| 21 Jul 2026
The use of remote sensing for studying environmental changes in a palsa mire zone of Northern Finland in the period 1985–2024
Abstract. The palsa mires – hummocks with a permafrost core – of northern Finland are among the southernmost periglacial features in the Northern Hemisphere and mark the southern boundary of permafrost range. The rapid warming of circumpolar regions demands a better understand of the complex of environmental changes in the marginal permafrost zone. The study aims to investigate changes in selected environmental features (vegetation, water conditions and land moisture) in northern Finland in the years 1985–2024 using Landsat satellite data, with particular emphasis on the differences taking place in palsa mires. The calculated spectral indices NDVI, NDWI, MNDWI, NDMI and LST were analysed in relation to each other and to environmental data. The analysis of changes in spectral indices over time showed a small but systematic increase in the vegetation index value, an increase in soil moisture, and varied responses in indices related to surface water. Principal component analysis (PCA), linear mixed-effects models (LMM) and redundancy analysis (RDA) indicated the multidimensional structure of remote sensing indices and revealed the main environmental factors determining changes in vegetation and moisture in the study area. One of the main predictors differentiating the spectral response was determined to be palsa mires, which differed from mineral soils, but also from other organic soils. The results confirmed the effectiveness of RS and multispectral indices based on Landsat data for studying environmental changes, while indicating the need for multidimensional analyses that include various indicators and local environmental predictors.
General comments
The study investigates long-term environmental changes in a palsa mire zone of Northern Finland using a 40-year Landsat time series (1985–2024) combined with environmental predictors and a range of statistical approaches, including mixed-effects models, PCA and RDA. The topic is timely and relevant given the rapid changes occurring in sub-Arctic permafrost regions and the increasing need for long-term monitoring approaches. The study benefits from a relatively long observation period, and an attempt to integrate multiple remote-sensing indicators describing vegetation, moisture, water conditions and surface temperature.
Overall, the study provides potentially valuable insights into environmental dynamics in a palsa mire region and demonstrates the usefulness of Landsat data for long-term environmental monitoring. However, I found that the central scientific message is currently obscured by the large number of analyses and the somewhat unclear connection between research objectives, methods, results and interpretation. Addressing the comments below would, in my opinion, substantially improve the clarity and impact of the manuscript.
Major comments
1. Research questions, study rationale and analytical framework need to be clarified
The scientific rationale of the study remains somewhat difficult to follow. While the Introduction outlines the general objective of analysing environmental change in a palsa mire zone using Landsat data, the specific research questions are not explicitly presented until the Figure 3. As currently written, the manuscript gives the impression that a broad set of indices and statistical analyses were first applied to the dataset and that the specific questions were formulated afterwards. I encourage the authors to explicitly formulate the main research questions (and ideally hypotheses or expectations) at the end of the Introduction and then structure the Results, Discussion and Conclusions around answering those questions.
2. Results section is difficult to follow and would benefit from a clearer hierarchy
The study contains many different analyses (correlations, PCA, LMM, post-hoc tests, period comparisons, regression models and RDA), which in turn leads to a large number of different results. This makes the Results section “heavy”, and it is difficult for the reader to identify the main findings of the study and link them to the research objectives and questions.
Several analyses appear to demonstrate similar patterns, particularly the separation between soil categories and the importance of soil type as a predictor. While all analyses may be statistically valid, the manuscript would benefit from a clearer hierarchy of findings and a stronger focus on the key messages emerging from the dataset. The results should be organised in the order of the research questions (RQs 1–4) and remove results that are not clearly contributing to them.
In addition, parts of the Results section contain already some ecological interpretation and speculation regarding permafrost degradation, hydrological transitions and ecosystem change. These interpretations of the results would fit better within the Discussion section. Keep the Results section as concise as possible, just report the relevant findings.
