1. Introduction
Post-wildfire land cover change affects the ecosystem dynamics, biodiversity system, and ecological processes [
1,
2]. According to the European Commission, land cover indicates “the visible surface of land on Earth (e.g. crops, grass, water, broad-leaved, forest or built-up area)” [
3]. Thus, land cover plays a key role in environmental conditions in multiple ways [
4]. Several studies have proved that fire events change soil characteristics, which increase runoff and erosion risk [
5,
6,
7], plant diversity [
8,
9,
10], habitat and species composition [
11,
12,
13], water quality [
14,
15,
16], climate [
17,
18,
19], and carbon storage [
20,
21]. Detecting the land cover change during a wildfire is critical, as it enables authorities to develop prompt disaster management strategies to prevent fatalities caused by landslides and flash floods following a wildfire [
22]. Furthermore, periodic assessments scrutinizing burned area recovery and vegetation dynamics after a wildfire provide valuable information for land management and biodiversity recovery monitoring.
According to Copernicus, Portugal is one of the countries most vulnerable to wildfire in the European Union [
23]. More than 200 forest fires happened annually in Portugal from 2013 to 2022. Portugal experienced the worst drought in 2017 since 1931, which caused 408 wildfires and involved 563,532 ha of burned area [
24,
25]. The Serra da Estrela Natural Park (PNSE), the largest protected area in Portugal, experienced the most significant fire since 1975 [
26], which involved a burned area of 21,942 ha in 2022 [
27]. This study focused on the major burned area in a wildfire that started on 6 August 2022 and ended on 2 September 2022 in PNSE [
27]. The fire spread beyond the PNSE to Estrela Geopark. As a recognized Biogenetic Reserve and Geopark [
28], the impacts of the large burned area post-wildfire 2022 in the PNSE are concerning. Close monitoring of post-wildfire burned area recovery and vegetation dynamics to observe the ecosystem dynamics and plan for disaster management is imperative.
The open data on land use and land cover (LULC) in Portugal is publicly available in vector format at a scale of 1:25,000 for Land Use and Occupation Map (COS), and in raster format with 10 m pixels for Conjunctural Land Occupation Map (COSc) from the National Geographic Information System (SNIG) [
29]. LULC maps are widely used by various stakeholders for research, analysis, planning, and monitoring purposes [
31,
32,
33]. However, continuous assessment and regular scrutiny of the land cover change are not achievable even with the annual update of the COSc [
29].
Researchers have been exploring the integration of machine learning (ML) algorithms and remote sensing (RS) data since the late 1990s [
34]. It has proven effective in image classification, object detection, and semantic segmentation since the 2010s [
35,
36,
37]. More recently, the integration of ML algorithms and RS data has been widely applied to convert satellite imagery into valuable spatial data for environmental monitoring [
11,
12,
13,
17,
18,
20,
21], urban planning [
19,
38], disaster management [
5,
6,
7], and more. Previous studies proved that integrating ML and RS techniques is more efficient in land cover classification and achieves higher accuracy [
39,
40] than conventional methods [
34].
ML is a subset of artificial intelligence (AI) that imitates intelligent human behavior and learns by trial and error from the database to perform predictions [
41,
42]. Many studies have found Support Vector Machines (SVM) and Random Forest (RF) the most efficient algorithms to be integrated with RS data for land cover classification, compared to other ML algorithms such as Artificial Neural Networks (ANN), K-Nearest Neighbours (K-NN), single decision trees (DTs), and boosted DTs [
32,
34,
39,
40,
43]. RF adopts an ensemble learning method formed by bagging, boosting, and stacking processes [
44]. The advantages of RF are that it is robust to outliers, accurate, and less overfitting [
45]. However, the right hyperparameters are decisive in generating high accuracy [
46]. Also, RF cannot forecast values other than the feature range of the training data [
47]. SVM uses kernel functions to maximize the margin between classes through a support vector and find the best hyperplane (optimal line) to define the best decision boundary for accurate predictions [
48]. SVM is effective for clear class separation, able to handle linear/nonlinear tasks, good for diverse land covers, and memory efficient [
49]. However, SVM can be computationally expensive and not suitable for large datasets [
50].
Object-Based Image Analysis (OBIA) and Pixel-Based Image Analysis (PBIA) are two common approaches for LULC [
38]. OBIA groups the neighbouring pixels with similar spectral properties and contextual information into meaningful objects. OBIA is useful for object identification in addition to pixel labeling. The quality of the object segmentation determines the final classification. Edge-based algorithms, region-based, and superpixel algorithms are common for OBIA segmentation [
51,
52]. PBIA is a traditional RS image classification commonly used in supervised and unsupervised techniques. PBIA classifies individual pixels based on their spectral properties without considering the adjacent pixels. Minimum Distance, Maximum Likelihood, and Spectral Angle Mapping are common algorithms available in many Geographic Information Systems (GIS) software for PBIA approaches [
52,
53,
54].
The LULC maps provided by the SNIG in Portugal are not ideal for continuous assessment and regular scrutiny. This study aims to identify the most appropriate LULC approach and classifier for land cover classification to resolve the challenges faced for continuous assessment in the study area while serving as a reference for other forest regions with similar characteristics.
This study was conducted with specific objectives: i) to identify the best model for land cover classification for the study area employing the integration of LULC classification with classifier on Google Earth Engine (GEE); ii) to compare the effectiveness of the two LULC classification methods: OBIA and PBIA; iii) to compare the accuracy of RF and SVM algorithms for land cover classification using GEE; iv) to assess the accuracy of the land cover classification generated by different models with different evaluation metrics (confusion matrix, overall accuracy (OA), kappa coefficient, and F1 score), burned territories from the Institute for Nature Conservation and Forests (ICNF), Normalized Difference Vegetation Index (NDVI) map, computed from the Sentinel-2A imagery, and COSc 2023 from SNIG; v) to detect the vegetation dynamic and burned area by computing NDVI and Normalized Burn Ratio (NBR) using Sentinel-2A; and vi) to detect the land cover change be-tween pre-wildfire 2022, post-wildfire 2022, and summer 2023, and vii) observe the burned area recovery with the highest accuracy land cover maps.
If an adequate model can be determined for the land cover classification in the study area, it will enable authorities to develop prompt disaster management strategies to prevent fatalities caused by landslides and flash floods following a wildfire and regular scrutiny on burned area recovery and vegetation dynamics. This provides valuable information for risk management, land management, and biodiversity recovery monitoring.
4. Conclusions
This research aimed to identify the best model for land cover classification of the study area using GEE. A land cover map provides important information for emergency planning and disaster management during a wildfire, as well as burned areas and vegetation monitoring after a wildfire. Sentinel-2A imagery was processed with two LULC classification approaches, OBIA and PBIA. The feature collections were subsequently trained with RF and SVM classifiers on GEE. The land cover maps were compared with the NDVI, BA22, dNBR, and COSc 2023 maps. The evaluation metric results were very consistent; the RF x OBIA model was the most accurate for land cover classification in the study area. Compared with the NDVI map and COSc 2023, the SVM x PBIA map resembled the maps better. However, the validation of land cover maps against COSc 2023 was not ideal as the land cover map generated in this research was based on a specific date, while the land cover classification in COSc 2023 was based on the monthly mean of a year. In addition, there is no access to field data, which is available for validation. Considering these limitations, the best model for land cover classification of the study area was not conclusively determined. Nonetheless, this research provided some important insights: i) The land cover classification with the highest values of evaluation metrics might not reflect the same level of accuracy in map presentation; ii) Concurrent with Lawrence and Moran’s studies, more classifiers should be tested to identify the most accurate model; iii) The PBIA approach had difficulties in distinguishing the vegetation; the OBIA approach tackled this problem better with the RF classifier but not with the SVM classifier; and iv) The SVM classifier had unstable performance where the accuracy in classifying the land cover classes fluctuates in different events.
In future work, it is ideal to test the model with imagery of other data with available field data to identify an optimized model to classify the land cover for periodic assessment in the study area. Since the SVM x PBIA model strongly resembled COSc 2023 while achieving low accuracy in evaluation metrics, different parameters can be tested and refined to optimize the result in future works. Furthermore, different ML and deep learning algorithms can be tested to improve the land cover classification result.
Figure 1.
Serra da Estrela Natural Park. (Source: Fire Station from Google Earth, Road/River/Reservoirs and Lagoons from OSM, DEM from FCUP, PNSE boundaries from ICNF, Estrela Geopark/Municipalities from DGT, Spain from GADM).
Figure 1.
