Understanding Land Changes for Sustainable Environmental Management: The Case of Basilicata Region (Southern Italy)

Sustainable land management is one of the key actions for the achievement of objectives set by the 2030 Agenda for Sustainable Development. In particular, land represents a fundamental resource to address issues of climate change, biodiversity preservation, maintaining ecosystem services, and at the same time ensuring shared prosperity and well-being. Therefore, it is necessary to activate strategies to monitor changes in land use and land cover in order to evaluate strategies for proper management. To do this, the new open source geospatial analysis tools and the increasing availability of remote sensed open data can allow the activation of methodologies for monitoring changes in land use and land cover in order to provide data usable in other research areas or, for example, to implement a decision support system for environmental sustainability. In this study, a GIS approach based on open remote sensing data has been used to perform a spatial analysis of land cover changes within the Basilicata region (Southern Italy) that is spatially expeditious yet accurate. The results showed a very evident land transformation with important repercussions on the environmental components. The ease of use of techniques makes this methodology replicable in other territory and can be used as a preliminary approach to sustainable development model.


Introduction
Sustainable Development Goal n.15 of the 2030 Agenda aims to "protect, restore, and promote the sustainable use of terrestrial ecosystems, sustainably manage forests, combat desertification, halt and reverse land degradation, and halt biodiversity loss." [1]. Integrated planning and management of land resources is an issue that had already emerged in Chapter 10 of Agenda 21 [2], which addresses the cross-cutting aspects of decision making for the sustainable use and development of natural resources, including soils, minerals, water, and the life that the earth comprises. This broad integrative view of land resources, which are essential to the life-support systems and productive capacity of the environment [3,4], underlies the discussions of Member States expressing the challenges posed to sustainable development by land consumption, desertification, and drought [5][6][7]. In this context, it is essential to activate all actions to monitor the land changes in order to prevent phenomena such as land degradation, which in many areas of the planet earth and in particular in the Mediterranean region, is causing enormous environmental and social impacts [8]. Therefore, for a sustainable environmental planning that allows land to become a renewable resource and to fight land consumption and degradation, it is important to use a multidisciplinary and holistic approach in which one of the preliminary analyses is based on the assessment of land system dynamics [9].
Land use (LU) and land cover (LC) data are powerful resources for various surveys and investigations, such as measuring physical and environmental impacts, setting well-The whole region is part of the Mediterranean biogeographic area and there is a climatic variability depending by the extreme orographic and morphological diversity of the territory which varies enormously even in a few kilometers. In fact, the south-western area has an oceanic/sub-oceanic climatic profile with precipitation values around 1500 mm of annual rainfall, which decreases almost to 300 mm going towards the internal and eastern areas that have continental profiles with areas classified as sub-arid. The Basilicata Region land in recent decades has undergone major transformations due mainly to land degradation [21], land abandonment [22], and increased oil production [23]. Moreover, the demographic trend shows an aging and shrinking of the population, with emigration of the workforce, as well as intra-regional migration to more urbanized areas [24].
The high land diversity and the low anthropic pressure (except in a few industrial areas) make the Basilicata region a hotspot for Mediterranean biodiversity which, however, needs new programming and planning strategies according to more stringent criteria of sustainability to make the territory more resilient.

MODIS Land Cover Database
MODIS is NASA's first mission to provide daily global coverage at a variable resolution starting at 250 m. The MODIS Moderate-resolution Imaging Spectroradiometer acronym consists of two satellites Earth and Water. The satellites are composed of a 36-band multispectral sensor and provide images of the entire Earth's surface every 1 and 2 days at different spatial resolutions 250 m, 500 m, 1 km, depending on the type of processing required and the level of correction needed. The two satellites Earth and Water were launched in December 1999 and May 2002 respectively. The spectral resolution is between 0.4 and 14.385 µ m.
For this analysis the product "MODIS Land Cover Type Combined" Earth + Water collection V6 was used, for the expeditious monitoring of land cover at a global scale with annual observation time intervals from 2001 to 2019 and with spatial resolution of about 500 m [25,26]. MODIS land cover products are a dataset widely used in the scientific community for several kinds of investigations [27,28]. The whole region is part of the Mediterranean biogeographic area and there is a climatic variability depending by the extreme orographic and morphological diversity of the territory which varies enormously even in a few kilometers. In fact, the south-western area has an oceanic/sub-oceanic climatic profile with precipitation values around 1500 mm of annual rainfall, which decreases almost to 300 mm going towards the internal and eastern areas that have continental profiles with areas classified as sub-arid.
The Basilicata Region land in recent decades has undergone major transformations due mainly to land degradation [21], land abandonment [22], and increased oil production [23]. Moreover, the demographic trend shows an aging and shrinking of the population, with emigration of the workforce, as well as intra-regional migration to more urbanized areas [24].
The high land diversity and the low anthropic pressure (except in a few industrial areas) make the Basilicata region a hotspot for Mediterranean biodiversity which, however, needs new programming and planning strategies according to more stringent criteria of sustainability to make the territory more resilient.

