Land Use Land Change analysis: are the Spatial Development Plans for conserving the nature reserves in Aruba implemented as prescribed?

(1)

Land Use Land Change analysis: are the Spatial Development

Plans for conserving the nature reserves in Aruba

implemented as prescribed?

Name: Fleur Drost Student number: 12096512

Date: May 29, 2021 Place: Amsterdam

Supervisor: Dhr. dr. K.F. (Kenneth) Rijsdijk Assessor: Dhr. dr. ir. J.H. (John) van Boxel

Advisor: Dhr. J. (Jim) Groot MSc Sharona S. Jurgen MSc

(2)

Earth Sciences

1

Summary

Human impact on nature is visible in the loss of biodiversity in many places. Aruba is one of those examples and knows a variety of species losses. Therefore, land use planning is required to keep the areas of nature reserves as large as possible. In 2009 and 2019 Spatial Development Plans (ROPs) were published by the government of Aruba, these contain prescriptions for land use planning in Aruba. Despite the allocation of regions as nature reserves by the government, merely a few are realised as nature reserves. Therefore, it was investigated if the areas of nature reserves in Aruba have been implemented as prescribed.

A Land Use Land Change analysis was done with the use of ArcGIS to investigate the extent to which the ROP was executed. Three cloudless satellite images of Aruba from 2008, 2012 and 2020 were derived from Google Earth Engine. In ArcGIS, the three maps were classified into five land use types water, vegetation, rural, build-up and sand. Next up the accuracy of the classified images was assessed which has led to an accuracy of 97% for the map of 2008 and 2020 and 96% for the 2012 map.

The area of nature reserves as stated in the ROP of 2009 was 67.9 km2, the classified map of 2020 showed an vegetated area of 72 km2. With these results, it could be concluded that the areas of nature reserves were mostly implemented. Besides, the allocation of nature reserves in the classified map corresponded with the prescribed areas in the ROP. But the land uses on the classified images were identified more specific than the ROP, it was therefore more difficult to compare these two maps. Less dense vegetated areas on the satellite images were classified as rural areas. In the ROP rural areas are described as mostly nature with a few buildings such as farms. In the classified map, these

landscape features are classified as vegetation and build-up, it was therefore challenging to compare them with the rural areas of the ROP. In further research, it is necessary to create a classified map with larger classified areas. This will make comparing a classified map to the ROP more accurate.

Keywords: Land use Land Change analysis, LULC, ArcGIS, Aruba, nature reserves, biodiversity, nature preservation, spatial development

(3)

2

Index

Summary ... 1 Introduction ... 3 Methods ... 6 Results ... 7 Discussion ... 14 Conclusion ... 16 References ... 17 Appendices ... 19 Appendix 1 ... 19 Appendix 2 ... 21 Appendix 3 ... 24

(4)

Earth Sciences

3

Introduction

The impact of thousands of years of inhabitants is visible in the nature of the Caribbean island Aruba (Nooren, 2008). Aruba is known for its white beaches and ocean blue sea, the temperatures across the year vary from 28 degrees in winter to 32 in summer. This makes it an attractive place for tourism, which is an important part of Aruba’s economy (Ridderstaat, Croes & Nijkamp, 2014). Land use has changed drastically as an effect of habitation and tourism, Njue et al. (2016) showed that land use change and a decrease of vegetation lead to a loss of biodiversity. The impact of this is visible in the biodiversity of Aruba, many plant and animal species are extinct (Hall, 2010). Especially the negative effect of tourism on the biodiversity of Aruba is significant (Hall, 2010). Therefore it is important to keep the areas of nature reserves as large as possible to reduce further biodiversity losses (Schmitz et al., 2015). This stresses the need for land use planning by the local authority to keep the biodiversity as high as possible. But the allocation of terrains as nature reserves is a challenge in the Caribbean islands. Commonly, not all proposed areas as nature reserves are realised (Alders, 2015). Aruba is a good example to investigate if the areas of nature reserves have been realised because it is a small island located in the Caribbean. I will investigate if the nature reserves were successfully

realised in Aruba. Besides, in a case of unsuccessful realisation, I will assess what types of land use have been realised instead.

