Remote Sensing Technologies for Monitoring Argane Forest Stands: A Comprehensive Review

Abstract

This comprehensive review explores the ecological significance of the Argane stands (Argania spinosa) in southwestern Morocco and the pivotal role of remote sensing technology in monitoring forest ecosystems. Argane stands, known for their resilience in semi-arid and arid conditions, serve as a keystone species, preventing soil erosion, maintaining ecological balance, and providing habitat and sustenance to diverse wildlife species. Additionally, they produce an extremely valuable Argane oil, offering economic opportunities and cultural significance to local communities. Remote sensing tools, including satellite imagery, LiDAR, drones, radar, and GPS precision, have revolutionized our capacity to remotely gather data on forest health, cover, and responses to environmental changes. These technologies provide precise insights into canopy structure, density, and individual tree health, enabling assessments of Argane stand populations and detection of abiotic stresses, biodiversity, and conservation evaluations. Furthermore, remote sensing plays a crucial role in monitoring vegetation health, productivity, and drought stress, contributing to sustainable land management practices. This review underscores the transformative impact of remote sensing in safeguarding forest ecosystems, particularly the Argane forest stands, and highlights its potential for continued advancements in ecological research and conservation efforts.

1. Introduction

The Argane tree, scientifically known as Argania spinosa (L.) Skeels, is a significant tree species endemic to southwestern Morocco, belonging to the Sapotaceae family. The tree is characterized by its fruit-forest stand form, variable presence of thorns, and an upright structure resembling the olive tree. It can reach heights of 8 to 10 m or more, depending on ecological conditions, with a dense, sprawling canopy. The tree’s branches are robust and often twisted, with a rugged texture on the bark and large branches bear thorns. The leaves are small, leathery, and come in various shapes, typically arranged in fascicles [1]. During periods of drought, the Argane stands showcase its resilience by retaining some of its leaves while shedding others in response to the degree of environmental stress [2]. However, its significance transcends its impressive adaptability—this species can promote regional and national sustainable development [1]. Argane forest stands serve a multitude of ecological functions, including the establishment of a conducive microclimate for diverse fauna and flora, mitigation of soil erosion, and prevention of desertification. Additionally, it serves as a crucial resource for local communities, which have deep-rooted socioeconomic ties to the diverse array of products derived from it (e.g., oil, soap, shampoo, cosmetic creams, and livestock feed) [1]. Furthermore, this review offers a comprehensive exploration of remote sensing technology, highlighting its indispensable role in gathering critical data and insights concerning the health and conditions of the Earth’s surface, particularly in the context of forests and trees. It underscores the versatility of remote sensing instruments and methodologies available to researchers, such as satellite imagery and aircraft technology, which as one of the featured technologies, enables researchers to capture large-scale data on forest cover, health, and changes over time [3]. This is especially beneficial for assessing the Argane stands population, identifying areas under stress due to factors like climate change or land use alterations, as well as evaluating the effectiveness of conservation efforts. LiDAR (light detection and ranging), another highlighted tool, provides high-resolution, three-dimensional data about forest structure. For the Argane stands and forest trees in general, LiDAR technology offers precise information about canopy height, density, and even individual tree health [4]. Unmanned aerial vehicles (UAVs) have emerged as a pivotal technology for evaluating forest parameters, thanks to their exceptional spatial resolution, versatility, and ease of operation [5,6]. Specifically, low-cost UAVs equipped with multispectral cameras have found extensive use in forest applications due to their remarkable precision in detecting tree-level parameters, including individual tree identification and measurement of tree height [6], health status assessment [7], and diameter at breast height [8]. This capability proves especially valuable in remote or challenging-to-access areas where conducting ground surveys may present logistical difficulties. Additionally, radar technology can penetrate forest canopies to some extent, allowing researchers to monitor soil moisture levels, which is crucial for understanding the Argane tree’s resilience during droughts. GPS is indispensable for tracking and mapping the exact locations of Argane stand populations. These data can be used to create accurate distribution maps, monitor changes in habitat, and assess the impact of human activities on these ecosystems. These cutting-edge tools provide scientists with the capability to remotely observe and monitor a wide range of environmental phenomena. Furthermore, this review discusses the importance of monitoring vegetation health, productivity, and drought stress using remote sensing technology. It introduces various vegetation indices, such as NDVI, EVI, SAVI, LAI, and others, that aid in assessing the well-being of vegetation [9], detecting stress factors [10], and contributing to sustainable land management practices. Furthermore, this review also delves into the utilization of remote sensing for crop yield estimation and drought condition monitoring, with specific emphasis on the manifold drought indices and their diverse applications. Overall, this comprehensive review provides a comprehensive overview of the Argane forest stands, remote sensing technology, and their multifarious applications in mapping and monitoring vegetation health and assessing productivity and susceptibility to drought-induced stress. It lays the groundwork for an in-depth investigation into the utilization of remote sensing for the examination of Argania spinosa in its ecosystem.

2. Argania spinosa Characteristics

Argania spinosa (Argane stand) is a species of profound ecological and socio-economic significance, particularly in southwestern Morocco. Comprehending its distinctive characteristics aids in elucidating its ecological role and associated advantages, thereby motivating our commitment to its surveillance via state-of-the-art accessible remote sensing technologies.

2.1. Taxonomical, Botanical and Phenological Features

The Argane tree, scientifically known as Argania spinosa (L.) Skeels, is a significant tree endemic to southwestern Morocco and a member of the Sapotaceae family, found within the order Ericales, and the only species of the genus Argania [11]. It is characterized by its fruit-forest tree form, variable presence of thorns, and an upright structure that resembles the olive tree. The tree’s size can reach 8 to 10 m or more, depending on ecological conditions, with a sprawling, dense canopy. Its branches are vigorous, often twisted and gnarled, arising from the fusion of close shoots. The bark and large branches exhibit a rugged texture, close to “snakeskin” cracks. Branches are densely distributed, with thorny tips. The leaves are small, alternate, leathery, and come in various shapes (Figure 1). They are typically arranged in fascicles with short petioles (2 to 3 cm), featuring dark green upper surfaces and paler undersides, with distinct midribs and fine, branched lateral veins [1]. During drought periods, Argane stand leaves partially persist, and the tree may shed some or all of its foliage [2]. Argane stand flowers are monoecious, hermaphroditic, and pentamerous, typically ranging from white to greenish-yellow. These flowers cluster at internodes and leaf axils, with a single cluster often containing 15 or more flowers [12,13]. The flower structure comprises five pubescent sepals, followed by two bracts and five rounded petals, though in some varieties, this number can extend to ten [14]. The androecium consists of five stamens with short threads and a prominent mucronate or obtuse anther. The ovoid ovary is pubescent and erect, topped by a short or conical style, sometimes surpassing the stamen length and the fruits generally ripen between June and August, depending on the location and environmental conditions [1]. Flowering occurs from February to June, with fruits maturing around August. Pollination is predominantly anemophilous (80%), with the remaining 20% being entomophilous [15], facilitated by flies like those from the Calliphoridae family. The fruit of the Argane stands is a false drupe [16] with a fleshy pericarp, enclosing one to three (or even four or five) cavities containing false oil-rich almonds (nuts) [1]. These fruits come in various shapes and sizes, changing in color as they mature [1,17,18]. Agane stands exhibit a dimorphic root system that appears to be associated with a flexible water absorption model [1,19]. This system comprises taproots that continue to extend into deeper layers and horizontally developing roots, facilitating nutrient uptake in proximity to the surface and water absorption across shallow to deep soil strata (up to 8 m depth), particularly in response to surface soil drying [1,19,20], with important mycorrhizal symbiosis [21]. Argane wood is dense and valuable for charcoal production. The longevity of the Argane stands is uncertain due to its irregular wood growth patterns, which relate more to vegetation periods than distinct years.

2.2. Geographical Distribution of the Argane Stands

The Argane stands are mainly found in the arid and semi-arid regions of southwestern Morocco [2]. Figure 2 illustrates the distribution area of the Argane forest stands in Morocco. It covers an extensive area of approximately 828,300 hectares [22] to 999,079 hectares [23] and is estimated to comprise around 21 million trees. Its geographical range spans from 29 to 32 degrees north latitude, extending along the Moroccan ocean coast from the northern mouth of the Oued Tensift to the southern mouth of the Oued Drâa. The central region where Argane stands are predominant stretches from the northeast of Essaouira to the Souss valley [1]. Today, the Argane stands have significantly expanded its habitat, covering a larger area. It can be found in various regions, including the Souss plain (Taliouine, Aoulouz, Taroudannt, and Agadir) and the southern and western slopes of the Western High Atlas (Haha and Ida-ou-Tanane), extending to Essaouira. The tree also thrives on the northeastern of the Western Anti-Atlas, reaching as far as Sidi-Ifni [24]. Two notable Argane stands areas are the upper Grou valley in the southeast of Rabat and the northwestern foothills of the Béni-Snassen region near Oujda.

2.3. Ecological Feature

The Argane stands are a thermophilic and xerophilic species that thrive in specific conditions of temperature and humidity, typically favoring a mild climate. Most Argane groves are located in the Mediterranean–Saharan transition zone [25], and the tree naturally grows in warm and temperate arid to semi-arid zones [2], including Saharan regions in southern Morocco. It exhibits remarkable resilience to the arid conditions of southwest Morocco, capable of enduring temperatures ranging from 30 °C to 50 °C, with the average temperature during the coldest month in its habitat ranging from 3 °C to 7 °C [2]. The Argane stands have demonstrated its ability to withstand short-duration extreme minimum temperatures, reaching as low as −2.6 °C in 1955. Moreover, it can thrive with minimal rainfall and can be found on steep slopes across a wide altitudinal range, spanning from sea level to approximately 1500 m [22]. The yearly rainfall within its primary range varies from 150 to 400 mm [1]. The Argane stand’s deep-rooted nature allows it to access water from considerable depths in the soil. The Argane stands exhibit adaptability to various soil types, including on shallow, rocky, schist, quartzite, limestone, and alluvial soils [26,27]. However, it tends to avoid mobile sand, soils, which can lead to root scouring and detrimentally impact its root system [26].