MODIS Land Cover Database
MODIS is NASA's first mission to provide daily global coverage at a variable resolution starting at 250 m. The MODIS Moderate-resolution Imaging Spectroradiometer acronym consists of two satellites Earth and Water. The satellites are composed of a 36-band multispectral sensor and provide images of the entire Earth's surface every 1 and 2 days at different spatial resolutions 250 m, 500 m, 1 km, depending on the type of processing required and the level of correction needed. The two satellites Earth and Water were launched in December 1999 and May 2002 respectively. The spectral resolution is between 0.4 and 14.385 µm.
For this analysis the product "MODIS Land Cover Type Combined" Earth + Water collection V6 was used, for the expeditious monitoring of land cover at a global scale with annual observation time intervals from 2001 to 2019 and with spatial resolution of about 500 m [25,26]. MODIS land cover products are a dataset widely used in the scientific community for several kinds of investigations [27,28].
The overall accuracy of the product is about 75% correctly classified, but the range in category-specific accuracies is great [26]. Within the product there are multiple layers, each describing the classification of land cover and land use according to different method-ological schemes. In particular, for this work the FAO (Food and Agriculture Organization) Land Cover Classification System land cover (LCCS1) and FAO Land Cover Classification System land use (LCCS2) classifications have been used.
The LCCS was defined to globally standardize land cover and land use classes. It is based on supervised classification and the use of ancillary data for post classification. In Table 1, the legend for the LCCS1 top-level classification that identifies land cover is reported and in Table 2 the legend for the land use classification resulting from the top-level analysis is reported. The two classifications, even if they have classes in common, allow for the characterization of the territory in a different way so as to be able to provide all the fundamental aspects for land planning.

Unclassified
Has not received a map label because of missing inputs.

Unclassified
Has not received a map label because of missing inputs.

LULC Changes: How to Detect
The "land cover" (LC) and "land use" (LU) terms are frequently improperly used as synonyms [29]. According to FAO [30], land cover is that portion of the globe's surface defined in space and recognized in terms of characteristics and properties included by features of the biosphere that are reasonably immutable or cyclically predictable. Land use, on the other hand, can be interpreted as the way space is occupied by human society, i.e., related to human activity and directly to the exploitation of land and its use by the population [31].
Thanks to the databases used in this study, it was possible to perform a two-stage transformation assessment immediately by taking advantage of common techniques for evaluating the dynamics of land changes [32]. All operations were performed with the open source software QGIS 3.16 [33].
In the first phase, the transformations of territory in terms of LU and LC have been evaluated through a comparison of the differences in quantitative terms of the surface area associated with each category of the LCCS1 and LCCS2. This first approach has been realized through simple operations: downloading the dataset from the official repository, clipping on administrative boundaries of Basilicata Region, realizing the layout maps, and computing the surfaces in hectares for each class of LU and LC. A year-by-year analysis was carried out in order to identify in more detail the trends of variations within each class. On the other hand, the second phase allows a detailed evaluation of the transformations. Using change detection techniques [34], a quantitative spatial analysis of the LULC data was performed. Using LULC maps from 2001 and 2019, it is possible to map changes and identify precisely how much of a given cover class has transformed into something else.
From a sustainable environmental planning perspective, the two approaches are indispensable and basic to effectively understand how and where LU and LC transformations have occurred.