Aruba is an attractive place for tourism, an important part of Aruba’s economy. Resulting in policy decisions supporting tourism and economic growth, causing resource depletion and loss of biodiversity (Hall, 2010). Currently, about half of the island is used for residential areas, among which tourist areas, and a small part of business destined areas (Overheid Aruba, 2020). The other half of the land exists of natural areas such as nature reserves, mangroves and salt sprays. Aruba has four nature reserves, the two most important ones are Parke Nacional Arikok and Salt Spray Park (Overheid Aruba, 2009). These parks are both located on the northeast side of the island and have an area of 35 km2_{and 32 km}2_{, the total area of nature reserves in Aruba is 69.7 km}2_{(Overheid Aruba, 2009).}

The government of Spatial Development, Infrastructure and Environment of Aruba is

responsible for the land use planning (Overheid Aruba, 2020). Every few years a Spatial Development Plan (ROP) with the policy about the land uses in Aruba is developed by this governance. The first ROP was published in 2009 and the second in 2019 (Figure 1). This ROP contains a map with the allocation of land uses such as nature reserves and residential areas. The goal of the ROP is among others keep and strengthen the natural values of these parks. On the south of the island reef islands and lagoons are located, they have an important ecologic function such as breeding ground for birds and fish (Overheid Aruba, 2009). Therefore these areas need to be preserved according to the ROP of 2009. Lastly, the preservation of the Spanish Lagoon, which is partially filled with fresh and partially with salt water, is addressed in the ROP.

(5)

4 Figure 1, This is the ROP of 2019 and shows the designated land uses of Aruba. The nature reserves are pictured in green colours. Build-up areas are coloured orange and economic zones are brown. The yellow coloured areas are rural (Overheid Aruba. 2019).

In the ROP of 2019 it was stated that the plans of 2009 were not accomplished as planned (Overheid Aruba, 2019). The cause of this is uncertain, it might be a result of patronage due to the colonised history of Aruba (Alders, 2015). Patronage is the permanent informal subordination of a country that has been a colony to the sovereign state. Aruba has been decolonized since 1954 but still experiences the negative impacts of its history in the form of a weaker developed government and the informal subordination of Aruba to the Netherlands. Several countries that have been colonized experience the same governmental shortcomings as Aruba does. These shortages might harm the implementation of the ROP and therefore the realisation of nature reserves.

This research will show the developments in the land use of Aruba from 2008 to 2020. The aim is to create a complete overview of the state on the development of the natural landscape of Aruba. Therefore it is important to create an impression of the current state of the natural landscape of Aruba and the extent to which the ROP of 2009 has been implemented. How much of the proposed natural areas are realised and did they deviate from the prescribed plans? If they deviated, what was the number of square kilometres of natural reserves that has been failed to realise? Since the first published ROPs was created in 2009 and the second in 2019, the timespan for the research will be from 2008 to 2020. These ROPs will be compared with satellite images of Aruba from 2008, 2012 and 2020. The starting date 2008 has been chosen because these images are taken just before the first ROP was published and the implementation has begun. The images will be compared to each other to give results about the implementation of the ROP and the land use changes in Aruba.

Therefore the research question is: How much area that was proposed by governmental bodies

is actually realised as nature reserves in Aruba between 2008-2020?

In the research, the Land use Land Change (LULC) of the three maps will be analysed with ArcGIS and compared to the ROP map of 2009. The results will show what parts of the ROP have been implemented and what has not been realised. The total area of nature reserves before the land use

(6)

Earth Sciences

5 planning and after are being calculated. If these areas are compared a quantitative result concerning the realisation of nature reserves in Aruba can be given. Next to this the classified map of 2020 will be compared to the ROP of 2009 to further compare the implementation of nature reserves. Besides, the land use of areas that are not nature reserves (build-up, rural area, sand and water) will be classified. If parts of the nature reserves have not been realised their current land use will be classified. With these results, the accomplishment of the ROP of 2009 can be examined.

The relevance of this research is to investigate to what extend the ROPs of Aruba have been realised since it was already mentioned in the ROP of 2019 that the ROP of 2009 was not fully implemented. Besides, a complete overview of the state of the development of the natural landscape will be given. This will be useful when developing a new ROP because the most recent land use chances are visualised. If the areas of nature reserves are not large enough subsequent ROPs can focus on this issue and create more space for nature reserves.

(7)

6

Methods

The ROPs of Aruba were created for the entire island, therefore the study area of the research was the entire island of Aruba. The Land Use Land Change (LULC) analysis identified the land uses in Aruba at a certain year. This led to classified maps which can be used to assess the extent to which the ROP has been implemented. Besides, these maps can be compared to analyse the changes in land use between different years.