3. Benefits and Uses of Argane

3.1. Ecological Interest of the Argane Stands

The Argane stands play an irreplaceable role in maintaining ecological balance. Its powerful root system acts as a shield against both wind and water erosion, effectively capturing water, stabilizing the soil, and reducing erosion. This essential function helps mitigate the advance of desertification, as noted by [2]. Additionally, the tree counteracts rain-induced erosion, especially in mountainous regions, making it an excellent soil stabilizer for mountain soils. The deep and robust root system of the Argane stands facilitate rainwater infiltration, aiding groundwater. Furthermore, its shading and soil-enhancing properties foster the development of a diverse range of biological entities, encompassing fauna, flora, and microflora [1,21]. This rich biodiversity significantly enhances natural areas and contributes to vital microbial activities such as nitrogen mineralization and improved phosphorus availability, as observed in studies by [26]. The presence of the Argane stands are directly linked to the thriving of these living organisms, including plants, animals, and microflora, with researchers identifying approximately 100 plant species growing beneath their canopy [26].

3.2. Socio-Economic Interest of Argane Stands

The Argane stands are a highly valuable resource with diverse applications, and each part of the tree, including its wood, leaves, fruit, and oil, serves as a source of sustenance and income for those who depend on it. This tree is essential to the socio-economic, botanical, and ecological dynamics within rural communities [1,28].

Wood: The Argane stands provide sturdy and resilient wood, widely used as a fuel source in the form of charcoal for heating, as well as in rural construction, joinery, and the crafting of traditional and household items [1].

Livestock Use: All components of the tree are essential for livestock. Argane leaves offer valuable grazing for goats and camels, while the oil cake, rich in carbohydrates and proteins, is used to fatten livestock.

Edible Argane Oil: Argane oil, extracted from the kernels, is a nutritious edible oil with a pleasing taste and exceptional dietary value. It contains approximately 80% unsaturated fatty acids and 20% saturated fatty acids, including significant amounts of linoleic and oleic acids and rich in polyphenols, tocopherols, and carotenes [29,30]. This oil has become one of the most valuable edible oils globally [31].

Cosmetic Application: Argane oil is also used in cosmetic products for its therapeutic potential [16,32,33], serving as a moisturizing oil for the face, hands, and feet. In addition to these functions and uses, the Argane stands play a crucial role in sustaining the livelihoods of rural communities, helping to alleviate rural migration pressures. Thus, the Argane stands are a versatile and indispensable resource, supporting both the economic well-being of local residents and the preservation of traditional practices, while also playing a vital role in addressing rural depopulation.

4. Remote Sensing Applications

Remote sensing technology involves the use of various tools and techniques for gathering data and information about the Earth’s surface and atmosphere remotely. These technologies allow for the observation, measurement, and monitoring of environmental and geographical phenomena without physical contact.

4.1. Land Use and Land Cover Assessment and Density Assessment

Remote sensing, particularly satellite imagery, plays a crucial role in both land use and land cover assessment (LULCA) and change detection within Argane stand habitats. It allows researchers to categorize and map different land cover types, such as forest-cover changes, global environmental changes [34], agricultural areas, urban development, and natural vegetation. Cameras mounted on satellite imagery, aircraft and radar provide a unique aerial perspective that serves mapping, land-use planning [35,36,37,38], and environmental monitoring purposes exceptionally well [3,39,40,41]; disaster response [42,43,44]; and sea ice monitoring [45,46,47,48]. These technologies provide invaluable insights into land cover, vegetation, and dynamic changes over time, making them indispensable tools for researchers. By delineating the extent and distribution of land cover categories, scientists can assess the impact of land use changes on Argane stand populations. LULC data help identify areas at risk due to factors like deforestation, urban expansion, or agricultural encroachment, providing essential insights for conservation planning. For example, Idbraim et al. [49], focused on using a CNN-based model for change detection to map deforestation of Argane forest stands, an UNESCO Biosphere Reserve located in the Souss-Massa region of Morocco from 2015 to 2020. Marzolff et al. [50] analyzed the tree-cover changes between 1967 and 2019 in Argane woodlands exhibiting different levels degrees of degradation across the provinces Chtouka-Aït Baha, Taroudant, and Agadir Ida-Outanane. Moumni et al. [51] conducted a study on the current state of forest species, specifically the Argane stands (Argania spinosa), in Essaouira Province to analyze forest changes (degradation) and create innovative approaches for preserving this native tree species. Ezaidi et al. [52] assessed and analyzed the degradation of vegetation cover over the past three decades, spanning from 1984 to 2018, in regions of the Arganeraie Biosphere Reserve in Morocco. Tree density assessment involves specifically measuring the number of trees per unit area within a defined forested or vegetated area. It is a metric used to understand the density of tree populations within a particular region, forest, or ecosystem, and it is particularly valuable for vegetation analysis. The assessment of Argane stand density has been conducted by various authors, including Smiej et al. [53] and Aouragh et al. [54]. In the eastern region of Morocco, the area occupied by the Argane forest stands covers 7.2 km², which is considerably lower compared to the southern region of Morocco, where it spans an area of 8300 km² [55]. Additionally, Marzolff et al. [50] reported that Argane stand densities can reach up to 300 trees and shrubs per hectare, with the average tree density increased from 58 trees per hectare in 1967 to 86 trees per hectare in 2019. This increase is primarily attributed to the proliferation of small trees with a diameter of less than 3 meters, often resulting from stump re-sprouting after the felling of larger trees in the provinces of Chtouka-Aït Baha, Taroudant, and Agadir Ida-Outanane.

To enhance the monitoring and conservation of Argane stands and other species, researchers should adopt advanced machine learning models for their high accuracy in detecting deforestation events from satellite imagery data. Integrating these models with multi-source data, including satellite imagery, on-ground observations, and historical records, can significantly improve the accuracy and comprehensiveness of land use and land cover assessment (LULCA) and change detection. This integration will allow for more precise identification of areas at risk due to deforestation, urban expansion, and agricultural encroachment. Policymakers should leverage these findings to develop and enforce targeted conservation strategies, focusing on high-risk regions identified through remote sensing studies. Additionally, the implementation of real-time monitoring systems, facilitated by remote sensing technologies, can provide stakeholders with actionable insights for adaptive land management practices. By encouraging collaboration between researchers, conservationists, and policymakers, we can ensure the sustainable management of Argane stands and other species, enhancing their resilience in the face of environmental challenges.

4.2. Dendrometric Parameters Using Light Detection and Ranging and Drones

The advent of LiDAR (light detection and ranging) technology has enabled the creation of precise 3D maps of landscapes. Leveraging laser-based measurements, LiDAR offers profound insights into terrain structure, enhancing our ability to assess flood risks [56,57,58,59], study forest canopy height [60,61,62,63,64], and plan urban landscapes [4,65,66,67]. It has evolved into a critical tool for managing natural resources and addressing the impacts of climate change. Drones, also known as unmanned aerial vehicles (UAVs), are compact, unmanned aircraft equipped with advanced cameras and sensors. These versatile devices play a pivotal role in collecting localized, high-resolution data. Whether itis monitoring plant health [9,68,69,70]; assessing the aftermath of disasters [43,44,71,72,73,74]; conducting density assessments [75]; measuring 3D canopy height [60,61,62,64,76,77,78], crown diameter [79,80,81], diameter at breast height (DBH) [8,82,83,84,85], and leaf area index (LAI) [23,86,87,88,89,90,91,92,93]; estimating evapotranspiration (ET) and water stress [9,94,95,96,97,98,99,100]; or inspecting critical infrastructure, drones have emerged as indispensable tools for a wide range of applications, owing to their agility and ability to access hard-to-reach areas. UAVs offer a practical alternative for collecting high-resolution forest inventory data due to their cost-effective features [6,77,78,79,101]. Marzolff et al. [102] surveyed 30 test sites, each covering 1 hectare, in Argane woodlands of different degradation stages using a UAV equipped with an RGB optical camera. They collected information on tree height and crown size from canopy height models (CHMs) and then analyzed various 3D point-cloud characteristics, including point profile shape, density, and layer structure. Various methods are used for individual tree identification and tree height extraction from CHM, including deep learning [103,104], local maxima [6,78,105], and extended-maxima transformation [106]. Among these, the local maxima algorithm is the most commonly used method for identifying treetops [78,107]. This analysis was conducted within a geographical information system (GIS).

Improving research and conservation efforts in ecosystems like Argane stands and other species involves exploring the integration of LiDAR technology and UAVs equipped with advanced sensors. LiDAR’s capability to create precise 3D maps enables detailed assessments of terrain structure and supports the management of natural resources, including monitoring forest canopy height, crown diameter, and diameter at breast height (DBH) and planning urban landscapes. Additionally, UAVs equipped with sensors facilitate the collection of high-resolution data essential for tasks such as monitoring plant health, conducting density assessments, and estimating ecological parameters like leaf area index and evapotranspiration. Machine learning algorithms, such as random forests (RF), artificial neural networks (ANN), and support vector regression (SVR), can be employed to predict dendrometric measurements from LiDAR and UAV data, enhancing the accuracy and efficiency of ecological assessments. Integrating these technologies can provide comprehensive insights into ecosystem dynamics and support effective conservation strategies tailored to specific environmental challenges and management needs.

4.3. Health Monitoring

Vegetation health is a critical component of ecosystem vitality and productivity. Assessing and monitoring vegetation health provide valuable insights into the condition of ecosystems, enabling timely interventions and informed decision making. Remote sensing technology has significantly advanced our ability to assess vegetation health by utilizing vegetation indices. These indices play a pivotal role in understanding the well-being of vegetation, detecting stress factors, and contributing to sustainable land management practices. Vegetation indices are fundamental tools in remote sensing for monitoring and assessing the health and vitality of plant ecosystems.

