Using Remote Sensing to Address Green Heritage of the City of Marrakech, Morocco

Abstract

Climate change and rapid urbanization have significant impact on green spaces and natural resources in African countries. To investigate this impact in the city of Marrakech, this study develops remote-sensing data to monitor changes in land cover and land use from 1990 to 2020. Results show almost 35% diminution of vegetation cover over the investigation period. In 1990, the city of Marrakech had a vegetation cover of 4.2 km², which fell to 2.7 km² in 2020. The main change occurred between 1990 and 2000 with a decrease of 13.7%, which is essentially due to the increase in build-up areas, related to the rapid growth of the city’s population. This evolution in land cover affects the urban environment negatively including air quality and temperature regulation. This research provides a better understanding of changing trends, confirms the importance of using satellite imagery to monitor vegetation cover in urban settings, helps determine efficient environmental management, and affects successful green infrastructure policy and planning, thereby allowing for improved adaptation and mitigation to climate change.

Highlights

• Remote sensing and GIS confirm the potential benefits of their use to develop urban planning policies that address green heritage.

• Land classification of Marrakech confirms that part of vegetation cover and fallow land were converted to built-up areas from 1990 to 2020.

• Green resources are under significant pressure due to the expansion of the city.

• To guarantee a decent ratio per capita, a plan for environmental improvements assess specific green spaces needs by implementing or preserving existing green heritage.

RÉSUMÉ

Le changement climatique et l‘urbanisation rapide ont d‘immenses répercussions sur les paysages et les ressources naturelles des pays africains. Pour étudier cet impact au sein de la ville de Marrakech, cette étude utilise des données de télédétection pour surveiller l’évolution des espaces verts et l‘utilisation des terres de 1990 à 2020. Les résultats montrent une diminution de près de 35% du couvert végétal au cours de la période étudiée. En 1990, la ville de Marrakech avait un couvert végétal de 4,2 km² contre 2,7 km² en 2020. Le principal changement s‘est produit entre 1990 et 2000 avec une diminution de 13,7% qui est essentiellement due à l‘augmentation des zones bâties, liée à la croissance rapide de la population de la ville. Cette évolution de l‘occupation des sols a des répercussions négatives sur l‘environnement urbain, notamment sur la qualité de l‘air et la régulation de la température. Cette recherche permet de mieux comprendre l‘évolution de ces tendances, confirme l‘importance de l‘utilisation de l‘imagerie satellitaire pour surveiller le couvert végétal en milieu urbain, aide à déterminer une gestion efficace de l‘environnement et influe sur la réussite de la politique et de la planification des infrastructures vertes, permettant ainsi d‘améliorer l‘adaptation au changement climatique et l‘atténuation de ses effets.

FAITS MARQUANTS

• La télédétection et les SIG offrent de nombreux avantages pour le développement de politiques de planification urbaine qui prennent en compte le patrimoine vert.

• La classification des terres de Marrakech confirme qu’une partie de la couverture végétale et des terres en friche ont été converties en zones bâties entre 1990 et 2020.

• Les ressources vertes sont soumises á une pression importante en raison de l’expansion de la ville.

• Pour garantir un ratio décent par habitant, un plan d’amélioration de l’environnement évalue les besoins spécifiques en espaces verts en mettant en œuvre ou en préservant le patrimoine vert existant.

Introduction

The world continues to urbanize. By 2030, predictions indicate that around 60.4% of the human population will have moved to urban areas (UN-Habitat 2020).

However, living in these cities, more people are directly facing the consequences of the global climate change, such as increased extreme weather events, loss of land ecosystems, decreased biodiversity, and a loss of the services normally provided by the ecosystems which have been lost (IPCC 2019; Liu and Russo 2021).

As these urban areas evolve diachronically, they impose a new spatial configuration that we need to understand in order to moderate the potential impact on social, economic, ecological, and physical systems. It is widely accepted that throughout human history, nature has provided food, energy, and protection (Xing et al. 2017). The presence of vegetation in cities that are strategically planned and managed help provide a range of ecological, social, and economic benefits (Wolch et al. 2014; Azagew and Worku 2020).

As many authors reported, vegetation helps reduce greenhouse gas emissions and mitigate climate change (Gill et al. 2007). Green spaces can also contribute to ameliorating human health (Pataki et al. 2011) physically, psychologically and socially (Jennings et al. 2016). Vegetation also provides esthetic benefits to cities (Kozlowski and Song 2022).

Despite these facts, significant parts of ecosystems are being lost, replaced with large impermeable surfaces (Xing et al. 2017) and degraded as a result of human activities (Directorate-General for Research and Innovation (European Commission) 2015).

This study fills a crucial gap in quantitative data on the evolution of green heritage in rapidly expanding African cities (Zhou et al. 2014; Jiao and Liu 2020). Landsat satellite imagery makes it possible to map urban vegetation efficiently thanks to its sensitivity to the spectral properties of vegetation (As-syakur et al. 2012). This approach has been applied to quantify changes in tree cover in many cities in different countries (Alqurashi and Kumar 2013).

In Morocco, the rapid urbanization of the post-Protectorate period (i.e., since 1956) has produced undesirable effects on urban territories, the most important of which are: residential sprawl, waste and scarcity of natural resources, socio-spatial and environmental segregation. In this context, the question of urbanization of green spaces is particularly relevant in these recent years and remains a major issue for the city of Marrakesh. Although its location is strategic because of its tourist dynamism and its worldwide influence, its vegetation has been and continues to be a source of its appeal, the evolution of these green areas is threatened as they are considered rare land reserves located in a dense intra-urban perimeter. From this perspective, a sustainable future for the city depends on reinventing and rethinking the current and future planning methods of urban green spaces.

One of the objectives in this paper is to investigate how the vegetation of the Marrakech has evolved over time and the distinctive characteristics and causes that triggered their neglect and abandonment. Preserving green spaces can be a concrete response to the current crisis of the city in its quest for renewal post covid. Hence, it plays a potentially important role in adapting to some of the anticipated impacts of climate change (Matthews et al. 2015). This paper examines the spatiotemporal evolution of vegetation in urban contexts using the city of Marrakech as a case study and addresses its major causes, challenges and resulting impacts. We first mapped and calculated the Normalized Difference Vegetation Index NDVI and Normalized Difference Built up Index NDBI from 1990 to 2020 based on the land cover data in ArcGIS. Following that, we utilized physical indexes such as NDVI and NDBI, related to climate data (Temperature and precipitation) (Li and Qu 2019) as support for mapping and monitoring land use and occupation, mainly vegetation in urban areas, due to, among other factors, changes in the energy balance and, consequently, increase heat of cities. By means of these indexes, it was possible to acquire accurate estimations for the land cover. To evaluate the vegetation cover changes in Marrakech, then we calculated a ratio per vegetation resident and compared three time periods (1990–2000, 2000–2010, and 2010–2020), and analyzed results for each 10-year period of the city. Finally, we examined the challenges behind the current states of vegetation based on documents review, and field observation. Although the use of these indices on time series images has been reported in previous literature, our study makes an important contribution by using a supervised classification approach to extract vegetation and land cover data from remotely sensed images. This method allowed us to obtain more accurate and detailed information about vegetation changes over time. It is hoped that these results will contribute to furthering our understanding of the spatial pattern of the city, and provide a helpful reference for future policies of planning and managing actions related to green spaces of Marrakech.

Materials and methods

Study area

Marrakesh city is located in the middle of Morocco between 7°53′∼8°5′ W longitude and 31° 37′ 48ʺ N latitude, with an area of 211,90 km2 (Figure 1). Marrakesh is the most visited city in Morocco and the frequently reported on by the international media (Kurzac-Souali 2011). It is well known for its heritage, its culture, its geographical proximity to Europe and its ability to attract tourists. The classification, since 1985, of the medina and the main historical monuments of Marrakech as UNESCO World Heritage Sites, is recognition of the importance of this city at the world scale.

Figure 1. Location of the study area Marrakech in Center of Morocco.

Founded nearly 1,000 years ago (Deverdun 2004), the urban landscape varies in an impressive way. It is composed of a plurality of systems retaining the traces of the succeeding dynasties: an old core called the medina, an urban space produced by the colonization as well as the peri-urban centers in extension (Figure 2).

Figure 2. Early green spaces created with traditional landscaping and hydrological technology in the urban fabric of Marrakech City. (a) Menara gardens planted with olive trees laid out during the period of the Almohad Caliphate (1121–1269); source: Les Jardins de l‘Agdal—Jardins et site historique sur My Little Kech (b) Marrakech palmeraie (“palm grove”) listed and protected as historical monuments.

Climatic context of Marrakech City from 1970 to 2020

Climate factors, such as temperature and precipitation, impact vegetation land cover (Li et al. 2017), for this reason, it is important to investigate the climate factors of Marrakech. To obtain monthly precipitation statistics from the Tensift Water Basin Agency and temperatures over the course of the last 40/50 years (1970–2020) for precipitation and (1981–2020) for temperature, we used the NASA Prediction of Worldwide Energy Resources (POWER) Project online resource (https://power.larc.nasa.gov/).

Figure 3 presents the variation of interannual monthly temperatures at 2 meters from the surface of the city of Marrakech, we can clearly see the gradual increase in temperatures over 30 years which can be explained by regional climate change. Figure 4 shown depicts the station’s monthly rainfall from 1970 to 2020. In 1970, the station’s rainfall reached an average of 450 mm; however, over time, maximum rainfall decreased by about 100 mm. In 2008, however, we notice an increase of 400 mm, and after this year, we start to see the effects of global change reflected by a decrease in rainfall over the previous century.

Figure 3. Average yearly temperature of Marrakech City. Source: NASA Prediction of Worldwide Energy Resources.

Figure 4. Average monthly rainfall of Marrakech City (1970–2020). Source: Tensift Water Basin Agency.