The LULC was analysed in ArcGIS, a Geographic Information System program. Within ArcGIS, satellite images can be classified and geographic maps can be created. The ROPs were created in ArcGIS, these files are available in PDF format on the Dutch governmental website of Aruba. These files can be found under the nature and environment tab where the information about spatial planning, including the ROPs, is available. Next up, cloud-free TM satellite images of Aruba from 2008, 2012 and 2020 were acquired. The reflective bands with 30x30 meter grid cells of the images were used for the LULC analysis (Liu, 2020). The satellite photography program used was Landsat, the three satellite images needed to be taken with the same satellite sensor. Landsat 7 sensor has made images from 1999 and therefore contains data from 2008, 2012 and 2020. Consequently, raw Landsat 7 Tier 1 Collection 1 images were used for the LULC analysis (Google Earth Engine, 2021). A margin of one or two years for the satellite images was necessary to receive the best quality material with the least amount of clouds. Through the script in Google Earth Engine the images with the least amount of cloud coverage between 2007 and 31-12-2008, 2011 and 31-12-2012, 01-01-2019 and 31-12-2021 were generated. The scripts used for retrieving the satellite images used in this research can be found in the appendix (Appendix 1). These steps led to three cloud-free images (Appendix 2).

Before the classification of the images the satellite data needed to be pre-processed. The TM scenes were geometric corrected using the Project Raster tool. This tool projected the satellite images in the correct UTM zone, for Aruba this is zone 19N with EPSG number 32619 (EPSG, 2021). This tool converted the pixels into square meters, this is necessary to quantify the land use changes in the final step.

The Classification Wizard Tool was used to train the samples for the classification (Remote Sensing, 2019). The classified land uses were build-up, rural area, water, vegetation and sand. Green coloured areas indicate areas with plants, trees and bushes and were classified as vegetation. Rural area differs from vegetation because these areas are mostly covered with sand and rocks and less dense vegetated, these fields consist of brown shades. With the Training Sample Manager, Training Data was created. A new schema with the classes build-up, rural area, water, vegetation and sand was created. Within these classes, training samples were sketched free-handed. When the satellite image of 2008 was selected the Classification Wizard Tool could be started. For this classification, the method was supervised and the classification was performed pixel-based. Next up the classification schema and training samples were used to run the classification. These steps were repeated for the images of 2012 and 2020.

This method has led to three classified maps. To check the accuracy of the classification a new set of training data was created the same way as the first set was generated. The Accuracy Assessment tool was started, as reference set the second set of Training Data was used, the sampling size was 500 and the sampling strategy Stratified Random. The outcome of the accuracy assessments was three Confusion Matrixes with the accuracy of the classes and the complete classified images (Rwanga & Ndambuki, 2017). The accuracy is expressed in a kappa coefficient whose minimum score was 0.8. If the kappa coefficients of the three classified maps were high enough the maps were finished.

Lastly, the land use changes are quantified. This was done with the use of the Zonal Statistics as Table Tool (Remote Sensing, 2019). This tool generated the land use per class in square meters per class. These tables were converted to Excel using the Table to Excel tool. In Excel, the areas of the classes were converted to square kilometre to make the numbers more comprehensive. With these areas, the exact amount of land change in Aruba in square kilometres could be calculated in squared kilometres and percentages. In the next paragraph, the results of LULC of Aruba are presented and compared to the ROP of 2009.

(8)

Earth Sciences

7

Results

In this section the results of the LULC of Aruba are visualised and explained. The LULC in ArcGIS of the three satellite images from 2008, 2012 and 2020 led to three classified maps (Appendix 3). These maps show the land use of Aruba in the given year. The goal of this research was to investigate to which extend the ROP of 2009 has been executed. The ROP of 2009 (Figure 2) was compared to the classified image of 2020 to analyse the extent to which the ROP has been executed. The allocation of the land uses vegetation, rural area, build-up, sand and water in the ROP will be compared to the allocation of them in the classified maps, deviations from the ROP are illustrated. Next up the LULC between the three classified maps of Aruba will be analysed using quantitative numbers for the land use change.

Figure 2, ROP of 2009. This ROP is compared to the classified map of 2020 to investigate to which extend the prescribed land uses are implemented. The north-east side of the island exists mostly of natural areas, pictured in the green and yellow colours. The south-west of the island is designated for living and economic zones (Overheid Aruba. 2009).

The three satellite images were classified in ArcGIS and the accuracy of the classifications was tested with use of the Accuracy Assessment Tool. Per map a Confusion Matrix was created with information about the accuracy of the classification per land use. The accuracy of the classified map of 2008 was 97%, this means that 97% of the pixels were class as the class they were identified as during the classification (Table 1). The accuracy of the second map, 2012, was 96% (Table 2). The land use allocations of the 2020 map was 97% accurate (Table 3).

(9)

8

Table 1, Confusion Matrix of the classfied map of 2008. This table shows the number of pixels which were classified as the land use of which they were intented to be classified as. The Total row shows the quantity of points that should have been identified as the given class according to the reference data (esri, 2021). The Total column shows the number of points that were identified as the given class according to the classified map. The kappa coefficient gives and indication of the overal accuracy of the classified maps.

Table 2, Confusion Matrix of the classified map of 2012.

Class Water Vegetation Rural Build-up Sand Total U_Accuracy Kappa

Water 498 0 0 0 0 498 1 0 Vegetation 0 10 1 0 0 11 0.91 0 Rural 0 0 8 0 0 8 1 0 Build-up 0 0 0 10 0 10 1 0 Sand 0 0 1 0 10 11 0.91 0 Total 498 10 10 10 10 538 0 0 P_Accuracy 1 1 0.8 1 1 0 1 0 Kappa 0 0 0 0 0 0 0 0.97

Table 3, Confusion Matrix of the classified map of 2020.

Water 498 0 0 0 0 498 1 0 Vegetation 0 10 2 0 0 12 0.83 0 Rural 0 0 7 0 0 7 1 0 Build-up 0 0 1 10 0 11 0.91 0 Sand 0 0 0 0 10 10 1 0 Total 498 10 10 10 10 538 0 0 P_Accuracy 1 1 0.7 1 1 0 0.99 0 Kappa 0 0 0 0 0 0 0 0.96

Next up the area of the classes per classified map in square kilometres were computed by exporting the created tables with the Zonal Statistics as Table Tool, of the classified maps to Excel (Table 4).

Table 4, This table shows the area in squared kilometres per land use of the classified maps per year.

Class Area (km2_{) 2008} _{Area (km}2_{) 2012} _{Area (km}2_{) 2020}

Water 1740.86 1740.11 1743.15

Vegetation 79.37 78.19 72.70

Rural 36.07 32.39 37.06

Build-up 58.83 63.51 64.50

Sand 7.57 8.52 5.30

Water 498 0 0 0 0 498 1 0 Vegetation 0 9 0 0 0 9 1 0 Rural 0 0 8 0 0 8 1 0 Build-up 0 0 0 10 0 10 1 0 Sand 0 1 2 0 10 13 0.77 0 Total 498 10 10 10 10 538 0 0 P_Accuracy 1 0.9 0.8 1 1 0 0.99 0 Kappa 0 0 0 0 0 0 0 0.96

(10)

Earth Sciences

9 The total area of vegetation on the classified map of 2020 is 72.7 km2_{(Table 6). This number}

contains all the densely vegetated areas including city parks, grass fields, agricultural fields, trees et cetera. This means that the actual area of nature reserves cannot be exactly derived from this

classification. The combined area of the nature reserves in Aruba is 67.9 km2_{, according to the ROP of}

2009 (Overheid Aruba, 2009).

In the Spatial Development Plan of 2009 the rural areas are characterised by agricultural features such as hedges, farms, agricultural fields and small brick stone walls (Overheid Aruba, 2009). According to the ROP of 2009 rural areas are located between the densely build-up area in the south-west of the island and the natural reserves in the north-east, this area is 27.7 km2_{(Overheid Aruba,}

2009). The classification of the 2020 map of Aruba shows that the rural area in 2020 is 37.06 km2

(Table 4).

Comparing the ROP of 2009 and the classified map of 2020 it can be seen that the area of nature reserves in the centre of the island are for the most part realised (Figure 3). Within the boundaries of the nature reserves there is no build-up visible.

Figure 3, This map is an magnification of the classified map from 2020 with an overlay of proposed area of nature reserves in the ROP of 2019. The light green area outlined with red is the proposed area of nature reserves. The darker green areas are classified as vegetation in 2020.

(11)

10 But on the ROP map of 2009 it is visible that the north-west top of the island is dedicated to be natural reserves, this part is coloured light green (Figure 4). But on the classified map this part exists mostly out of sand and rural area, there is barely any vegetation visible in this area (Figure 5).

Figure 4, This map visualises an magnification of the classified map of 2020 with an overlay the proposed area of nature reserves in the ROP of 2009. The light green area outlined with red is the proposed area of nature reserves and the underlaying map is the classified map of 2020.

(12)

Earth Sciences

11

Figure 5, this is an magnification the north-west end of Aruba from the classified map from 2020, this area was classified as mostly rural and sand.

Besides a large part of the nature reserves located in the middle of the island is classified as rural area instead of vegetation (Figure 6). Similarly the less densely vegetated areas along the coast of the island are classified as rural areas instead of vegetation (Figure 6).

Figure 6, this figure pictures the middle of Aruba. The light green area outlined with red are the nature reserves as prescribed in the ROP of 2009. The underlaying map is the classified map of 2020 where along the coast rural areas are identified.