The normalized difference vegetation index (NDVI), developed by [108], is a widely employed index that quantifies vegetation health by contrasting the reflectance of near-infrared (NIR) and red, with higher values indicating healthier vegetation (Figure 3). NDVI is directly used to monitor and characterize canopy growth and plant vigor [109]. Healthy plants with robust photosynthetic activity can be analyzed by contrasting the reflectance of near-infrared (NIR) and visible light [9,109]. Vegetation indices also offer insights into plant growth, as healthy plants reflect more near-infrared and absorb more visible light [110]. Variations in NDVI values could be due to water availability issues, either deficiency or excess. Therefore, to obtain a more comprehensive understanding, other indices such as the soil moisture index and the evapotranspiration index should be used alongside NDVI. Ezaidi et al. [52] conducted a study on ecosystem degradation, utilizing multi-temporal Landsat-derived NDVI (Normalized Difference Vegetation Index) to assess the scenario and identify the significant factors contributing to degradation. During the winter season, both olive and Argane stands exhibit NDVI values ranging from 0.3 to 0.55, while in the summer, these values decrease to a range of 0.2 to 0.3; this similarity in the NDVI signatures of the two vegetation classes is clearly evident, as reported by [111]. The reduction in NDVI values for these trees during the summer can be attributed to water stress caused by insufficient water availability and elevated temperatures during this period, as well as the loss of leaves due to pruning, which also contributes to the decrease in NDVI values, as noted by [51,111].

The Enhanced Vegetation Index (EVI) improves upon NDVI by correcting for atmospheric and canopy conditions [112,113], enhancing its suitability for vegetation health monitoring. Unlike NDVI, which can be affected by atmospheric noise and canopy background variations, EVI provides more accurate readings under these conditions, making it particularly useful in densely vegetated areas. In contrast, the Normalized Difference Water Index (NDWI) specializes in quantifying water content by analyzing near-infrared and shortwave infrared reflectance, making it valuable for detecting water stress [114,115]. NDWI is particularly effective in regions prone to periodic droughts, as it can detect changes in water content within vegetation and soil. To mitigate soil brightness interference in sparsely vegetated regions, the Soil-Adjusted Vegetation Index (SAVI) offers enhanced accuracy in assessing vegetation health [116]. SAVI is especially useful in arid and semi-arid regions where soil brightness can significantly affect NDVI readings. Additionally, the Modified Soil-Adjusted Vegetation Index (MSAVI) proposed by [117], further improves accuracy in vegetation cover assessment.

The Leaf Area Index (LAI) measures canopy leaf density [86], aiding in the evaluation of overall vegetation health and productivity. Mouafik et al. [23] applied random forest algorithms to estimate the Leaf Area Index (LAI) of Argania spinosa. Their approach involved vegetation indices extracted from both drone and Mohammed VI imagery, resulting in a higher R² compared to data obtained from Sentinel 2 imagery. This demonstrates the potential of advanced machine learning techniques in enhancing the accuracy of vegetation health assessments. Many studies have used machine learning methods for LAI prediction, such as those by [87,88,118,119,120,121,122], indicating a trend towards integrating remote sensing data with machine learning for improved vegetation monitoring. Chlorophyll indices, like the Chlorophyll Index (CI) [123,124] and Transformed Chlorophyll Absorption in Reflectance Index (TCARI) [125], focus on chlorophyll content, a key indicator of plant biomass [126] and plant health. The Chlorophyll Index green (CIg) [124] evaluates chlorophyll concentration and, consequently, vegetation health, using the green spectral region, valuable for detecting nutrient deficiencies related to chlorophyll production and photosynthetic activity. The Plant Senescence Reflectance Index (PSRI) captures chlorophyll and carotenoid levels to detect plant stress and senescence, while the Photochemical Reflectance Index (PRI) proxy of ecophysiological parameters monitor photosynthetic activity by assessing changes in xanthophyll cycle pigments [127], valuable for monitoring plant responses to environmental stressors like drought and disease. In drought-prone regions, the Water Stress Index (WSI) developed by [128], is instrumental in identifying water stress, enabling the assessment of drought impacts [129]. The Vegetation Health Index (VHI) proposed by [130], a composite index, integrates multiple parameters, including NDVI and surface temperature, to provide a holistic evaluation of vegetation health widely used for drought monitoring. These indices collectively constitute a comprehensive toolkit for researchers and environmental scientists engaged in the critical task of vegetation health assessment and management, providing comprehensive assessments that consider multiple aspects of vegetation health, as summarized in

Table 1. Summary of vegetation indices used for health and drought monitoring.

Index	Name	Formulation	Reference
NDVI	Normalized Difference Vegetation Index	${\displaystyle \frac{NIR-Red}{NIR+Red}}$	[108]
EVI	Enhanced Vegetation Index	$2\ast \left({\displaystyle \frac{NIR-Red}{NIR+6\ast Red-7.5\ast Blue+1}}\right)$	[113]
NDWI	Normalized Difference Water Index	${\displaystyle \frac{NIR-SWIR}{NIR+SWIR}}$	[115]
SAVI	Soil Adjusted Vegetation Index	$1.5\ast \left({\displaystyle \frac{NIR-Red}{NIR+Red+0.5}}\right)$	[116]
MSAVI	Modified Soil Adjusted Vegetation Index	${\displaystyle \frac{2\ast NIR+1-\sqrt{{(2\ast NIR+1)}^2-8\ast (NIR-Red)}}{2}}$	[117]
LAI NDVILog	Leaf Area Index from Log NDVI	${\displaystyle \frac{1}{2}}\ast {\mathit{log}}_e\left({\displaystyle \frac{NDVI+0.1}{0.2}}\right)$	[131,132,133]
CI green	Chlorophyll Index green	${\displaystyle \frac{NIR}{\mathrm{G}\mathrm{r}\mathrm{e}\mathrm{e}\mathrm{n}}}-1$	[123,124]
CI red-edge	Chlorophyll Index red-edge	${\displaystyle \frac{NIR}{\mathrm{R}\mathrm{E}}}-1$	[123,124]
TCARI	Transformed Chlorophyll Absorption in Reflectance Index	$3\ast \left[\left(RE1-Red\right)-0.2\ast \left(RE1-Green\right)\ast (RE1/Red)\right]$	[125]
PSRI	Plant Senescence Reflectance Index	${\displaystyle \frac{Red-Green}{RE2}}$	[134]
VHI	Vegetation Health Index	$VHI=a\times VCI+(1-a)\times TCI$	[130,135]
VCI	Vegetation Condition Index	${\displaystyle \frac{{\left(NDVI\right)}_{current}-{\left(NDVI\right)}_{min}}{{\left(NDVI\right)}_{max}-{\left(NDVI\right)}_{min}}}$	[136]
TCI	Temperature Condition Index	${\displaystyle \frac{{\left(LST\right)}_{max}-{\left(LST\right)}_{current}}{{\left(LST\right)}_{max}-{\left(LST\right)}_{min}}}$	[137]