Moreover, the manuscript would benefit from a more selective use of supplementary material. The extensive number of supplementary figures and tables contributes to the overall complexity of the study and reinforces the impression that numerous analyses were conducted to describe broadly similar patterns. While supplementary material is important for transparency, the current structure requires frequent movement between the main text and the Supplement, making the manuscript difficult to follow. The authors may wish to reconsider which analyses are essential for the main narrative and whether some supplementary content could be consolidated, reduced, or integrated into the main manuscript.
3. The role of PCA, RDA and LST within the analytical framework should be clarified
The conceptual role of PCA, RDA and LST within the overall analytical framework could be explained more clearly. PCA based on multi-year means and RDA performed for three temporal periods appear to identify broadly similar gradients and highlight the same dominant explanatory factors, particularly soil category and topographic controls.
Similarly, the added value of LST remains somewhat unclear. A substantial part of the manuscript is devoted to discussing the limitations and interpretational challenges of Landsat-derived LST, yet the ecological conclusions derived from LST are relatively modest. The authors should more clearly explain what additional insight these analyses provide beyond the vegetation and moisture indices and beyond the mixed-effects modelling framework already applied.
4. Interpretation of permafrost degradation should be more cautious
The manuscript frequently links observed changes in RS indices to permafrost degradation and palsa deterioration. While these interpretations are plausible and consistent with current understanding of palsa dynamics, the study does not include direct measurements of permafrost conditions, active-layer thickness, palsa extent, or geomorphological change. The results therefore provide indirect evidence consistent with permafrost degradation rather than demonstrating it directly. The distinction between observation and interpretation should be communicated more clearly throughout the manuscript.
5. Discussion is sometimes weakly linked to the study results
Several parts of the Discussion read more like a literature review than an interpretation of the results presented in this study. Extensive sections discuss Arctic warming, permafrost degradation, and climate feedback, but the links between these topics and the actual findings of the study are not always sufficiently discussed. Therefore, I strongly recommend restructuring the Discussion around the main findings of the study and explicitly discussing how each result supports, challenges or complements previous studies and how they answer to the set research questions.
Minor comments
1. Because most palsa features are considerably smaller than Landsat pixels, greater attention should be given to the implications of mixed-pixel effects and the extent to which the observed signal represents the palsa itself versus the surrounding mire environment. In addition, the areal change of palsas during the 40-year study period should be discussed more and the trends observed in the study should be reflected on it.
2. I was left wondering about the rationale for the selection of the 245 sampling points. A brief explanation of how the points were chosen (e.g. random, stratified, or systematic sampling) would help readers better understand the representativeness of the dataset.
3. The description of data availability and year exclusions appears inconsistent across the manuscript and Supplement. Different sections refer to different years as lacking suitable imagery or being omitted from the analyses (e.g. 1999 and 2012 in Section 2.2, 2010 and 2015 in Section 2.3, and 1992, 1999 and 2008 in the Supplement). I recommend clarifying the image selection workflow and explicitly listing which years were excluded from the final analyses and why.
4. The environmental predictors play a central role in the statistical analyses and interpretation of the results. While their derivation is described in the Supplement, the manuscript would benefit from a slightly more detailed introduction to the selected predictors and the rationale for their inclusion in the main text. Now they are only shortly mentioned at the end of the Statistical methods -section while climate and RS data have their own sections.
5. Consider reorganising Figure 3 so that the workflow progresses from research questions to data sources and finally to analytical methods. This would make the structure of the study easier to follow.
6. The manuscript would benefit from more consistent use of terminology throughout.
7. The description of the study area could be streamlined. In particular, the detailed presentation of bedrock geology appears only weakly connected to the subsequent analyses and interpretation. Consider reducing this section and focusing more on environmental characteristics directly relevant to the research questions.
8. Consider defining abbreviations in figure and table captions, even if they have already been introduced in the main text. This would improve the readability of individual figures and tables and make them easier to interpret as stand-alone elements.
Small corrections and suggestions: