian forests is the same as that of batch Mondrian forests. Mondrian forests achieve competitive predictive performance comparable with existing online random forests and periodically re-trained batch random forests, while being more than an order of magnitude faster, thus represe
We study approximation algorithms for the forest cover and bounded forest cover problems. A probabilistic $2+ε$ approximation algorithm for the forest cover problem is given using the method of dual fitting. A deterministic algorithm with a 2-approximation ratio that rounds the optimal solution to a linear program is given next. The 2-approximation for the forest cover is then used to give a 6-approximation for the bounded forest cover problem. The use of the probabilistic method to develop the $2+ε$ approximation algorithm may be of independent interest.
In this paper, we introduce the notion of Cartesian Forest, which generalizes Cartesian Trees, in order to deal with partially ordered sequences. We show that algorithms that solve both exact and approximate Cartesian Tree Matching can be adapted to solve Cartesian Forest Matching in average linear time. We adapt the notion of Cartesian Tree Signature to Cartesian Forests and show how filters can be used to experimentally improve the algorithm for the exact matching. We also show a one to one correspondence between Cartesian Forests and Schröder Trees.
A new approach called NAF (the Neural Attention Forest) for solving regression and classification tasks under tabular training data is proposed. The main idea behind the proposed NAF model is to introduce the attention mechanism into the random forest by assigning attention weights calculated by neural networks of a specific form to data in leaves of decision trees and to the random forest itself in the framework of the Nadaraya-Watson kernel regression. In contrast to the available models like the attention-based random forest, the attention weights and the Nadaraya-Watson regression are represented in the form of neural networks whose weights can be regarded as trainable parameters. The first part of neural networks with shared weights is trained for all trees and computes attention weights of data in leaves. The second part aggregates outputs of the tree networks and aims to minimize the difference between the random forest prediction and the truth target value from a training set. The neural network is trained in an end-to-end manner. The combination of the random forest and neural networks implementing the attention mechanism forms a transformer for enhancing the forest predic
The Turán number of a graph $H$, denoted by $ex(n,H)$, is the maximum number of edges in any graph on $n$ vertices containing no $H$ as a subgraph. A linear (star) forest is a forest consisting of paths (stars). A path-star forest $F$ is a forest consisting of paths and stars. In this paper, we determine $ex(n,F)$ for sufficiently large $n$ and characterize the corresponding extremal graphs, and our result generalizes previous known results on the Turán numbers of linear forests and star forests.
Forests will have two notable economic roles in the future: providing renewable raw material and storing carbon to mitigate climate change. The pricing of forest carbon leads to longer rotation times and consequently larger carbon stocks, but also exposes landowners to a greater risk of forest damage. This paper investigates optimal forest rotation under carbon pricing and forest damage risk. I provide the optimality conditions for this problem and illustrate the setting with numerical calculations representing boreal forests under a range of carbon prices and damage probabilities. The relation between damage probability and carbon price towards the optimal rotation length is nearly linear, with carbon pricing having far greater impact. As such, increasing forest carbon stocks by lengthening rotations is an economically attractive method for climate change mitigation, despite the forest damage risk. Carbon pricing also increases land expectation value and reduces the economic risks of the landowner. The production possibility frontier under optimal rotation suggests that significantly larger forests carbon stocks are achievable, but imply lower harvests. However, forests' societall
The first year of data from the Dark Energy Spectroscopic Instrument (DESI) contains the largest set of Lyman-$α$ (Ly$α$) forest spectra ever observed. This data, collected in the DESI Data Release 1 (DR1) sample, has been used to measure the Baryon Acoustic Oscillation (BAO) feature at redshift $z=2.33$. In this work, we use a set of 150 synthetic realizations of DESI DR1 to validate the DESI 2024 Ly$α$ forest BAO measurement. The synthetic data sets are based on Gaussian random fields using the log-normal approximation. We produce realistic synthetic DESI spectra that include all major contaminants affecting the Ly$α$ forest. The synthetic data sets span a redshift range $1.8<z<3.8$, and are analysed using the same framework and pipeline used for the DESI 2024 Ly$α$ forest BAO measurement. To measure BAO, we use both the Ly$α$ auto-correlation and its cross-correlation with quasar positions. We use the mean of correlation functions from the set of DESI DR1 realizations to show that our model is able to recover unbiased measurements of the BAO position. We also fit each mock individually and study the population of BAO fits in order to validate BAO uncertainties and test our
Large-scale delineation of individual trees from remote sensing imagery is crucial to the advancement of ecological research, particularly as climate change and other environmental factors rapidly transform forest landscapes across the world. Current RGB tree segmentation methods rely on training specialized machine learning models with labeled tree datasets. While these learning-based approaches can outperform manual data collection when accurate, the existing models still depend on training data that's hard to scale. In this paper, we investigate the efficacy of using a state-of-the-art image segmentation model, Segment Anything Model 2 (SAM2), in a zero-shot manner for individual tree detection and segmentation. We evaluate a pretrained SAM2 model on two tasks in this domain: (1) zero-shot segmentation and (2) zero-shot transfer by using predictions from an existing tree detection model as prompts. Our results suggest that SAM2 not only has impressive generalization capabilities, but also can form a natural synergy with specialized methods trained on in-domain labeled data. We find that applying large pretrained models to problems in remote sensing is a promising avenue for futu
Changes in the frequency, duration, and severity of climate extremes are forecast to occur under global climate change. The impacts of climate extremes on forest productivity and health are complicated by potential interactions with disturbance events and stand dynamics. The effects of stand dynamics on forest responses to climate and disturbance are particularly important given forest characteristics driven by stand dynamics can be modified through forest management with the goal of increasing forest resistance and resilience to climate change. We develop a hierarchical Bayesian state-space model allowing climate effects on tree growth to vary over time and in relation to climate extremes, disturbance events, and stand dynamics. We apply the model to a dendrochronology dataset comprising measurements from forest stands of varying composition, structure, and development stage in northeastern Minnesota. Results indicate average forest growth was most sensitive to variables describing climatic water deficit. Forest growth responses to water deficit were partitioned into responses driven by climatic threshold exceedances and interactions with forest tent caterpillar defoliation. Fores