.

Effective implementation of these indices requires a nuanced understanding of the specific context and conditions of the study area. For instance, while NDVI is broadly useful, its accuracy can be compromised in areas with high soil brightness or sparse vegetation, necessitating the use of SAVI or MSAVI. Integrating multiple indices can provide a more robust assessment. For example, combining NDVI with soil moisture and evapotranspiration indices offers a more detailed picture of water stress and its impact on vegetation health. The application of advanced machine learning techniques, as demonstrated by [23], can further enhance the predictive accuracy of vegetation health assessments. These methods allow for the integration of diverse data sources, such as drone and satellite imagery, to produce more reliable and actionable insights. Future research should focus on developing more sophisticated models that incorporate a wider range of environmental variables and leverage the latest advancements in remote sensing technology. Establishing standardized protocols for the application and interpretation of vegetation indices is also crucial to ensure consistency and comparability across different studies and regions. It is recommended that future studies explore the synergistic use of various vegetation indices and machine learning algorithms to improve the accuracy and applicability of vegetation health assessments in diverse ecological contexts.

4.4. Productivity Monitoring

Photosynthesis serves as the primary source for crop yield formation, yet only a fraction of sunlight is harnessed through leaf photosynthesis [138]. To describe a plant’s capacity to absorb atmospheric CO2 via photosynthesis, plants convert light into chemical energy and accumulate organic dry matter, introducing the concept of vegetation productivity. Vegetation productivity comprises two key components: gross primary productivity (GPP) and net primary productivity (NPP) [138]. GPP quantifies the total organic carbon fixed by green plants through photosynthesis over a specific time and area unit, while NPP represents the portion of GPP remaining after accounting for plant autotrophic respiration, which consumes organic matter. As NPP and GPP are closely linked to crop biomass, they are frequently employed in studies related to crop yield estimation [138]. Crop yield estimation via remote sensing methods may be categorized into three primary approaches: statistical analysis of remote sensing data, production efficiency models, and crop growth models [138]. The first approach involves the development of relationship models between remote sensing data, such as different band combinations or various remote sensing indices, along with crop yield components [139]. The second approach, based on production efficiency models, posits that crop yields under non-stressed conditions exhibit a linear correlation with the absorbed photosynthetically active radiation. This approach first estimates aboveground crop dry matter using remote sensing data and subsequently transforms it into crop yield [140,141]. The third method involves using satellite data to calibrate physiologically based crop simulations, emulating real crop growth dynamics to ultimately estimate crop yield [142,143].