Results
The mapping operations of land use and land cover with MODIS data have allowed for the realization of an expeditious assessment of the dynamics of transformation occurring in the Basilicata Region so as to draw some indications of criticalities useful for sustainable environmental planning.

Year-to-Year Quantitative Assessment of LULC
The first phase involved the quantitative year-by-year assessment of changes in LULC. The use of geographic tools coupled with an open source spreadsheet allowed one primarily to map and then to export the data in table format and then to graph them. Tables A1 and A2 show the year-by-year data from the LCCS1 FAO Land Cover Classification. Their analysis showed that the Deciduous Broadleaf Forests, Open Forests, Sparse Forests, and Dense Herbaceous classes are those quantitatively most representative of the study area and in which the dynamics of transformation are most evident. Their year-to-year trends are shown in Figure 2.

Year-to-Year Quantitative Assessment of LULC
The first phase involved the quantitative year-by-year assessment of changes in LULC. The use of geographic tools coupled with an open source spreadsheet allowed one primarily to map and then to export the data in table format and then to graph them. Tables A1 and A2 show the year-by-year data from the LCCS1 FAO Land Cover Classification. Their analysis showed that the Deciduous Broadleaf Forests, Open Forests, Sparse Forests, and Dense Herbaceous classes are those quantitatively most representative of the study area and in which the dynamics of transformation are most evident. Their year-toyear trends are shown in Figure 2. From the analysis of the data obtained from the classification that concerns the land cover ( Figure 3), it is shown in particular that the Dense Herbaceous class, in which a vegetation cover with annual coverage of 60% is dominant, has suffered a substantial loss from 2001 to 2019 of about 110,000 hectares while the Deciduous Broadleaf Forest class is the one that has shown a substantial growth from 2001 to 2019 equal to 26,300 hectares. The Sparse Forests class shows a major increase of almost 80,000 hectares as well.
An identical analysis was performed for the land use assessment based on LCCS2. MODIS data were tabulated for all years from 2001 to 2019 (Tables A3 and Tables A4). Land use classification allows for an analysis that is equally useful for sustainable planning purposes as it considers the presence of anthropic activities on the land. Analyzing the graphs shown in Figures 4 and 5, the decrease of the class Herbaceous Cropland class, which identifies cultivated areas with an area lost of about 100,000 hectares in the last 20 years and a growth of the forested areas identified by the Dense Forests and Open Forests classes, is immediately evident. In addition, an increase in areas classified as Urban and Built-Up should be noted, demonstrating a land consumption that should be evaluated in more detail.
Both analyses show that some classes, for some years, are not present in terms of area. This is due to the classification algorithms used that, for those years, did not find some classes. From the analysis of the data obtained from the classification that concerns the land cover ( Figure 3), it is shown in particular that the Dense Herbaceous class, in which a vegetation cover with annual coverage of 60% is dominant, has suffered a substantial loss from 2001 to 2019 of about 110,000 hectares while the Deciduous Broadleaf Forest class is the one that has shown a substantial growth from 2001 to 2019 equal to 26,300 hectares. The Sparse Forests class shows a major increase of almost 80,000 hectares as well.
An identical analysis was performed for the land use assessment based on LCCS2. MODIS data were tabulated for all years from 2001 to 2019 (Tables A3 and A4). Land use classification allows for an analysis that is equally useful for sustainable planning purposes as it considers the presence of anthropic activities on the land. Analyzing the graphs shown in Figures 4 and 5, the decrease of the class Herbaceous Cropland class, which identifies cultivated areas with an area lost of about 100,000 hectares in the last 20 years and a growth of the forested areas identified by the Dense Forests and Open Forests classes, is immediately evident. In addition, an increase in areas classified as Urban and Built-Up should be noted, demonstrating a land consumption that should be evaluated in more detail.
Both analyses show that some classes, for some years, are not present in terms of area. This is due to the classification algorithms used that, for those years, did not find some classes.

LULC 2001-2019 Dynamics Mapping
As reported earlier, a mapping operation was performed for each annual MODIS dataset from LCCS1 and LCCS2, resulting in maps such as those shown in Figure 6.