Population of Marrakech

According to the latest census in 2014, the population of the province of Marrakech totaled 1,330,468 inhabitants (MONOGRAPPHIE-DE-LA-REGION-DE-MARRAKECH-SAFI 2015), compared to 1,070,838 in 2004 (Huang et al. 2021), an average annual growth rate of 2.2%. The population is expected to increase through 2040. Thus, according to calculations of the Master Plan for Urban Planning and Development (SDAU), 5,00 additional hectares are needed for the future development of the city. In this study, the inhabitants corresponding to the years (1990–2000–2010–2020) were downloaded from the following site: https://www.city-facts.com/marrakech as they nearly match the numbers of the census made by the government.

Data collection and preparation

The majority of the data used in this study come from Landsat 4-5 Thematic Mapper (TM) Level-2 and Sentinel-2 L2A satellite images, which were downloaded from the Sentinel Hub website at www.sentinel-hub.com (accessed on October 7, 2022). These images cover the years 1990, 2000, 2010, and 2020 (Table 1). In order to improve the accuracy of the indices that will be processed, we downloaded images in the spring, where the weather is less cloudy specifically between April and May. In order to reduce the number of outside influences that lower the quality of satellite products, we applied an atmospheric correction to the images. Then we cropped the photographs to correspond to the administrative borders of Marrakech.

Table 1. Remote sensing image information.

Year	Date acquired	Type	Landsat Scene Identifier	Path	Row
1990	1990-04-04	Landsat 4-5 Thematic Mapper (TM) Level-2	LT52020381990094FUI00	202	38
2000	2000-04-15	Landsat 4-5 Thematic Mapper (TM) Level-2	LT52020382000106FUI00	202	38
2010	2010-05-29	Landsat 4-5 Thematic Mapper (TM) Level-2	LT52020382010149MTI00	202	38
2020	2020-05-26	Sentinel-2 L2A	L1C_T29SNR_A025731_20200526T112106	–	–

Satellite image analysis and mapping

Supervised classification

In this study, we use Geographic Information Systems (GIS) to manage spatial information about urban, vegetation, and many other natural resources. To extract Marrakech’s urban and vegetation, we use the supervised classification with ArcMap toolbar, which classifies images using spectral signatures derived from training samples. In order to track the progress of urban areas and compare them to green spaces throughout the entire city of Marrakech, we compare 4 images from 1990 to 2020.

The choice of a supervised classification is justified. Previous studies have shown that the spectral signatures of urban vegetation can be accurately distinguished from other surfaces using training data (Liu and Yang 2015). The use of local training data makes it possible to optimize the results (Maxwell et al. 2018). This supervised approach is standard in vegetation remote sensing.

NDVI

NDVI has been used to examine temporal trends and spatial patterns of vegetation (Li et al. 2017) during the past decades (Fensholt et al. 2012; Jeyaseelan et al. 2007; Jong et al. 2012; Julien et al. 2006). We use the 4 satellite images that were used in the supervised classification earlier to calculate the NDVI. Sentinel Hub was used to select the satellite images (Datum: 84 WGS/EPSG: 4326). We have selected satellite images taken during this time of year because spring marks the peak of biological activity in Morocco’s climate-controlled regions. Using the (GIS) and bands 3 and 4 from Landsat 4–5, as well as bands 8 and 4 from Sentinel 2, we estimated the area occupied by Marrakech’s vegetation. The formula (Equation (2)) used to determine NDVI is shown on Table 2.

Table 2. (1): Equation for NDVI.

Index Satellite image	Satellite image
NDVIL5 = (Band4 − Band3)/(Band4 + Band3)	Landsat 4, 5 TM
NDVIS2 = (Band8 − Band4)/(Band8 + Band4)	Sentinel 2

For Landsat 4-5, NIR is the reflectance at the wavelength that corresponds to the near infrared (band 4), while RED is the reflectance at the wavelength that corresponds to red (band 3) for Seninel 2. The NDVI values range from −1.0 to +1.0; healthy, active, and dense vegetation has NDVI values above 0.5; scarce vegetation and/or a low-activity plant phenophase are associated with NDVI values between 0.2 and 0.5 (Omar and Kawamukai 2021). NDVI values for the 4 images taken in spring are between −0.29 and 0.91. Positive NDVI values are increasing, which shows that there is more green vegetation present. The NDVI values close to 0 and decreasingly negative values represent non-vegetated objects, such as water, snow, ice, clouds, and barren surfaces (rock and soils).

NDBI

Built-up areas, one of the significant land cover types, are introduced by the Normalized Difference Built-up Index (NDBI) (Zha et al. 2003). Urban built-up lands are extracted from Landsat TM/ETM and Sentinel 2 images to show how the Normalized Difference Built-Up Index works. The NDBI index is dependent on the integration of bands 4, 5, and 8 and 11 in Sentinel 2 and TM, respectively, as shown by the Equation (2) in Table 3. NDBI has a value that lies between 0 and 1. According to research, positive NDBI values correspond to urban land areas, whereas negative NDBI values correspond to non-urban land regions (Zheng et al. 2021).