(13)

12 The LULC analysis between 2008, 2012 and 2020 showed a decrease in vegetation (Table 5). This is visible in the classified maps, many densely vegetated areas became rural areas over time (Figure 7 & 8). The area covered with vegetation in 2008 was 79.37 km2_{, in 2012 this lowered to}

78.19 km2_{(Table 5). The percentual change of area vegetation from 2008 to 2012 was 1.5% (Table 6).}

In 2020 the area of vegetation decreased further with 7% compared to 2012, in 2020 the area of vegetation in Aruba was 72.7 km2_{. From 2008 to 2020 the amount of vegetation decreased with 8.2%.} Table 5. In this table the Land Use Land Changes in squared kilometres between the different classes are pictured.

Table 6. In this table the percentual changes in land use between de areas of land use are visualised.

Percentual change Class 2008-2012 2012-2020 2008-2020

Water -0.04% 0.17% 0.13% Vegetation -1.48% -7.02% -8.40% Rural -10.21% 14.44% 2.75% Build-up 7.95% 1.56% 9.63% Sand 12.48% -37.80% -30.04% .

Besides the LULC shows that the area of build-up increased, areas that were covered with vegetation became build-up (Figure 9 & 10). The area on the beach made place for build-up, the seashore of Aruba decreased with 30% from 2008 to 2020 (Table 5) . The area of build-up in 2008 was 58.83 km2_{, in 2012 this already increased with 8% to 63.51 km}2_{(Table 4 & 5). But compared to}

2012 in 2020 the area only increased with 1.6% to 64.5 km2_{. The total increase of build-up from 2008}

to 2020 was 9.6%. Difference km2 Class 2008-2012 2012-2020 2008-2020 Water -0.76 3.04 2.28 Vegetation -1.18 -5.49 -6.67 Rural -3.68 4.68 0.99 Build-up 4.68 0.99 5.67 Sand 0.94 -3.22 -2.27

Figure 7, this image visualises an magnification of one of the nature reserves of Aruba in 2008.

Figure 8, this images magnifies the same area as figure 7 but in 2020. It is visible that the area vegetation has decreased and the rural area has increased.

(14)

Earth Sciences

13 The land use types that changed the most were vegetation, build-up and sand. The results will be discussed further in the next paragraph.

Figure 9, this image shows an area of the build-up in Aruba in 2008. Figure 10, in this image from 2020 it is visible that the build-up area has increased compared to 2008. Especially near the coast and around the vegetated area.

(15)

14

Discussion

The results generated with this research are to a certain extend helpful when investigating the implementation of the ROP of 2009. The area of vegetation on the classified map of 2020 was 72.6 km2_{, this is higher than the area given in the ROP of 2009 which was 67.9 km}2_{. But the classified map}

included every form of vegetation on the island and not only the nature reserves. Therefore the research question: ‘How much area that was proposed by governmental bodies is realised as a nature reserves in Aruba between 2008-2020?’, cannot be answered with exact numbers. What can be stated is that the area of vegetation from 2008 to 2020 decreased with 8.2% to 72.6 km2_{. This number will be}

close to the area of nature reserves which is 67.9 km2_{. The areas destinated for nature reserves in the}

classified maps mostly consist of rural and vegetated areas. There are barely any build-up areas visible in the classified maps. Therefore it can be concluded that the area of nature reserves by the

government of Aruba are for the most part realised. Due to the high accuracy score, it could be stated that the classification was done correctly. Nevertheless, comparing the results of the LULC to the ROP is difficult and has therefore not led to the aimed exact quantitative results.

The rural area generated during the classification of the satellite images was larger than the 27.7 km2_{stated in the ROP of 2009, the LULC showed an rural area of 37.06 km}2_{. In the classified}

map of 2020, the rural areas are classified more precisely than in the ROP because the classification in ArcGIS was pixel-based. The class ‘rural’ in the LULC was described as less densely vegetated areas. The areas that are classified as rural area in the ROP of 2009 were described as predominantly natural areas with a few buildings, located between the dense build-up area and the nature reserves (Overheid Aruba, 2009). This makes it difficult to compare rural area between the classified map and the ROP because the classification in GIS was pixel-based and the classification of land uses in the ROP was rougher with larger areas. Along the coast of the island rural areas are classified when these were classified as nature in the ROP. Due to the difference of rural areas in the ROP and LULC, less densely vegetated areas are classified as rural areas but they could still be a part of the nature reserves.

One of the explanations for the increase in rural areas could be climate change. Researches state that as an effect of climate change the yearly precipitation of Aruba decreases which will lead to a decrease of vegetation (Bishop & Payne, 2012) (Cashman, Nurse & John, 2020) (Taylor,

Stephenson, Chen & Stephenson, 2012). Areas covered with vegetation will become less densely vegetated because there is not enough water available for the plants to grow. This could explain the decrease in vegetation and increase in rural areas over the past 12 years.

