Merge pull request #83 from TUDelftGeodesy/joss_improvement

add more sections

Author: rogerkuou

paper/paper.bib

@@ -82,3 +82,39 @@ @article{moreira2013tutorial
   year={2013},
   publisher={IEEE}
 }
+
+@article{LUMBANGAOL2025117551,
+title = {Peat subsidence and dynamics in Midden-Delfland, the Netherlands, from time series InSAR analysis and the SPAMS model},
+journal = {Geoderma},
+volume = {463},
+pages = {117551},
+year = {2025},
+issn = {0016-7061},
+doi = {https://doi.org/10.1016/j.geoderma.2025.117551},
+url = {https://www.sciencedirect.com/science/article/pii/S0016706125003921},
+author = {Yustisi Lumban-Gaol and Philip Conroy and Simon {van Diepen} and Freek {van Leijen} and Ramon Hanssen},
+keywords = {InSAR, Pastures, Peatlands, SPAMS, Subsidence},
+abstract = {Peat subsidence occurs when parts of the peat soil interact with air, usually due to water table lowering, then triggers peat consolidation, shrinkage, and oxidation, releasing substantial CO2 emissions. Managing and mitigating these impacts requires a comprehensive understanding of the mechanisms and the spatio-temporal variations of the subsidence. Advanced space geodetic techniques, particularly InSAR, enable surface displacement monitoring. While time series InSAR analysis effectively estimates displacement, its precision, accuracy, and representativity are compromised by temporal decorrelation, noise, and dynamic soil movement, especially over pastures on peat soils. Moreover, loss-of-lock events caused by an irrecoverable loss of coherence disrupt the time series and introduce arbitrary unintelligible phase offsets. Strategies such as multilooking using contextual information have improved the reliability of the InSAR displacement estimates. However, more experience in the efficacy of InSAR-based surface dynamics assessments is required. This study estimates and analyzes surface motion in a regional peat area in Midden-Delfland, The Netherlands, using Sentinel-1 data and the SPAMS model. SPAMS incorporates precipitation and evapotranspiration information to estimate surface motion parameters, distinguishing between reversible and irreversible subsidence. The results reveal an average subsidence rate of −5.4±0.7 mm/year within the study area. Irreversible subsidence is strongly correlated with climatic conditions, with the most significant subsidence observed during a prolonged dry period in the summers of 2018 and 2022. Mitigating peatland subsidence includes preserving soil water content, especially during dry periods. Integrating InSAR and SPAMS provides a valuable tool for monitoring peat surface elevation, water management, and reducing peatland degradation.}
+}
+
+@ARTICLE{brouwer11131159,
+author={Brouwer, Wietske S. and Hanssen, Ramon F.},
+journal={IEEE Transactions on Geoscience and Remote Sensing}, 
+title={On the Definition of an Independent Stochastic Model for InSAR Time Series}, 
+year={2025},
+volume={63},
+number={},
+pages={1-11},
+keywords={Stochastic processes;Noise;Market research;Mathematical models;Bars;Vectors;Time series analysis;Estimation;Covariance matrices;Reliability;Interferometry SAR (InSAR);parameter estimation;stochastic model;surface displacements},
+doi={10.1109/TGRS.2025.3600893}}
+
+@INPROCEEDINGS{conroy10642504,
+  author={Conroy, Philip and Lumban-Gaol, Yustisi and van Diepen, Simon and van Leijen, Freek and Hanssen, Ramon F.},
+  booktitle={IGARSS 2024 - 2024 IEEE International Geoscience and Remote Sensing Symposium}, 
+  title={First Wide-Area Dutch Peatland Subsidence Estimates Based on InSAR}, 
+  year={2024},
+  volume={},
+  number={},
+  pages={10732-10735},
+  keywords={Heart;Soil measurements;Atmospheric modeling;Time series analysis;Stars;Soil;Predictive models},
+  doi={10.1109/IGARSS53475.2024.10642504}}