Table 3. (2): Equation for NDBI.

Index Satellite image	Satellite image
NDBI L5 = (Band5 − Band4)/(Band5 + Band4)	Landsat 4, 5 TM
NDBI S2 = (Band11 − Band8)/(Band11 + Band8)	Sentinel 2

Land surface temperature (LST)

For many ecological, hydrological, and atmospheric processes, land surface temperature (LST) is a crucial physical parameter. LST represents the flow of surface energy and moisture between the atmosphere and the biosphere in physical form (Li et al. 2023). In this study, the land surface temperatures for urban areas in Marrakech City are retrieved using data from the Landsat TM5 satellite and Landsat 7 pictures. Bands 1 through 5 and Band 7 of the satellite photos have a spatial resolution of 30 meters each. Band 6 (thermal infrared) spatial resolution is 120 meters, but is resampled to 30-meter pixels. From 1990 through 2020, we collect 4 satellite photos, selecting one image every 10 years.

The goal of the study was to evaluate the effectiveness of LST retrieval methods using a variety of NDVI-based land surface emissivity (LSE) models using information from past and present Landsat missions. Data from Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper Plus (ETM+), and Landsat 8 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS) were used to evaluate LST retrieval algorithms using the Radiative Transfer Equation (RTE) method, Single Channel Algorithm (SCA), and Mono-Window Algorithm (MWA) (Jiménez-Muñoz and Sobrino 2003; Price 1983; Qin et al. 2001; Schmugge et al. 2002).

There are various processes in the LST calculation process. This procedure was adopted from the USGS Landsat 8 Data User Handbook (Vaughn 2016). The Thermal Infrared Digital Number (DN) must first be converted into Top of Atmosphere Radiance (TOR) (TOR) (Equation (3)(3) $$\mathrm{L}\lambda =\text{ML}*\text{QCAL}+\text{AL}–\text{Oi}$$(3) ). (3) $$\mathrm{L}\lambda =\text{ML}*\text{QCAL}+\text{AL}–\text{Oi}$$(3)

TIRS 1 is band 10 (10.6–11.19 m). When L is the TOR spectral radiance in (Watts/(m² sr m)), 100 m of Landsat 8 have been utilized. ML is the multiplicative band value for Radiance. The radiance add band value is AL. A quantized calibrated pixel value in DN is known as QCAL. Oi, which equals 0.29, is the Band 10's correction value.

The spectral radiance data must then be transformed into Top of Atmosphere (TOA) Brightness Temperature in the subsequent phase using Equation (4)(4) $$\text{BT}=\left(\mathrm{K}2/\left(\hspace{0.17em}\ln \hspace{0.17em}\left(\mathrm{K}1/\mathrm{L}\lambda \right)+1\right)\right)–273.15$$(4) : (4) $$\text{BT}=\left(\mathrm{K}2/\left(\hspace{0.17em}\ln \hspace{0.17em}\left(\mathrm{K}1/\mathrm{L}\lambda \right)+1\right)\right)–273.15$$(4)

BT refers to the temperature at the top of the atmosphere. L is the TOR spectrum radiance as determined by Equation (3)(3) $$\mathrm{L}\lambda =\text{ML}*\text{QCAL}+\text{AL}–\text{Oi}$$(3) . In the image set metadata, K1 and K2 represent the K1 and K2 constant values of Band 10, respectively. The mean value of surface elements as determined by NDVI is known as land surface emissivity. First, using Equation (5)(5) $$\text{PV}=\left(\left(\text{NDVI}–\text{NDVI min}\right)\left(\text{NDVI max}-\text{NDVI min}\right)\right)2$$(5) , the Proportion of Vegetation (PV) has been determined. (5) $$\text{PV}=\left(\left(\text{NDVI}–\text{NDVI min}\right)\left(\text{NDVI max}-\text{NDVI min}\right)\right)2$$(5)

Here, NDVI has been identified using Equation (1) (Table 1). The DN values for NDVI's minimum and maximum are designated as NDVImin and NDVImax. The calculation of Emissivity (E) has since been done using Equation (6)(6) $$\mathrm{E}=0.004*\text{PV}+0.986$$(6) . (6) $$\mathrm{E}=0.004*\text{PV}+0.986$$(6)

PV was gathered from Equation (5)(5) $$\text{PV}=\left(\left(\text{NDVI}–\text{NDVI min}\right)\left(\text{NDVI max}-\text{NDVI min}\right)\right)2$$(5) , and the associated correction value is 0.986. The final stage was calculating the LST (7) using the BT and E values from Equations (4)(4) $$\text{BT}=\left(\mathrm{K}2/\left(\hspace{0.17em}\ln \hspace{0.17em}\left(\mathrm{K}1/\mathrm{L}\lambda \right)+1\right)\right)–273.15$$(4) and (6)(6) $$\mathrm{E}=0.004*\text{PV}+0.986$$(6) . (7) $$\text{LST}=\left(\text{BT}\right)/\left(1+\left(\lambda *\text{BT}/\left(\text{hc}/\mathrm{x}\right)\right)*\hspace{0.17em}\ln \hspace{0.17em}\left(\mathrm{E}\right)\right)$$(7) where λ is the emission wavelength, c the speed of the light travel, h the Planck’s constant and s the Boltzman’s constant. The entire LST extraction code may be seen in the section Algorithm.