Serra da Estrela Natural Park. (Source: Fire Station from Google Earth, Road/River/Reservoirs and Lagoons from OSM, DEM from FCUP, PNSE boundaries from ICNF, Estrela Geopark/Municipalities from DGT, Spain from GADM).
Figure 2.
Methodology diagram.
Figure 2.
Methodology diagram.
Figure 3.
The dNBR maps of the study area for the 2022 wildfire, based on classifications proposed by MLC (left), and UGSG (right).
Figure 3.
The dNBR maps of the study area for the 2022 wildfire, based on classifications proposed by MLC (left), and UGSG (right).
Figure 4.
NDVI maps of the study area for: (a) pre-wildfire 2022, (b) post-wildfire 2022, and (c) summer 2023, computation was made from Sentinel-2A data and classified according to NDVI values range suggested by USGS.
Figure 4.
NDVI maps of the study area for: (a) pre-wildfire 2022, (b) post-wildfire 2022, and (c) summer 2023, computation was made from Sentinel-2A data and classified according to NDVI values range suggested by USGS.
Figure 5.
F1 score ranking for different land cover classes across various models.
Figure 5.
F1 score ranking for different land cover classes across various models.
Figure 6.
Comparison of the burned area retrieved from ICNF (red border), and the post-wildfire land cover maps classified by different models: (a) RF x OBIA, (b) RF x PBIA, (c) SVM x OBIA, (d) SVM x PBIA.
Figure 6.
Comparison of the burned area retrieved from ICNF (red border), and the post-wildfire land cover maps classified by different models: (a) RF x OBIA, (b) RF x PBIA, (c) SVM x OBIA, (d) SVM x PBIA.
Figure 7.
Comparison of land cover maps classified by different models: (a) RF x OBIA, (b) RF x PBIA, (c) SVM x OBIA, (d) SVM x PBIA for pre-wildfire 2022.
Figure 7.
Comparison of land cover maps classified by different models: (a) RF x OBIA, (b) RF x PBIA, (c) SVM x OBIA, (d) SVM x PBIA for pre-wildfire 2022.
Figure 9.
Comparison of land cover maps classified by different models: (a) RF x OBIA, (b) RF x PBIA, (c) SVM x OBIA, (d) SVM x PBIA for summer 2023.
Figure 9.
Comparison of land cover maps classified by different models: (a) RF x OBIA, (b) RF x PBIA, (c) SVM x OBIA, (d) SVM x PBIA for summer 2023.
Figure 10.
The land cover maps classified using RF x OBIA model: (a) pre-wildfire 2022, (b) post-wildfire, and (c) summer 2023.
Figure 10.
The land cover maps classified using RF x OBIA model: (a) pre-wildfire 2022, (b) post-wildfire, and (c) summer 2023.
Figure 11.
The land cover maps classified using SVM x PBIA model: (a) pre-wildfire 2022, (b) post-wildfire, and (c) summer 2023.
Figure 11.
The land cover maps classified using SVM x PBIA model: (a) pre-wildfire 2022, (b) post-wildfire, and (c) summer 2023.
Table 1.
Sentinel-2A imageries applied in this study.
Table 1.
Sentinel-2A imageries applied in this study.
Acquisition Date |
Event |
Cloud cover (%) |
2 August 2022 |
Pre-wildfire 2022 |
0.000657 |
26 September 2022 |
Post-wildfire 2022 |
0.152309 |
28 July 2023 |
Summer 2023 (1 Year after fire) |
0.000352 |
Table 2.
Description of the 6 land cover classes used in this research.
Table 3.
Analysis of the land area affected by different fire severities in the study area after the 2022 wildfire, based on the dNBR ranges proposed by MLC and USGS.
Table 3.
Analysis of the land area affected by different fire severities in the study area after the 2022 wildfire, based on the dNBR ranges proposed by MLC and USGS.
Fire Severity |
dNBR range (scaled by 103) |
MLC (100m2) |
% |
dNBR range (scaled by 103) |
USGS (100m2) |
% |
Unburned |
<= 100 |
53291 |
2.20 |
<= 100 |
53291 |
2.20 |
Low |
100 - 320 |
409450 |
16.89 |
100 - 270 |
292044 |
12.05 |
Moderate |
320 - 650 |
1042929 |
43.03 |
270 - 660 |
1193272 |
49.23 |
High |
> 650 |
918192 |
37.88 |
> 660 |
885255 |
36.52 |
Total |
|
2423862 |
100 |
|
2423862 |
100 |
Table 4.