paper/paper.md

@@ -28,7 +28,7 @@ bibliography: paper.bib
 
 ## Summary
 
-Satellite-based Synthetic Aperture Radar (SAR) provides invaluable data for Earth Observation. The Interferometric SAR (InSAR) technique [@hanssen01], which utilizes a stack of SAR images in Single Look Complex (SLC) format, plays a significant role in various surface motion monitoring applications, e.g. civil-infrastructure stability [@chang2014detection; @chang2017railway;ozer2018applicability], and hydrocarbons extraction [@fokker2016application; @ZHANG2022102847]. To facilitate advanced data processing for InSAR communities, we developed `SARXarray`, a Xarray extension for handling SLC SAR stacks. 
+Satellite-based Synthetic Aperture Radar (SAR) provides invaluable data for Earth Observation. The Interferometric SAR (InSAR) technique [@hanssen01], which utilizes a stack of SAR images in Single Look Complex (SLC) format, plays a significant role in various surface motion monitoring applications, e.g. civil-infrastructure stability [@chang2014detection; @chang2017railway; @ozer2018applicability], and hydrocarbons extraction [@fokker2016application; @ZHANG2022102847]. To facilitate advanced data processing for InSAR communities, we developed `SARXarray`, a Xarray extension for handling co-registered SLC SAR stacks. 
 
 ## Statement of Need
 
@@ -38,6 +38,23 @@ To facilitate efficient processing of SLC SAR stacks and minimize code customiza
 
 `SARXarray` leverages two well-established Python libraries `Xarray` and `Dask` from the [Pangeo community](https://www.pangeo.io/). It utilizes Xarray’s support on labeled multi-dimensional datasets to stress the space-time character of an SLC SAR stack. `Dask` is used to perform lazy evaluation of operations and block-wise computations. SARXarray can be integrated into existing Python workflows of InSAR processing and deployed on a variety of computational infrastructures. 
 
+## State of the field
+
+A similar open-source library [`xarray-sentinel`](https://github.com/bopen/xarray-sentinel) exists for handling raw Sentinel-1 GRD and SLC data as lazy-loaded Xarray Datasets. Despite the similar goals of digesting SAR data into lazy-loaded Xarray Datasets, `SARXarray` and `xarray-sentinel` are designed for different applications:
+
+- `SARXarray` is designed to handle co-registered SLC stacks, instead of raw Sentinel-1 data. It supports the output of two common SLC co-registration tools: [`DORIS`](https://doris.tudelft.nl/) and [`SNAP`](https://step.esa.int/main/download/snap-download/), instead of raw Sentinel-1 data products from European Space Agency (ESA).
+- `SARXarray` supports interferometric stacks from other sensors than Sentinel-1, as long as they can be co-registered by SNAP or DORIS. It reads the output from the two co-registration tools, and relies on them to handle the specificities of different sensors.
+
+## Software design
+
+`SARXarray` is designed as an extension of Xarray using accessors. This design is motivated by [Xarray community’s recommendation](https://docs.xarray.dev/en/stable/internals/extending-xarray.html), in order to isolate the extension from API changes of the core Xarray library. The software has three main components: an I/O module that lazily loads binary SLCs and related metadata, a Stack accessor that provides SAR-specific operations (amplitude/phase extraction, multi-looking, Mean-Reflection-Map generation, coherence calculation), and utility functions.
+
+## Research impact statement
+
+`SARXarray` is a dependency of [CAROLINE](https://github.com/TUDelftGeodesy/caroline) (Contextual and Autonomous processing of satellite Radar Observations for Learning and Interpreting the Natural and built Environment), which is an InSAR processing framework developed by the InSAR group of Delft University of Technology. It has facilitated the data processing of multiple InSAR research projects [@LUMBANGAOL2025117551; @brouwer11131159; @conroy10642504].  
+
+`SARXarray` is a recognized related project of the Xarray ecosystem, see the [Xarray user-guide page](https://docs.xarray.dev/en/stable/user-guide/extensions.html).
+
 ## Tutorial
 
 We provide a tutorial as a Jupyter notebook to demonstrate the functionalities of `SARXarray`:
@@ -60,4 +77,10 @@ The authors express sincere gratitude to the Dutch Research Council (Nederlandse
 
 We would also like to thank Dr. Francesco Nattino and Dr. Meiert Willem Grootes of the Netherlands eScience Center for the insightful discussions, which are important contributions to this work.
 
+## AI usage disclosure
+
+In writing the software documentation and the JOSS paper, GPT-5 was used for language improvements for grammar correction and style edits.
+
+Additionally, various language models were used via [GitHub Copilot](https://copilot.github.com/) in Pull Requests reviews, for pre-filtering obvious issues such as typos, small logical errors, and for suggesting code improvements. All suggestions from AI were reviewed and verified by the authors before merging into the codebase. The correctness of the code was ensured by the unit tests.
+
 ## References