Algorithm

of LST extraction

1. var landsat = ee. ImageCollection ('LANDSAT/LC08/C01/T1_SR')

2. filterBounds (geometry)

3. filterDate('2020-11-01, '2021-01-31′)

4. filterMetadata('CLOUD_COVER', 'less _than’, 5). mean (). clip (geometry);

5. var ndvi = landsat.normalizedDifference (['B5', 'B4'1).clip(geometry);

6. var thermal = landsat.select ('B10').multiply(0.1);

7. var fv =(ndvi.subtract (min). divide(max.subtract (min))).pow(ee. Number (2)). rename ('FV');

8. var a = ee. Number (0.004);

9. var b = ee. Number (0.986);

10. var EM-fv.multiply(a).add(b). rename ('EMM');

11. var LST = thermal.expression(

12. '(Tb/(1 + (0.00115* (Tb/1.438))*log (Ep)))-273.15′,

13. 'Tb’: thermal. select ('B10'),

14. 'Ep’: EM.select('EMM')

15). rename ('LST'). clip (geometry);

World Settlement Footprint

The World Settlement Footprint (WSF) 2019 is a binary mask with a 10-m resolution that depicts the spread of human settlements worldwide using multitemporal Sentinel-1 (S1) and Sentinel-2 (S2) imagery from 2019 (Asam et al. 2022). In order to generate temporal statistics for both S1- and S2-based indices, it is assumed that settlements behave more consistently than most land-cover classifications. The map was created using QGIS and the WSF plugin.

Global Human Settlement Layer

We downloaded the data relating to the global human settlement layer from the website https://ghsl.jrc.ec.europa.eu/ for 4 prospective years 1990, 2000, 2010, and 2020 in order to end up with maps that will mark the urban development on the city of Marrakech. The Global Human Settlement Layer (GHSL) project creates global geographical data about the evolution of human habitation on the earth. This in the form of settlement maps, built-up maps, and maps of population density. New geographic data-mining tools were used to generate this information utilizing knowledge and analytics based on empirical data.

Vegetation indicators and calculations

As previously mentioned, the classification map shows three different types of cover (in this case: vegetation, built-up areas, fallow land). We further determined the area occupied by each component of the city. Table 4 records the total of the three components for the city of Marrakech. Based on the NDVI values, the urban vegetation cover ratio per capita provides a better understanding of the availability of green vegetation for each individual in the city and allows for tracking changes over time. This can help decision makers evaluate the effectiveness of urban development and environmental management policies.

Table 4. Changes over 1990, 2000, 2010, and 2020 in the indicators for the vegetation categories at the city of Marrakech per resident.

	Area m^2 (1990)	Area m^2 (2000)	Area m^2 (2010)	Area m^2 (2020)
Vegetation cover	52,340,279.36	47,114,042.47	40,800,174.9	37,665,056.05
Population	801,158	1,027,426	1,245,217	1,463,008
Vegetation cover per resident	65.33	45.85	32.76	25.74

The percentage of urban vegetation cover is also important because it provides a measure of the amount of vegetation relative to the total area of the city. This can help planners to compare vegetation levels between different urban areas and identify areas that require additional efforts to increase vegetation.

In general, these urban vegetation indicators can help the policy and environmental management decisions in urban areas, contributing to more sustainable and environmentally friendly planning. For the calculation of vegetation cover, we generated the following two indicators for the city for the years 1990, 2000, 2010, and 2020 based on the urban vegetation cover as follows: $$\text{Urban}\text{ vegetation }\text{per resident}\left({\mathrm{m}}^2/\text{person}\right)=\text{urban}\text{ vegetation }\text{cover}\left({\mathrm{m}}^2\right)/\text{population}$$ $$\begin{array}{l}\text{Urban vegetation}\left(\%\right)\\ =\text{urban}\text{ vegetation }\text{cover}\left({\mathrm{m}}^2\right)/\text{total area}\left({\mathrm{m}}^2\right)\\ \times 100,\text{based }\text{on}\text{ the }\text{values}\text{ of }\text{the}\text{ vegetation }\\ \text{cover}\text{ per }\text{person}\text{ for }1990,2000,2010\text{ and }2020\end{array}$$

To assess the indicators’ temporal changes between 1990 and 2020, we made the following calculation (Equation (8)(8) $$\Delta \mathrm{t}2\text{-}\mathrm{t}2\left(\%\right)=\left(\text{parameter t}2-\text{parameter t}1\right)/\text{parameter t}1\times 100$$(8) ): (8) $$\Delta \mathrm{t}2\text{-}\mathrm{t}2\left(\%\right)=\left(\text{parameter t}2-\text{parameter t}1\right)/\text{parameter t}1\times 100$$(8) where t1 and t2 represent the starting and the ending year, respectively.