Recently, machine learning models have demonstrated significant promise in retrieving biophysical and physiological parameters, particularly in crop yield prediction from remote sensing data. In the field of machine learning, the Linear Regression model establishes relationships among independent variables and multiple dependent variables [144] in estimating vegetation parameters [23]. Proficiency in a machine learning framework is achieved by utilizing datasets and reducing error or loss (such as root mean square error (RMSE), mean absolute error (MAE), or mean square error (MSE)) through regression algorithms [145]. Various parameters such as precipitation, temperature, soil conditions, and vegetation health are utilized in crop yield prediction through the application of machine learning [146]. Figure 4 and Figure 5 illustrate prediction algorithms used in crop yield forecasting across several studies and a spectrum of performance evaluation metrics utilized in prior crop yield prediction research, respectively, as cited by [145]. It is noted that random forest (RF), artificial neural networks (ANN), and support vector regression (SVR) are highlighted as among the most commonly used methods for crop yield prediction [145]. Among these machine learning models, the random forest (RF) algorithm emerges as a particularly robust tool for yield prediction, with widespread applications in agricultural research [146,147]. Random forest generates a multitude of regression trees derived from an extensive set of decision trees, enabling accurate regression computations and offering resilience against overfitting and noise [145,148,149]. Numerous studies have demonstrated the superior performance of random forest in crop yield prediction. For instance, Prasad et al. [146] used random forest to predict cotton yield, achieving promising results by utilizing various predictor variables such as land surface temperature (LST), Standardized Precipitation Index (SPI), Vegetation Condition Index (VCI), and historical yield data. Similarly, Jhajharia and Mathur [150] compared four machine learning algorithms RF, decision trees (DT), SVR, and Lasso regression (LassoR) and found that RF and DT outperformed the other methods in predicting crop yield. Sharifi [151] observed the same results in barley yield prediction, and Kuradusenge et al. [152] noted the superior performance of RF in predicting Irish potato and maize yields. Additionally, Fukuda et al. [153] reported successful use of RF for mango yield estimation, while Jeong et al. [154] achieved excellent results in wheat yield prediction. On the other hand, some studies have shown that other machine learning algorithms can also outperform in yield forecasting. For example, Abbas et al. [144] demonstrated that SVR models, using soil properties and NDVI as predictors, outperformed linear regression and K-nearest neighbor (KNN) algorithms in predicting potato tuber yield. Han et al. [155] showed that random forest, Gaussian process regression and support vector machine models are robust for predicting winter wheat yield using a combination of climate data, vegetation indices and soil information. In addition to RF and SVR, artificial neural networks (ANNs) represent a commonly employed machine learning technique for predicting crop yield, particularly suitable for modeling complex nonlinear relationships [145,156]. Mouafik et al. [157] demonstrated that the XGBoost model achieved the best performance in predicting the crop yield of Argania sinosa, followed by GBDT, RF, and ANN. Moreover, Mariadass et al. [158] utilized the XGBoost model for annual crop yield prediction in Malaysia, achieving an R-squared value of 0.98 and outperforming current models. Overall, the choice of algorithm for crop yield prediction is influenced by various factors, including training time, model structure, cost, and the handling of incomplete data. Each algorithm has its strengths and can be chosen based on the specific requirements of the prediction task.

Future research should focus on integrating multiple remote sensing indices and machine learning algorithms to enhance the robustness and applicability of crop yield predictions across diverse agricultural landscapes. Standardized protocols for data collection, preprocessing, and model calibration are essential to ensure consistency and comparability of results. Furthermore, leveraging the spatial and temporal resolution capabilities of remote sensing technologies can provide real-time insights into crop health, growth dynamics, and environmental conditions. Collaborative efforts among researchers, agronomists, and policymakers are crucial to implement these advanced techniques effectively in precision agriculture. By optimizing crop management practices based on accurate remote sensing data, stakeholders can mitigate risks associated with climate variability and contribute to global food security.

4.5. Drought Stress Monitoring

Drought, a severe climatic condition marked by prolonged periods of water deficiency due to an imbalance in water resources, is considered one of the most complex and poorly comprehended natural phenomena. In recent decades, the study of drought has intensified due to its escalating frequency and consequential losses [159]. It is typically categorized into four types: agricultural droughts, meteorological droughts, hydrological droughts, and economic and social droughts [138,160], leading to diminished vegetation, crop yields, and land degradation [161]. Meteorological drought arises from decreased precipitation, whereas agricultural drought involves inadequate water availability in soil for plant growth. Hydrological drought indicates a deficit in streamflow supply, while socioeconomic drought is characterized by an insufficient water supply to meet the demand for economic activities, affected by the drought conditions [160,162,163]. Numerous drought indices have been developed using diverse remote sensing data across various fields. Monitoring drought conditions involves the extensive use of these indices, which measure deviations in hydrologic processes [164]. Remote-sensing-based drought indicators are extensively employed in crop drought monitoring due to their effectiveness [138], especially in regions with limited access to meteorological stations [160,165,166]. These drought indices utilize distinct methods to quantify and analyze drought phenomena, each with varying requirements for remote sensing data, thereby providing valuable insights into this critical environmental challenge. Remote sensing also offers significant advantages for large-scale, real-time monitoring compared to traditional meteorological stations [138]. Prominent drought indicators include the Vegetation Condition Index (VCI), Temperature Condition Index (TCI), Precipitation Condition Index (PCI), Standardized Precipitation Index (SPI), Soil Moisture Condition Index (SMCI), and Evapotranspiration Condition Index (ETCI). These indices vary in their methodologies and data requirements, providing nuanced perspectives on drought severity and impacts across different landscapes. The VCI, as generalized by [136], serves as a robust indicator of moisture deficiency, distinct from the NDVI, by separating near-term climate effects from enduring biological signals [167]. It scales NDVI values from 0 to 1, using minimum and maximum thresholds, making it crucial indicator for evaluating drought impacts on agricultural areas [168]. Complementarily, the Temperature Condition Index (TCI), formulated by [137] under NOAA’s oversight in the United States, uses land surface temperature (LST) to assess vegetation stress from elevated temperature and humidity levels. Numerous studies have highlighted its positive correlation with VCI, making it valuable for monitoring drought events [160,161,168,169]. Building upon this correlation, Kogan [135] introduced the Vegetation Health Index (VHI) by integrating VCI and TCI for agricultural drought detection. The PCI is a condition index that provides insights into weather conditions [162], while SPI, originally proposed by [170], identifies and tracks local droughts based on long-term precipitation data, typically spanning 3, 6, or 12 months, requiring at least 30 years of precipitation dataset [171]. It remains one of the most widely used drought indices globally. SMCI tracks soil moisture dynamics and is particularly suited for monitoring drought duration and impact across large geographical territories [163,166,172]. It has shown strong correlations with short-term SPI [160] and VCI indices [172]. The Evapotranspiration Condition Index (ETCI) significantly influences agriculture and water resource management, reflecting crop water deficits level in agricultural zones [173]. It is valuable for assessing vegetation’s efficient use of available water resources. Recent studies have increasingly combined these indices for drought monitoring and detection in agriculture and meteorology. For instance, Du et al. [162] proposed the Synthesized Drought Index (SDI) based on MODIS and TRMM data, combining VCI, PCI, and TCI for vegetation growth monitoring. Similarly, Liu et al. [172] introduced composite drought indices (MCDIs) combining VCI, TCI, PCI, and SMCI to detect drought in Shandong Province on the east coast of China. The same combination of indices from MODIS, TRMM, and AMSR-E was used by [174] to establish the Optimized Meteorological Drought Index (OMDI) and Optimized Vegetation Drought Index (OVDI). Additionally, several composite drought indices have been proposed, including the Scaled Drought Condition Index (SDCI) [166], Combined Drought Index (CDI) [157], Combined Drought Monitoring Index (CDMI) [175], Microwave Integrated Drought Index (MIDI) [163], Multivariate Drought Index (MDI) [176], and Process-based Accumulated Drought Index (PADI) [177], yielding significant findings for researchers.