The LULC showed the area with build-up increased by 9.6%. This increase can be explained by the growing population and increase in tourism which results in the need for more buildings (Cole & Razak. 2009). Research has shown that an increase in build-up leads to a decrease in vegetation (Hula, 2010). This is in line with the results of the LULC where the area of vegetation decreased and the area build-up increased in the past 12 years.

Despite the increase in build-up, the areas of nature reserves in Aruba have still been conserved. This could be a result of the land planning in the ROP where the exact areas and

boundaries of land uses had been indicated. A research of Pierce et al. (2005) showed that awareness of the importance of biodiversity and nature preservation an important factor is when implementing spatial development plans. In the ROP the need for nature preservation was emphasised which could have led to the successful implementation of the nature reserves.

Other researches have shown that not only political but also economic factors are important when implementing spatial development plans (Sissoko, 2020) (Wang et al., 2020). When developing the ROP of 2009 the government of Aruba focussed on land use planning as well on economic growth and development (Overheid Aruba, 2009). This could be an important factor for accomplishing the allocation of nature reserves.

Nevertheless, there were two issues with the satellite images. First of all, the cloud coverage of the satellite images was chosen over the time of year. This means that the images are not taken at the same time each year. One might be taken in November and the other in May, which affects the amount of vegetation. The precipitation increases in November which results in the growth of vegetation and an increase in the amount of green (Stoffers, 1956). This results in more green pixels and therefore more area classified as vegetation. During the dry periods, these plants are less green and could be

(16)

Earth Sciences

15 classified as rural area instead of vegetation. But since the seasonal variations in Aruba are minor this should not have had a large impact on the results. Secondly, the colours of the image of 2008 appear to be more vivid. Especially the vegetated areas appear to be greener. Therefore the pixels which are coloured between green and brown could have been classified as vegetation in the 2008 map but as rural are in the 2012 and 2020 map. Yet there are no significant differences in the areas vegetation and rural between the classified 2008 map and the 2012 and 2020 maps. Besides, the accuracy of the maps was high enough to reject this uncertainty. To improve the obtained results, it would be better if the satellite images would be derived during the same time of year. This will eliminate the difference in the amount of vegetation due to the seasonal differences. Besides, it is important to make sure the satellite images have the same vividness to avoid inconsistency in different colours for the same land use.

If the classified maps are going to be compared to maps with several large areas with the same land use, such as the ROP, it is necessary to create a land use map with the same layout. This will make comparing the maps more accurate. This can be done after the classification by creating a polygon of areas that have the same land use. These polygons can be sketched free handed and their location will be estimated based on the classification.

Aquatic natural zones were not involved in this research. It is difficult to classify natural zones within the ocean and coast. Since the biodiversity losses in oceans are significant it is important to focus on this issue too. In further research, there can be focussed on these aquatic and maritime nature reserves. It was also beyond the scope of this research to investigate the capacity of the government of Aruba. But since the outcome of this research contradicts with the statements in the research of Alders (2015) about the problems with the government of Aruba and the effect of patronage it might be interesting to investigate how the land use planning in Aruba could have been a success.

(17)

16

Conclusion

The area of nature reserves in Aruba prescribed by the ROP of 2009 should be 67.9 km2_{. In this}

research the area densely covered with vegetation derived from the classified map of 2020 is 72.6 km2_.

Yet this area includes regions which are not part of the prescribed nature reserves such as agricultural fields. Moreover less densely covered areas were classified as rural during the LULC but were intended to be nature reserves according to the ROP. These rural areas are part of the nature reserves despite the fact they were not identified as vegetation. Besides the rural areas located between the build-up and nature reserves on the ROP are classified as vegetation, build-up and small parts of rural area. This happened because the classification in GIS was pixel based and the ROP identified larger areas of land uses. The vegetated area has decreased from 2008 to 2020, while the build-up and rural areas have increased. The decrease in vegetated area is a result of the increase in build-up, where vegetated areas needed to make place for buildings. The decrease in vegetation might also be a result of the decrease in precipitation in Aruba which leads to droughts, this could explain the increase in rural areas which area less dense vegetated. Lastly, the raw satellite images were not taken during the same time of year, this could have had an effect on the vegetated areas.

Nevertheless, it can be concluded that the area of nature reserves is implemented as prescribed in the ROP of 2009. The successful implementation could be a result of the stress for the need of nature preservation. Besides, the government of Aruba has focussed on economic growth which is an important factor for successfully implementing spatial development plans.

In further research, it might be helpful to divide the island into natural zones after the LULC classification. This will make comparing the classified maps to the proposed areas of land use in the ROP more accurate. Aquatic zones are important natural reserves but were not included in this research, further research on preserving aquatic zones is therefore important. Besides, it might be interesting to further investigate why some governments successfully implement spatial development plans and others are not.