The flowchart of Figure 5 summarizes the methodology.

Figure 5. Methodology flowchart.

Results

NDVI, NDBI, and LST

Figure 6 shows the NDVI maps for Marrakech City. The minimum values of the NDVI, which vary from one zone to the next and are denoted by a drop over time, can be used to interpret the decrease of the green zone that is evidently occurring in the city’s north and southwest. The map makes it possible to compare various areas visually and serves as a baseline for future requirements that will drive urban transformations by highlighting urban patterns.

Figure 6. NDVI maps for Marrakech City from 1990 to 2020.

The NDBI index maps suggest that Marrakech City is seeing a development (Figure 7). The variation of NDBI index in the urban area of Marrakech NDBI value found that the built-up value has changed significantly between 1990 and 2020, respectively 0.32 and 0.87. In the middle of the city’s built-up area, these effects are particularly noticeable.

Figure 7. Variation of NDBI in the urban area of Marrakech (1990–2000–2010–2020).

The LST maps show the various surface temperatures across the metropolitan area of Marrakech City in both time and space (Figure 8). Effects of urbanization and spatial variation of surface temperature is highly exhibited. In 1990, it is evident that the temperature varies from 12° to 30° across Marrakech; however, the temperature is rising in the southeastern part of the city while it is decreasing in the north and southwest. The maximum and minimum temperatures in the city as a whole increased by 7 °C in 2000, after 10 years, but there was still a geographic contrast, with the exception of a slight increase in the north. In 2010, the maximum temperature increased to reach 51, and the majority of the city’s districts experienced an increase in temperature, with the exception of the city center and certain types of structures. In 2020, the minimum temperature increased to 26° and is characterized by a continued maximum temperature of 50 °C. This change is manifested in heat waves that last over lengthy periods, specifically in the summer.

Figure 8. Variation of LST in the urban area of Marrakech (1990–2000–2010–2020).

Numerous researches explore the LST pattern based on the correlation between LST and NDVI (Gutman and Ignatov 1998; Guha and Govil 2021). According to these studies, high LST is related to low vegetal covered areas in mixed urban land (Voogt and Oke 2003). Thus, significant relationship of LST with spectral indices (moderate negative for NDVI and strongly positive for NDBI) supports a sustainable urban planning (Mondal et al. 2021).

World Settlement Footprint and Global Human Settlement Layer

The World Settlement Footprint (WSF) map shows the progression (extension) from 1985 to 2015 for the region of Marrakech (Figure 9). This map depicts several orientations of the expansion along 5 axes (Barthel, 2013):

Figure 9. WSF map evolution from 1985 to 2015 of Marrakech City.

• Northern extension: exceeding the limits defined by the Oued Tensift and the Palmeraie. Beyond these natural limits, residential and tourist projects are installed at the expense of the green belt (the Palmeraie) and blue one (Oued Tensift).

• North-West extension: the creation of an industrial and residential zone.

• Western extension: Establishment of large-scale residential projects combining social housing and average medium standing.

• South-Western extension: With a greater amplitude, the current sprawl has been to the detriment of the irrigated perimeter and planted areas to be developed.

• South and South-East extension: Providing views of the Atlas Mountains, this area is characterized by residential and tourist activities.

Figure 10 shows the Global Human Settlement Layer (GHSL) for 1990, 2000, 2010, and 2020 as well as the distribution maps for the same years. This was done in order to have a clear idea on the evolution of the human presence on the city of Marrakech and to specify how this urban development was done during the period from 1990 to 2020. We note the extension of the central urban meshes to urban areas with human settlements and automatically they will have low NDVI values and loss of vegetation. The study shows that the distribution of green infrastructure across the city is being impacted as urban areas become more densely populated as confirmed by Lin et al. (2015).

Figure 10. GHSL (1990–2000–2010–2020) for Marrakech City and distribution maps of urban area of Marrakech City for the same period.

Supervised classification

Multi-temporal LULC maps (Weng 2006) covering three major classes: vegetation, fallow land and built-up areas of 1990, 2000, 2010, and 2020 are shown in Figure 11.

Figure 11. Land cover maps of Marrakech during years 1990, 2000, 2010 and 2020.

Relationship between LST and LUCC is widely reported (Chen et al. 2006; Xiao and Weng 2007; Thi Van and Xuan Bao 2010; Hereher 2017; Guha and Govil 2023). Evolution of land surface feature is affected directly by urbanization and the rapidly growing population. Moreover, the LST-NDBI correlation is strong on green vegetation; strong to moderate on built-up and open land (Guha et al. 2021). In order to manage this serious threat of human systems on the ecological one, this study recommends to well manage the remaining areas to promote the urban environment of Marrakech. It is worth mentioning that, according to the classification, Marrakech City shows extensive green coverage that constitutes an important green historical heritage. Despite this characteristic, built-up areas have a densification scenario significantly simulating the urban sprawl.