Analysis of the study area size subject to NDVI values range suggested by USGS.
Table 4.
Analysis of the study area size subject to NDVI values range suggested by USGS.
NDVI |
Pre-wildfire 2022 (100m2) |
% |
Post-wildfire 2022 (100m2) |
% |
Summer 2023 (100m2) |
% |
<= 0.1 |
647 |
0.03 |
23563 |
0.98 |
25378 |
1.05 |
0.1 - 0.5 |
844708 |
34.85 |
2229573 |
91.98 |
1820739 |
75.11 |
> 0.5 |
1578507 |
65.12 |
170726 |
7.04 |
577745 |
23.84 |
Total |
2423862 |
100 |
2423862 |
100 |
2423862 |
100 |
Table 5.
Confusion Matrix of SVM x PBIA model [Post-wildfire 2022].
Table 5.
Confusion Matrix of SVM x PBIA model [Post-wildfire 2022].
SVM x PBIA |
Agriculture |
Artificial |
Bareland |
Forest |
Shrub |
Water |
Precision |
Agriculture |
797 |
2 |
16 |
249 |
79 |
0 |
0.6973 |
Artificial |
11 |
207 |
6 |
0 |
0 |
0 |
0.9241 |
Bareland |
18 |
13 |
2554 |
0 |
0 |
0 |
0.988 |
Forest |
185 |
0 |
0 |
860 |
11 |
0 |
0.8144 |
Shrub |
223 |
2 |
0 |
11 |
83 |
0 |
0.2602 |
Water |
0 |
0 |
0 |
0 |
0 |
53 |
1 |
Recall |
0.6459 |
0.9241 |
0.9915 |
0.7679 |
0.4798 |
1 |
|
Table 6.
Overall accuracy and kappa coefficient of each model.
Table 6.
Overall accuracy and kappa coefficient of each model.
|
RF x OBIA |
RF x PBIA |
SVM x OBIA |
SVM x PBIA |
OA |
κ |
OA |
κ |
OA |
κ |
OA |
κ |
Pre-fire 2022 |
0.99 |
0.98 |
0.90 |
0.86 |
0.74 |
0.66 |
0.81 |
0.73 |
Post-fire 2022 |
0.99 |
0.98 |
0.94 |
0.91 |
0.84 |
0.76 |
0.85 |
0.77 |
Summer 2023 |
0.99 |
0.99 |
0.94 |
0.91 |
0.97 |
0.96 |
0.88 |
0.83 |
Table 7.
Analysis of the land cover size after wildfire 2022 for each model.
Table 7.
Analysis of the land cover size after wildfire 2022 for each model.
F22 |
RF x OBIA |
RF x PBIA |
SVM x OBIA |
SVM x PBIA |
Class |
Area (ha) |
% |
Area (ha) |
% |
Area (ha) |
% |
Area (ha) |
% |
0 |
1,655.68 |
6.83 |
2,989.13 |
12.33 |
3,684.57 |
15.20 |
5,172.30 |
21.34 |
1 |
191.49 |
0.79 |
373.98 |
1.54 |
639.68 |
2.64 |
593.50 |
2.45 |
2 |
20,512.27 |
84.63 |
18,566.33 |
76.60 |
18,710.02 |
77.19 |
17,346.21 |
71.56 |
3 |
1,152.91 |
4.76 |
1,527.02 |
6.30 |
754.61 |
3.11 |
391,02 |
1.61 |
4 |
724.23 |
2.99 |
779.75 |
3.22 |
447.71 |
1.85 |
733.09 |
3.02 |
Table 8.
Analysis of the land cover size in summer 2023 for each model and COSc 2023.
Table 8.
Analysis of the land cover size in summer 2023 for each model and COSc 2023.
Area(ha) |
Agriculture |
Artificial |
Bareland |
Forest |
Shrub |
COSc 2023 |
1479.32 |
5.78 |
5820.55 |
1168.25 |
15757.47 |
RF_OBIA_S23 |
6867.58 |
349.50 |
10425.99 |
2316.96 |
4276.57 |
RF_PBIA_S23 |
10254.29 |
187.83 |
8555.31 |
1892.89 |
3342.48 |
SVM_OBIA_S23 |
9822.13 |
1041.57 |
9404.46 |
2196.01 |
1771.86 |
SVM_PBIA_S23 |
14189.20 |
472.56 |
7944.88 |
752.46 |
876.15 |
Table 9.