(18)

Earth Sciences

17

References

Alders, A. A. (2015). Obstacles to 'Good Governance' in the Dutch Caribbean: Colonial-and postcolonial development in Aruba and St. Maarten (Master's thesis).

Aruba Native. (July 19, 2019). Despues di 10 aña por fin ROP 2019 ta efectua! Retrieved at 24-05-2021 from: https://arubanative.com/2019/07/19/despues-di-10-ana-por-fin-rop-2019-ta-efectua/ Bishop, M. L., & Payne, A. (2012). Climate change and the future of Caribbean development. The Journal of Development Studies, 48(10), 1536-1553.

Cashman, A., Nurse, L., & John, C. (2010). Climate change in the Caribbean: the water management implications. The Journal of Environment & Development, 19(1), 42-67.

Cole, S., & Razak, V. (2009). How far, and how fast? Population, culture, and carrying capacity in Aruba. Futures, 41(6), 414-425.

EPSG. (2021). WGS 84 / UTM zone 19N. Retrieved at 16-04-2021 from: https://epsg.io/32619 esri. (2021). Compute Confusion Matrix (Spatial Analyst). Retrieved at 26-05-2021 from:

https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-analyst/compute-confusion-matrix.htm Google Earth Engine. (2021). Google Earth Engine. Retrieved at 07-04-2021 from:

https://code.earthengine.google.com/?scriptPath=Examples:Datasets/LANDSAT_LE07_C01_T1 Google Earth Enginge. (2021). Landsat. Retrieved at 07-04-2021 from:

https://developers.google.com/earth-engine/datasets/catalog/landsat

Hall, C. M. (2010). An island biogeographical approach to island tourism and biodiversity: An exploratory study of the Caribbean and Pacific Islands. Asia Pacific Journal of Tourism Research, 15(3), 383-399.

Hula, M. A. (2010). Population dynamics and vegetation change in Benue State, Nigeria. Journal of Environmental Issues and Agriculture in Developing Countries, 2(1), 53-69.

Kabinet Wever-Croes. (2017). Regeerprogramma. Retrieved at 20-04-2021 from: https://www.ministeriejustitie.com/assets/docs/hunto_pa_aruba.pdf

Karimi, A., Yazdandad, H., & Fagerholm, N. (2020). Evaluating social perceptions of ecosystem services, biodiversity, and land management: Trade-offs, synergies and implications for landscape planning and management. Ecosystem Services, 45, 101188.

Klimaat aruba. (2021). Retrieved at 13-03-2021 from: https://klimaatinfo.nl/klimaat/aruba/

Liu, D., Chen, N., Zhang, X., Wang, C., & Du, W. (2020). Annual large-scale urban land mapping based on Landsat time series in Google Earth Engine and OpenStreetMap data: A case study in the middle Yangtze River basin. ISPRS Journal of Photogrammetry and Remote Sensing, 159, 337-351. Njue, N., Koech, E., Hitimana, J., & Sirmah, P. (2016). Influence of land use activities on riparian vegetation, soil and water quality: An indicator of biodiversity loss, South West Mau Forest, Kenya. Open Journal of Forestry, 6(05), 373.

(19)

18 Overheid Aruba. (2009). Ruimtelijk Ontwikkelingsplan Aruba Planbescrhijving 2009. Retrieved at 14-05-2021 from:

https://drive.google.com/file/d/1V1s6_RKNOvhZA57VzCi8B8mfFwKDLbP5/view

Overheid Aruba. (2018). Ruimtelijke ontwikkeling en gronduitgifte beleid Dienst Infrastructuur en Planning. Retrieved at 15-03-2021 from:

https://www.overheid.aw/document.php?m=46&fileid=52354&f=9ee3a69c12c37a8d679d21c8ce4733 1c&attachment=0&c=40450

Overheid Aruba. (2019). ROPV Aruba. Retrieved at 15-03-2021 from:

https://www.arcgis.com/apps/webappviewer/index.html?id=38891684b5404974a15860c00a0defd6 Overheid Aruba. (2019). ROP Toelichting met bijlages. Retrieved at 15-04-2021 from:

https://www.overheid.aw/informatie-dienstverlening/natuur-en-milieu-subthemas_47096/item/algemeen_37936.html

Overheid Aruba. (3-10-2020) ROPV. Retrieved at 15-04-2021 from https://dip-infra.com/wp-content/uploads/2020/03/cROPv-1.pdf

Pierce, S. M., Cowling, R. M., Knight, A. T., Lombard, A. T., Rouget, M., & Wolf, T. (2005). Systematic conservation planning products for land-use planning: interpretation for implementation. Biological Conservation, 125(4), 441-458.