The percentage of each land cover classes for the 4 LULC maps are presented in Figure 12. Results from classified maps indicate that in 1990 area occupied by different classes; vegetation was about 24.70%, built-up areas covered 34.38%, fallow land occupied about 40.92% of the area. On the other hand, in 2010 about 19.25% of the area was covered by vegetation against 22.23% area in 2000 showed a decrease in vegetative areas. Fallow land and built-up area covered 32.29% and 48.46% respectively in 2010 while, the value was 38.04% and 39.72% in 2000.

Figure 12. Land Coverage in Marrakech, 1990–2020. Note that, during the 30-year period under investigation, built-up areas nearly doubled, as a percentage of total area, while vegetation decreased by about 25%.

Results from the classified image of 2020 illustrate that more than 66% of the area was covered by a built-up area, whereas vegetation was 17,77%. Fallow land occupied 16,22% (Figure 12). The analysis conducted in Marrakech between 4 situations revealed changes related to the urbanization process that corresponds to a general increase in built-up areas and a decline in areas with vegetation.

Change in vegetation cover for the city of Marrakech

While the population increased between 1990 and 2020, the urban vegetation areas decreased (Table 4). The multidirectional expansion of Marrakech City and the population growth occurred at the same time as these changes in the vegetation cover. We also determined the vegetation cover per person and their transformation through time. Table 4 displays the change for each parameter between 1990 and 2020. The declines in the vegetation indicators are shown in the last column. We also note the increase of population with growth rates of 1.41%, 1.05% and 0.87% (1990–2000, 2000–2010, and 2001–2020), respectively which led to greater changes.

To obtain more insight into the correlation between urbanization and impact on green space, we address the percentage of change of the vegetation cover in m²/person. Figure 13 shows the relative values of changes between 1990 and 2020 for the vegetation cover per person for the three stages (1990–2000), (2000–2010), (2010–2020) and the general duration (1990–2020). The third stage was slightly lower (8.3%), while the overall change (1990–2020) was more significant (34.89%). Academic research has widely approved the high pressure on green areas in growing cities, due to rapid urbanization in association with social and ecological systems (Fanan et al. 2011; Jim 2013; Qureshi 2010; Schetke et al. 2016).

Figure 13. Percent change of vegetation in m²/person for Marrakech.

The deceleration of change in green space for the period 2010–2020 is most likely due to renewed interest in planting vegetation in the city. To evaluate further the results, a visual representation at a higher resolution was made to identify this major phenomenon in the city. Figure 14 shows high-resolution images from Google Earth that demonstrate land use change over the period (2004–2020). This is one example among many of the replacement of vegetation with urban development. The land use change from green land to urban areas represents constant reclaim of Marrakech City and its expansion.

Figure 14. examples of visual comparisons between the accuracy of vegetation classification. (a) Location of the zoomed area in Marrakech City; (b) google earth images for the selected area that demonstrate land use change over the period (2004–2020).

Discussion

The mapping of green spaces was digitized, and areas were mapped in ArcGIS environment, we extracted areas and calculated the changes of green vegetation cover per person from 1990 to 2020. The analysis of these data was presented in maps and the results (supervised classification and changes per inhabitant) were presented in tables.

Over the years studied (1990–2020), most of the vegetation areas in Marrakech City were intensively converted to build up area land (either for residential, commercial, industrial use or for infrastructure roads and railways). If it continues unchecked, the alarming rate of change will have an immense impact on the local, regional and national environment. The land cover changes show a marked impact of increase in population in the study area combined with different climate context (augmentation of temperature and irregularity of rainfall).

Originally known as a garden city, the history of Marrakech from its foundation aimed to implement major green areas benefiting from its proximity to water sources, the Atlas Mountains and the rich soil of the Haouz plain. The extremely sophisticated green areas are the large basins of Menara and Agdal which made green spaces available and close to the entire population, this reflection has allowed the creation of a green belt and large areas of agriculture, not only sufficient for the city itself but even for the surrounding accompanied by an ingenious hydraulic system (El Faïz 2002). With the arrival of the French Protectorate system in 1912 and the introduction of the new system of green space such as: Promenades, integrating planted strips, trees, covered pedestrian crossings to protect them from dust and sun, of various widths…

Despite its high urbanization rate and the strong growth of the population during the last decades, Marrakech counts many public and private green spaces (73 gardens according to SDAU). Some of these spaces are still visible today, such as Arsat Moulay Abdeslam, Jnane El Harti.

The recent Marrakech of nowadays considers the green spaces in the heart of the city as large clearings and empty fragments that survive with a kind of compromise of coexistence with compact urban spaces. Population growth, lack of green policy management were analyzed as causes of land cover changes and loss of vegetation cover during the last decades.