Table 2. Classification of drought indices.

Drought Severity	TCI, VCI, PCI, SMCI, ETCI & VHI Values	SPI Values
Exceptional drought	VCI ≤ 10	SPI ≤ −2
Critical drought	10 < VCI ≤ 20	−2 < SPI ≤ −1.5
Moderate drought	20 < VCI ≤ 30	−1.5 < SPI ≤ −1
Slight drought	30 < VCI ≤ 40	−1 < SPI ≤ 0
No drought	VCI ≥ 40	SPI ≥ 0

provides a classification of drought indices discussed by [157].

Researchers should prioritize the development of integrated platforms that combine remote sensing indices with machine learning algorithms to enhance real-time drought monitoring capabilities. This approach, combined with the integration of multiple indices like VCI, TCI, PCI, SMCI, and ETCI, provides nuanced insights into drought severity and impacts across diverse landscapes. In addition, it will empower stakeholders with actionable insights for implementing adaptive agricultural practices and optimizing water resource management strategies in response to increasing climate variability. Such integrated systems are crucial for fostering resilience in agriculture and ensuring sustainable water use in the face of evolving environmental challenges.

5. Challenges and Limitations

While remote sensing offers remarkable capabilities for monitoring Argania spinosa, it also faces several challenges and limitations. In arid regions where Argania spinosa thrives, cloud cover can impede satellite imagery acquisition, limiting data availability. Additionally, the need for ground truth data to validate remote sensing results can be challenging, especially in remote areas with dense tree cover that can hinder GPS accuracy and accessibility to specific points. In such environments, satellite signals can be obstructed or weakened by foliage, causing inaccuracies in location determination, while navigating dense vegetation to reach data collection points can be labor-intensive, impacting fieldwork efficiency and data reliability. Data accuracy may also vary due to factors such as sensor calibration and atmospheric interference. The spatial resolution of satellite images can indeed vary, and lower-resolution images may not be sufficient for detailed image classification, especially in heterogeneous landscapes where high-resolution data are crucial for distinguishing small-scale features and ensuring accurate monitoring. Furthermore, image distortions in uneven terrain can indeed affect the quality and accuracy of remote sensing data. Terrain variations can cause geometric distortions and shadowing effects, which can complicate image analysis and interpretation. Additionally, the digitization footprint associated with different sensing techniques, including the storage, processing, and computational demands, can present significant challenges. The interpretation of remote sensing data requires expertise, and there can be limitations in distinguishing between stress factors, like drought and disease, solely through spectral signatures. Moreover, the cost of acquiring and processing remote sensing data can be a barrier for some research and conservation efforts. Despite these challenges, ongoing advancements in technology and collaboration among researchers offer promising avenues to address these limitations and further enhance the effectiveness of remote sensing in monitoring Argania spinosa health, productivity, and response to drought stress.

6. Conclusions

In summary, this review article has underscored the pivotal role of remote sensing technology in advancing our understanding of forest ecosystems, with a specific focus on the Argane forest stands (Argania spinosa). Remote sensing tools, including satellite imagery, LiDAR, drones, radar, and GPS precision, have revolutionized our capacity to remotely gather data on forest cover, health, and responses to environmental changes. The Argane stand, known for its resilience in arid conditions and ecological significance in combating desertification, serves as a compelling case study for remote sensing applications. These technologies provide precise insights into canopy height, density, and individual tree health, enabling assessments of Argane stand populations, stress detection, and conservation evaluations. Furthermore, remote sensing plays a crucial role in monitoring vegetation health, productivity, and drought stress through various indices, contributing to sustainable land management practices. Additionally, the review briefly touched on the role of remote sensing in estimating crop yield and monitoring drought conditions, showcasing the diverse applications of this technology. This review underscores the transformative impact of remote sensing in safeguarding forest ecosystems, particularly the Argane stands, and highlights its potential for continued advancements in ecological research and conservation efforts.