Remote Sensing, Future Planet Studies. (2019). LULC_China_RS_Mod9_2019.pdf. retrieved at 15-03-2021 from:

https://canvas.uva.nl/courses/10084/pages/assignment-text-module-9?module_item_id=414672

Ridderstaat, J., Croes, R., & Nijkamp, P. (2014). Tourism and long‐run economic growth in Aruba. International Journal of Tourism Research, 16(5), 472-487.’

Rwanga, S. S., & Ndambuki, J. M. (2017). Accuracy assessment of land use/land cover classification using remote sensing and GIS. International Journal of Geosciences, 8(04), 611.

Sampford, C. (2002). Environmental governance for biodiversity. Environmental Science & Policy, 5(1), 79-90.

Schmitz, O. J., Lawler, J. J., Beier, P., Groves, C., Knight, G., Boyce, D. A., Bulluck, J., Johnston, K. M., Klein, M. L., Muller K., Pierce, J. D., Singleton, W. R., Strittholt, J. R., Theobald, D. M.,

Trombulak, S. C. & Trainor, A. (2015). Conserving biodiversity: practical guidance about climate change adaptation approaches in support of land-use planning. Natural Areas Journal, 35(1), 190-203. Sissoko, F. (2020). Land-use planning implementation uncertainty in Bamako District. African Journal on Land Policy and Geospatial Sciences, 3(1), 32-44.

Stoffers, A. L. (1956). The vegetation of the Netherlands Antilles (No. 135). Kemink.

Taylor, M. A., Stephenson, T. S., Chen, A. A., & Stephenson, K. A. (2012). Climate change and the Caribbean: Review and response. Caribbean studies, 169-200.

Van Nooren, K. (2008). Human Impact on the Vegetation of Aruba. In Second National Congress for the Preservation of Aruban Wildlife, Wild Aruba, 2, 25–26 August 2008, Oranjestad.

Wang, J., Wu, Q., Yan, S., Guo, G., & Peng, S. (2020). China’s local governments breaking the land use planning quota: A strategic interaction perspective. Land Use Policy, 92, 104434.

(20)

Earth Sciences

19

Appendices

Appendix 1

These appendices contain the scripts for downloading the satellite image from 2008, 2012 and 2020 of Aruba. For the classification Collection 1 Tier 1 raw Landsat 7 images were required. The code for downloading Landsat 7 data could already been found on the site of Google Earth Engine:

ee.ImageCollection("LANDSAT/LE07/C01/T1"). Next up the dates can be filtered, this code comes directly after the Landsat collection. Lastly the boundaries of the research area, Aruba, were created by creating a polygon, ‘geometry’. This polygon had the geometry points -70 and 13. Next up the

composite can be made cloud free by first creating a composite by setting the percentile to 100 and the cloud score range to 0. The composites are displayed by the drawn polygon ‘geometry’. The bands used for Landsat 7 are B3, B2 and B1. Lastly, the image can be exported to Google Drive, the name of the file can be changed with the description. The scale of the map is 30x30 metres grid cells and the part of the image that will be exported is the polygon ‘geometry’.

var collection = ee.ImageCollection("LANDSAT/LE07/C01/T1_RT") .filterDate('2007-01-01', '2008-12-31')

.filterBounds(ee.Geometry.Point(-70, 13));

var composite = ee.Algorithms.Landsat.simpleComposite(collection); var customComposite = ee.Algorithms.Landsat.simpleComposite({ collection: collection,

percentile: 100, cloudScoreRange: 0 });

Map.centerObject(geometry);

Map.addLayer(composite, {bands: ['B3', 'B2', 'B1'], max: 128}, 'Aruba'); Export.image.toDrive({ image: composite, description: 'Image2008', scale: 30, region: geometry });

(21)

20 Script satellite image 2012

Script satellite image 2020

(22)

Earth Sciences

21

Appendix 2

This is a Landsat 7 Collection 1 Tier 1 Raw satellite image from 2008 from Aruba. This image was derived from Google Earth engine with the first script in Appendix 1.

(23)

22 This is a Landsat 7 Collection 1 Tier 1 Raw satellite image from 2012 from Aruba, this image is derived from Google Earth Engine with the second script in Appendix 1.

(24)

Earth Sciences

23 This is a Landsat 7 Collection 1 Tier 1 Raw satellite image from 2020 from Aruba, this image is derived from Google Earth Engine with the third script in Appendix 1.

(25)

24

Appendix 3

This is the classified map from 2008, classified with ArcGIS. The classification was based on the satellite image of 2008 in Appendix 2 and trainings data created in ArcGIS.

(26)

Earth Sciences

25 Classified map of 2012 based on the satellite image of 2012 in Appendix 2.

(27)

26 Classified map of 2020 based on the satellite image of 2020 in Appendix 2.