As it was stated by the results of this research, these useful green sites suffer from progressive abandonment and widespread damage. The existence of urban landscapes that are connected with nature is a big asset to improve attractiveness for both locals and tourists. In relation to the multitemporal variation of spectral indices, it was possible to observe the reduction, practically by 35%, of vegetation cover during expansion in some parts of the urban fabric of the city, especially in its Central Region (médina) completely dense. This reduction is possibly related to the increase in constructed areas, as presented by the NDBI and, consequently, suppression of vegetation, considering that vegetation degradation is noticeable in the visual comparison between the years analyzed. Thus, the vegetation distribution highlights strategical areas to restore lost vegetation.

The growing city warmth is a major issue for sustainable city planning (Shahfahad et al. 2022). To support our findings, LST was introduced (Algretawee 2022). In fact, the presence of vegetation reduces urban heat and helps cooling urban areas and the results were comparable to other related studies performed worldwide. Rapid urbanization accelerates the land conversion process creating warming status for cities raising the rate of LST. Thus, most of the lands are converted into settlements and consequently the city knows an increase in built-up area and a decrease in forest-cropland (Pandey et al. 2022).

In future, many additional research works may be included. First, the in-situ vegetation data (ratio per inhabitant, areas change) may be compared and confronted to satellite data. Using technically suitable and viable alternative to Landsat-5 TM is also preferable to monitor vegetation cover (Gill et al. 2010). Finally, water-saving strategies adapted for the weather of Marrakech may be analyzed and taken into consideration in order to integrate appropriate eco-friendly irrigation solutions for a sustainable preservation of these vegetative covers. In the perspective that they ensure delivering various environmental, social and economic benefits. In different studies, methods to determine annual and seasonal changes observed in irrigated areas (Ben Khalfallah et al. 2021) can draw promising indicators to limit vegetation degradation.

Other urban territories located in Mediterranean climate zones, such as Lisbon in Portugal (De Almeida et al. 2022), have also experienced similar changes, highlighting the urgency of environmental improvement and sustainable development. A contrary positive illustration is the Friendship Square Park in Addis Ababa (Azagew and Worku 2020). Areas where land had previously provided various ecosystem services were transformed to promote the park development, resulting in the removal of precarious and unhealthy habitat areas.

Furthermore, the findings are in agreement with results published in the literature, highlighting the decrease and high pressure on natural areas as a direct consequence of rapid urbanization and climate change experienced over 30 years (Jiao et al. 2021). The city of Marrakech has experienced a remarkable revival of interest in green spaces over the past 5 years, both quantitatively and qualitatively by launching appropriate studies (Marrakech green plan in 2019), and preservation initiatives.

Conclusions

Multitemporal cartographic analysis of Marrakech reveals that the city undergone a multi-directional expansion and has experienced very significant changes in its land cover. Starting from the quality of the actual green areas, it is important to preserve them, which will generate a bigger variety of environmental services, and will increase the touristic and economic value of the city.

The findings reveal that, following the city’s development from 1990 to 2020, the green coverage has decreased, while the built-up area increased due to high population growth. Using the NDVI, NDBI and LST database has been generated for future in-depth studies aiming to evaluate vegetation cover. Hence, proper management of land use, including vegetation planning action, is mandatory to avoid their neglect and rather participate to their maintenance.

The method employed here provides extensive information about the spatiotemporal history of green spaces in the city of Marrakech that can be used for effective urban planning and for developing policy at the local and national levels. It can be used to promote stakeholder involvement.

The overall decrease in green space per capita in Marrakech is cause for concern, and this study of vegetation represents an initial step toward developing actionable policy to encourage a reversal of this trend. Fortunately, the city has an unusual resource upon which to draw: its own past. As noted above, Marrakech has extensive examples of traditional proactive creation of green space, particularly the Menara Gardens and Palmeraie, employing complex irrigation systems that have lasted for many centuries. Perhaps these can become the inspiration for modern urban planners wishing to add further vegetation in keeping with the history and esthetic of the cityscape. Beyond that, it is hoped that this article represents an important first step toward creating dialogue about development of urban green space throughout Morocco and all over the African continent.

As far as short-term prospects go, we intend to: (i) apply AI algorithms with combination of the proposed approach to improve the accuracy of vegetation cover monitoring; (ii) include other studies relative to economic and environmental analysis to support appropriate decisions. It is also recommended to add a quantitative validation based on a field sampling, as suggested by Guo et al. (2011). This would help to consolidate confidence in the results. Nevertheless, the 30-year multi-temporal analysis strengthens credibility, as other studies have shown (Du et al. 2016).

Acknowledgments

The authors would like to thank the Geosciences Laboratory of the Earth Sciences Department, faculty of science at Semlalia Cadi Ayyad University Marrakech, thank you to all the contributors who helped make this article possibly, providing all types of support and motivation. We thank also the local agencies for their support and data.

Declaration of competing interest

The authors declare that they have not known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

The data presented in this study are available in this article.