LULC 2001-2019 Dynamics Mapping
As reported earlier, a mapping operation was performed for each annual MODIS dataset from LCCS1 and LCCS2, resulting in maps such as those shown in Figure 6.
Using the raw raster map allowed for a change detection to spatialize the transformations. This operation is critical to understanding trajectories of change, i.e., identifying how different LULC categories have evolved over the years (Figures 7 and 8).
These data, even if very complex, can be aggregated and used to calculate changes (Tables A5 and A6). Moreover, they are a source of ancillary geodata that can be exploited for multidisciplinary surveys in a perspective of integrated spatial planning. All operations are easily implemented within common open source GIS applications without excessive time consumption and high performance tools.
The extrapolated data show the usefulness of both types of LUCL datasets. In fact, although there are similarities, different aspects of land transformation dynamics emerge from their analysis. The areas that have not changed are almost the same in terms of hectares, but from the analysis of LCCS1 we see that the most evident change in percentage terms is that between Dense Herbaceous to Sparse Forests. This information is generic but it is not possible to define how much of this change affected natural grasslands and how much affected cereal agricultural areas. Deepening the analysis with the LCCS2 dataset, it is possible to differentiate between Natural Herbaceous and Herbaceus Cropland classes by evaluating how much of the two classes has been transformed into the different forest classes. Both analyses show that some classes, for some years, are not present in terms of surface. This is due to the classification algorithms used which, for those years, did not find some classes. Pollutants 2021, 1, FOR PEER REVIEW 9 Using the raw raster map allowed for a change detection to spatialize the transformations. This operation is critical to understanding trajectories of change, i.e., identifying how different LULC categories have evolved over the years (Figures 7 and 8).  These data, even if very complex, can be aggregated and used to calculate changes (Tables A5 and A6). Moreover, they are a source of ancillary geodata that can be exploited for multidisciplinary surveys in a perspective of integrated spatial planning. All operations are easily implemented within common open source GIS applications without excessive time consumption and high performance tools.  These data, even if very complex, can be aggregated and used to calculate changes (Tables A5 and A6). Moreover, they are a source of ancillary geodata that can be exploited for multidisciplinary surveys in a perspective of integrated spatial planning. All operations are easily implemented within common open source GIS applications without excessive time consumption and high performance tools.

LULC Changes in Basilicata Region: The Land Abandonment Issue
The main dynamic that emerges from Basilicata region survey is that of abandonment of agricultural areas (mostly arable land and pastures characterized by herbaceous cover) and their subsequent reforestation through natural processes of secondary succession. Agricultural abandonment processes started in the 1950s but especially today we hear more about them as the rates of increase have increased greatly [35]. The causes are multiple in that this phenomenon is extremely complex and is very evident in the mountainous inland areas. The causes can be different and interconnected: abandonment due to low profitability of land, soil conditions unfavorable to agriculture, depopulation, lack of innovation, socioeconomic problems at local or regional level, and wrong agricultural policies [36,37]. This phenomenon has several positive and negative impacts, with trade-offs largely dependent on the specific spatial context [38]. One of the typical effects, especially in the Mediterranean context, is that of natural reforestation with secondary positive and negative consequences that need further study and investigation [39].
In Basilicata these processes are currently underway [40] and have also emerged from this study, but compared to other areas it seems to be proceeding faster with evident ecological and socio-economic problems [41].
In this area, as in other similar areas in southern Italy, following an increase in cereal cultivation favored by the "Agrarian Reforms for Southern Italy" and an increase in mechanization, there was an increase in the agricultural area used even in marginal areas with little vocation for cereal cultivation. In the following decades, with the changes in socio-economic conditions, the crises of the agricultural sector that have made economically unfavorable, for many areas, cereal farming and some agrarian reforms related to the Common Agricultural Policy (in particular the "set-aside" REG. CEE 1272/88), there has been a steady abandonment of agricultural activity [42]. These areas give rise to different landscapes in relation to climatic conditions, age of abandonment, management before and after abandonment, and disturbances that are triggered successively [43]. In some areas of Basilicata, the continuous process of abandonment could lead to significant problems of land degradation that should be investigated [44,45].
Finally, the survey shows that abandoned areas are subject to natural renaturalization and afforestation that make the study of territorial dynamics even more intricate because, even for this phenomenon, the impacts can be positive or negative in relation to the territorial context. [46,47]. All this information, obtained in this expeditious manner, can be easily obtained and subsequently incorporated within planning actions of sustainable models aimed at improving the resilience of territory. In this specific study, the abandonment of agricultural areas and the consequent renaturalization of some areas can determine positive or negative impacts related to the consumption of natural resources (soil, water etc.) and the CO 2 balance. Therefore, the spatialization of land use and land cover is the basis for all subsequent analyses.