Each land cover class's major land cover change from pre-fire 2022 to post-fire 2022, and from post-fire 2022 to summer 2023 [RF x OBIA model].
Table 9.
Each land cover class's major land cover change from pre-fire 2022 to post-fire 2022, and from post-fire 2022 to summer 2023 [RF x OBIA model].
Major change from |
Prefire 2022 to Postfire 2022 |
Postfire 2022 to Summer 2023 |
To |
Area (ha) |
% |
To |
Area (ha) |
% |
Agriculture |
Bareland |
1593.07 |
6.57 |
Agriculture |
690.74 |
2.85 |
Artificial |
Bareland |
70.53 |
0.29 |
Artificial |
58.79 |
0.24 |
Bareland |
Bareland |
2094.21 |
8.64 |
Bareland |
9965.35 |
41.11 |
Forest |
Bareland |
15847.11 |
65.38 |
Forest |
653.9 |
2.7 |
Shrub |
Bareland |
907.34 |
3.74 |
Agriculture |
302.43 |
1.25 |
Table 10.
The major land cover change of each land cover class from pre-fire 2022 to post-fire 2022, and from post-fire 2022 to summer 2023 [SVM x PBIA model].
Table 10.
The major land cover change of each land cover class from pre-fire 2022 to post-fire 2022, and from post-fire 2022 to summer 2023 [SVM x PBIA model].
Major change from |
Prefire 2022 to Postfire 2022 |
Postfire 2022 to Summer 2023 |
To |
Area (ha) |
% |
To |
Area (ha) |
% |
Agriculture |
Bareland |
3305.88 |
13.64 |
Agriculture |
3881.37 |
16.01 |
Artificial |
Bareland |
28.07 |
0.12 |
Agriculture |
370.35 |
1.53 |
Bareland |
Bareland |
877.69 |
3.62 |
Agriculture |
9354.4 |
38.59 |
Forest |
Bareland |
11373.65 |
46.92 |
Forest |
243.37 |
1 |
Shrub |
Bareland |
1760.7 |
7.26 |
Agriculture |
450.55 |
1.9 |
Table 11.
Land Cover Changes: Prefire 2022, Postfire 2022, and Summer 2023 [RF x OBIA].
Table 11.
Land Cover Changes: Prefire 2022, Postfire 2022, and Summer 2023 [RF x OBIA].
ID |
Land Cover |
Prefire 2022 |
Postfire 2022 |
Summer 2023 |
Area (ha) |
% |
Area (ha) |
% |
Area (ha) |
% |
0 |
Agriculture |
2452.94 |
10.12 |
1655.68 |
6.83 |
6867.58 |
28.34 |
1 |
Artificial |
173.57 |
0.72 |
191.49 |
0.79 |
349.5 |
1.44 |
2 |
Bareland |
2256.58 |
9.31 |
20512.27 |
84.63 |
10425.99 |
43.02 |
3 |
Forest |
17846.92 |
73.64 |
1152.91 |
4.76 |
2316.96 |
9.56 |
4 |
Shrub |
1506.59 |
6.22 |
724.23 |
2.99 |
4276.57 |
17.65 |
Total |
24236.6 |
100 |
24236.6 |
100 |
24236.6 |
100 |
Table 12.
Land Cover Changes: Prefire 2022, Postfire 2022, and Summer 2023 [SVM x PBIA].
Table 12.
Land Cover Changes: Prefire 2022, Postfire 2022, and Summer 2023 [SVM x PBIA].
ID |
Land Cover |
Prefire 2022 |
Postfire 2022 |
Summer 2023 |
Area (ha) |
% |
Area (ha) |
% |
Area (ha) |
% |
0 |
Agriculture |
5292.29 |
21.83 |
5,172.30 |
21.34 |
14189.20 |
58.55 |
1 |
Artificial |
63.35 |
0.26 |
593.50 |
2.45 |
472.56 |
1.95 |
2 |
Bareland |
984.99 |
4.06 |
17,346.21 |
71.56 |
7944.88 |
32.78 |
3 |
Forest |
14884.9 |
61.41 |
391,02 |
1.61 |
752.46 |
3.1 |
4 |
Shrub |
3010.71 |
12.42 |
733.09 |
3.02 |
876.15 |
3.62 |
Total |
24236.24 |
100 |
24236.12 |
100 |
24235.25 |
100 |