arXiv subjects

Scott R. Saleska

Publications and source records attributed to Scott R. Saleska.

4 recordsLinked to original sources

arXiv2026

TerraMARS: A Domain-Adapted Small-Language-Model Pipeline for Mars Terraforming Literature

Researchers are interested in learning about Mars so that it may eventually become habitable for humans. To achieve this, there is a need for comprehensive knowledge of the planet's atmosphere, hydrology, surface chemistry, radiation environment, and spatial features through the scientific literature. These contain valuable information and meaningful quantitative constraints that can be used in other models and studies, such as habitability assessment and future terraforming studies. We present TerraMARS, an end-to-end information extraction pipeline that combines a domain-adapted Small Language Model to answer Mars terraforming-related questions and convert unstructured Mars science text into machine-readable structured outputs in JavaScript Object Notation (JSON) format. A corpus of open-access papers is collected and processed using a multistage retrieval and chunking framework. Google Gemma 3 1B was adapted to the domain using Quantized Low-Rank Adaptation (QLoRA) fine-tuning on Mars-specific question-answering and information extraction datasets. The resulting pipeline generates both types of output and provides a foundation for integrating knowledge from scientific literature into downstream applications like digital twins and habitability modeling for Mars. The output from this pipeline looks promising, but further improvements are needed to increase extraction accuracy and factual consistency.

cs.CL↗

arXiv2025

Seeing Forests Through Clouds: Comment on "Recent global temperature surge amplified by record-low planetary albedo" (arXiv:2405.19986)

Goessling et al. (1) link the record-breaking warming anomaly of 2023 to a global albedo decline due to reduced low-level cloud cover. What caused the reduction remains unclear. Goessling et al. considered several geophysical mechanisms, including ocean surface warming and declining aerosol emissions, but did not discuss the biosphere. We propose that disruption of global biospheric functioning could be a cause, as supported by three lines of evidence that have not yet been jointly considered.

physics.ao-ph↗

arXiv2023

The role of ecosystem transpiration in creating alternate moisture regimes by influencing atmospheric moisture convergence

The terrestrial water cycle links the soil and atmosphere moisture reservoirs through four fluxes: precipitation, evaporation, runoff, and atmospheric moisture convergence (net import of water vapor to balance runoff). Each of these processes is essential for human and ecosystem well-being. Predicting how the water cycle responds to changes in vegetation cover remains a challenge. Recently, changes in plant transpiration across the Amazon basin were shown to be associated disproportionately with changes in rainfall, suggesting that even small declines in transpiration (e.g., from deforestation) would lead to much larger declines in rainfall. Here, constraining these results by the law of mass conservation, we show that in a sufficiently wet atmosphere, forest transpiration can control atmospheric moisture convergence such that increased transpiration enhances atmospheric moisture import and resulting water yield. Conversely, in a sufficiently dry atmosphere increased transpiration reduces atmospheric moisture convergence and water yield. This previously unrecognized dichotomy can explain the otherwise mixed observations of how water yield responds to re-greening, as we illustrate with examples from China's Loess Plateau. Our analysis indicates that any additional precipitation recycling by additional vegetation increases precipitation but decreases local water yield and steady-state runoff. Therefore, in the drier regions and early stages of ecological restoration, the role of vegetation can be confined to precipitation recycling, while once a wetter stage is achieved, additional vegetation enhances atmospheric moisture convergence and water yield. Evaluating the transition between regimes, and recognizing the potential of vegetation for enhancing moisture convergence, are crucial for characterizing the consequences of deforestation and for motivating and guiding ecological restoration.

physics.ao-ph↗

arXiv2022

How transpiration by forests and other vegetation determines alternate moisture regimes

The terrestrial water cycle links the soil and atmosphere moisture reservoirs through four fluxes: precipitation, evaporation, runoff and atmospheric moisture convergence. Each of these fluxes is essential for human and ecosystem well-being. However, predicting how the water cycle responds to changes in vegetation cover, remains a challenge (Lawrence and Vandecar, 2015; Ellison et al., 2017; te Wierik et al., 2021). Recently, rainfall was shown to decrease disproportionally with reduced forest transpiration following deforestation (Baudena et al., 2021). Here, combining these findings with the law of matter conservation, we show that in a sufficiently wet atmosphere forest transpiration can control atmospheric moisture convergence such that increased transpiration enhances atmospheric moisture import. Conversely, in a drier atmosphere increased transpiration reduces atmospheric moisture convergence and runoff. This previously unrecognized dichotomy can explain the seemingly random observations of runoff and soil moisture sometimes increasing and sometimes reducing in response to re-greening (e.g., Zheng et al., 2021). Evaluating the transition between the two regimes is crucial both for characterizing the risk posed by deforestation as well as for motivating and guiding global ecosystem restoration.

physics.ao-ph↗