|
| 1 | +@article{legrand2024, |
| 2 | + bibtex_show = {true}, |
| 3 | + author = {Legrand, Juliette and Pimont, François and Dupuy, Jean-Luc |
| 4 | + and Opitz, Thomas}, |
| 5 | + publisher = {French Statistical Society}, |
| 6 | + title = {Bayesian Spatiotemporal Modelling of Wildfire Occurrences and |
| 7 | + Sizes for Projections Under Climate Change}, |
| 8 | + journal = {Computo}, |
| 9 | + year = 2024, |
| 10 | + url = {https://computo.sfds.asso.fr/published-202407-legrand-wildfires/}, |
| 11 | + doi = {10.57750/4y84-4t68}, |
| 12 | + issn = {2824-7795}, |
| 13 | + type = {{Research article}}, |
| 14 | + domain = {Statistics}, |
| 15 | + language = {R}, |
| 16 | + repository = {published-202407-legrand-wildfires}, |
| 17 | + langid = {en}, |
| 18 | + abstract = {Appropriate spatiotemporal modelling of wildfire activity |
| 19 | + is crucial for its prediction and risk management. Here, we focus on |
| 20 | + wildfire risk in the Aquitaine region in the Southwest of France and |
| 21 | + its projection under climate change. We study whether wildfire risk |
| 22 | + could further increase under climate change in this specific region, |
| 23 | + which does not lie in the historical core area of wildfires in |
| 24 | + Southeastern France, corresponding to the Southwest. For this |
| 25 | + purpose, we consider a marked spatiotemporal point process, a |
| 26 | + flexible model for occurrences and magnitudes of such environmental |
| 27 | + risks, where the magnitudes are defined as the burnt areas. The |
| 28 | + model is first calibrated using 14 years of past observation data of |
| 29 | + wildfire occurrences and weather variables, and then applied for |
| 30 | + projection of climate-change impacts using simulations of numerical |
| 31 | + climate models until 2100 as new inputs. We work within the |
| 32 | + framework of a spatiotemporal Bayesian hierarchical model, and we |
| 33 | + present the workflow of its implementation for a large dataset at |
| 34 | + daily resolution for 8km-pixels using the INLA-SPDE approach. The |
| 35 | + assessment of the posterior distributions shows a satisfactory fit |
| 36 | + of the model for the observation period. We stochastically simulate |
| 37 | + projections of future wildfire activity by combining climate model |
| 38 | + output with posterior simulations of model parameters. Depending on |
| 39 | + climate models, spline-smoothed projections indicate low to moderate |
| 40 | + increase of wildfire activity under climate change. The increase is |
| 41 | + weaker than in the historical core area, which we attribute to |
| 42 | + different weather conditions (oceanic versus Mediterranean). Besides |
| 43 | + providing a relevant case study of environmental risk modelling, |
| 44 | + this paper is also intended to provide a full workflow for |
| 45 | + implementing the Bayesian estimation of marked log-Gaussian Cox |
| 46 | + processes using the R-INLA package of the R statistical software.} |
| 47 | +} |
| 48 | + |
| 49 | +@article{pishchagina2024, |
| 50 | + bibtex_show = {true}, |
| 51 | + author = {Pishchagina, Liudmila and Rigaill, Guillem and Runge, |
| 52 | + Vincent}, |
| 53 | + publisher = {French Statistical Society}, |
| 54 | + title = {Geometric-Based {Pruning} {Rules} for {Change} {Point} |
| 55 | + {Detection} in {Multiple} {Independent} {Time} {Series}}, |
| 56 | + journal = {Computo}, |
| 57 | + year = 2024, |
| 58 | + url = {https://computo.sfds.asso.fr/published-202406-pishchagina-change-point/}, |
| 59 | + doi = {10.57750/9vvx-eq57}, |
| 60 | + issn = {2824-7795}, |
| 61 | + type = {{Research article}}, |
| 62 | + domain = {Statistics}, |
| 63 | + language = {R}, |
| 64 | + repository = {published-202406-pishchagina-change-point}, |
| 65 | + langid = {en}, |
| 66 | + abstract = {We address the challenge of identifying multiple change |
| 67 | + points in a group of independent time series, assuming these change |
| 68 | + points occur simultaneously in all series and their number is |
| 69 | + unknown. The search for the best segmentation can be expressed as a |
| 70 | + minimization problem over a given cost function. We focus on dynamic |
| 71 | + programming algorithms that solve this problem exactly. When the |
| 72 | + number of changes is proportional to data length, an |
| 73 | + inequality-based pruning rule encoded in the PELT algorithm leads to |
| 74 | + a linear time complexity. Another type of pruning, called functional |
| 75 | + pruning, gives a close-to-linear time complexity whatever the number |
| 76 | + of changes, but only for the analysis of univariate time series. We |
| 77 | + propose a few extensions of functional pruning for multiple |
| 78 | + independent time series based on the use of simple geometric shapes |
| 79 | + (balls and hyperrectangles). We focus on the Gaussian case, but some |
| 80 | + of our rules can be easily extended to the exponential family. In a |
| 81 | + simulation study we compare the computational efficiency of |
| 82 | + different geometric-based pruning rules. We show that for a small |
| 83 | + number of time series some of them ran significantly faster than |
| 84 | + inequality-based approaches in particular when the underlying number |
| 85 | + of changes is small compared to the data length.} |
| 86 | +} |
| 87 | + |
1 | 88 | @article{susmann_adaptive,
|
2 | 89 | bibtex_show = {true},
|
3 | 90 | author = {Susmann, Herbert and and Chambaz, Antoine and Josse, Julie},
|
|
0 commit comments