How to Use LULC Data in Sustainable Environmental Planning
Appropriate LULC data have a key role in several sectors of planning as for example in natural resource management, food security, hydrology modelling, etc.
The problem of land transformation is one of those research fields that needs LULC geodata that are as up-to-date as possible and at the same time allow a standardized and repeated survey over time [48,49]. This is because the data considered are often raster or vector cartography of different origins, scales, reference systems, etc. [50,51]. In addition, replicating the study, even for the interest of monitoring the phenomenon, becomes problematic since the data used are not available in time and in the same way.
Many of the problems can be solved using a remote sensing approach and online data made freely available according to well-defined standards [52]. This ensures that it is possible to monitor the phenomenon and replicate studies over time and in different spatial contexts. In particular, the open source tools discussed here can be used to support the implementation of appropriate techniques to represent the dynamics of land transformation and at the same time be used for landscape monitoring and planning [16].
In view of this, in this case study the applicability of the MODIS LCSS1 and LCSS2 datasets were verified for the analysis of land transformations both from the point of view of land cover and land use and to evaluate the replicability of use in a perspective of implementation of sustainable environmental planning models.
The MODIS products, which can be freely used, provide an important support for assessments and preliminary quantifications of environmental and territorial transformations. Indeed, different types of data are available for a myriad of different applications [53,54]. This work considers the land cover products classified according to FAO nomenclature which allows for the differentiation of the land use from the land cover. In fact, this aspect is fundamental to understanding the actual transformation of territory not only from a physical point of view but also from a socio-economic one.
Compared to other products found online with similar accuracy, the LCCS MODIS data allows for annual mapping of the last two decades in an expeditious manner. This also allows their use as ancillary data for supervised classifications with other higher resolution remote sensing data. Moreover, since they are already classified, they can be easily used outside the scientific research environment. Finally, the lightness in terms of bytes of data allows analysis even with low performance hardware. The limits of applicability of this methodology are related to the resolution of data used which often does not allow for the identification of local dynamics that frequently have impacts that affect a larger scale.

Conclusions
In view of the achievement of some of the Sustainable Development Goals of the 2030 Agenda, environmental planning assumes a fundamental relevance as it allows for confronting territorial problems, such as the abandonment of agricultural areas in Basilicata, which are causing high negative impacts. Sustainable environmental planning is a process that aims to integrate ecological with the socio-economic principles for an overview of land management for present and future generations. In order to implement strategies for environmental planning, it is critical to collect, edit, and distribute timely and accurate data and to apply advanced land assessment and technologies to create scientific knowledge for suitable decision support systems. It is essential to implement tools that can allow an overview of the territory and that can ensure an effective and dynamic decision support system, thus meeting the needs of public decision makers to verify past management policies and developing new strategies. One of the basic approaches to developing more complex investigations and studies is the one illustrated in this work, which allows for the extrapolation of remote sensing data on land cover and land use and carries out analysis in an expeditious way both in research and in local actions.
Author Contributions: The design and conduct of this research is equally shared between the authors. The five authors collaborated to produce this paper. G.C., B.T., and V.S. proposed and developed the research design, methodology, manuscript writing, and data analysis, G.N. and A.L. supervised the work, provided additional comments on the results and interpretation, and reviewed and approved the final version. All authors have read and agreed to the published version of the manuscript. Data Availability Statement: The raw LCCS1 and LCCS2 data can be downloaded from the official MODIS databases at: https://modis.gsfc.nasa.gov/data/dataprod/mod12.php (accessed on 5 May 2021).The elaborated data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.