USING GEOGRAPHIC INFORMATION SYSTEM (GIS) AS A TOOL TO MAP AND TYPE FINE SCALE WETLAND DATA

(1)

Master Thesis

submitted within the UNIGIS MSc programme at the Department of Geoinformatics - Z_GIS

University of Salzburg

USING GEOGRAPHIC

INFORMATION SYSTEM (GIS) AS A TOOL TO MAP AND TYPE FINE

SCALE WETLAND DATA

by

Miss Namhla Mbona

446938

A thesis submitted in partial fulfilment of the requirements of the degree of

Master of Science (Geographical Information Science & Systems) – MSc (GISc) Advisor:

Ann Olivier

Pretoria South Africa, February 2016

(2)

i

DECLARATION OF AUTHORSHIP

I, Namhla Mbona declare that the thesis/dissertation, which I hereby submit for the degree Master of Science in GeoInformatics, is my own work and has not previously been submitted by me for a degree at this or any other tertiary institution.

Signature: ..……….. Date: ……….

(5)

iv

FOREWORD

Publications which form part of and/or included in the research presented in this thesis, is a work done for Water Research Commission by Namhla Mbona as a first author with the guidance of the steering committee and collaboration with other authors as referenced below:

Mbona N, Job N, Smith J, Nel J, Holness S, Memani S, Dini J. 2015. Supporting better decision-making around coal mining in the Mpumalanga Highveld through the development of mapping tools and refinement of spatial data on wetlands. WRC Report No TT 614/14, Water Research Commission, Pretoria.

(6)

v

ABSTRACT

Wetlands in South Africa are increasingly coming under threat from different land use like agriculture; mining and urban development; and are rapidly disappearing. In response to the many threats to wetlands, South Africa has seen an increased interest in wetland research, which has introduced many methods to help standardize the approach to research, management and conservation of wetlands. To best manage and conserve wetlands require the development of countrywide and, related, provincial wetland-focused strategies that address planning and sustainability, as well as water quality and quantity issues. One of the important tools for wetland management is to create an inventory with the location, extent and attributes. Wetland attributes are important to understand the wetland type. To know where wetlands are, their diversity and what pressures they are under, will aid in setting appropriate policies, development of specific awareness and management guidance for wetland conservation.. The generated information will need to prioritise, assist in rehabilitation and protection of the most appropriate wetlands. As a result, Geographical Information System (GIS) is a powerful tool that can be used to map the location, extent and characteristics of wetlands. The main aim of this study was to determine if Geographical Information System (GIS) techniques could be used to map wetland location and size, and to explore the potential of GIS to apply classification system to assign ecosystem type. Wetlands were mapped more accurately at fine scale using a GIS method. The classification system was applied at accuracy levels of above 50% at this scale at a faster and uniform rate.

(Keywords: wetlands, GIS, mapping, classification system red edge position, Mpumalanga Highveld grassland, groundtruth

(7)

vi

ACKNOWLEDGEMENTS

I would like to thank the following parties for their contribution in this project work:-

 Water Research Council (WRC) and South Africa National Biodiversity Institute (SANBI) for funding;

 The members of the Reference Group of the WRC Project for their support, guidance and the constructive engagement throughout the duration of the project; Nancy Job as a project member on the WRC project

 I would also like to thank Siyabulela Memani and Anisha Dayaram of World Wildlife Fund-Mondi Wetlands Programme for desktop digitising assistance;

 Hannes Marais of Mpumalanga Tourism and Parks Agency (MTPA) for assistance in the field and contribution of data;

 Douglas MacFarlane of EcoPulse, Ronell Niemand of MTPA and Andre Beetge of Working for Wetlands for contribution of data;

 Vaughan Koopman of WWF-Mondi Wetlands Programme for assistance in the initial training;

 Ursula Franke of the Endangered Wildlife Trust, and all those who participated in the workshops;

 Mervyn Lotter of MTPA for guidance to the project team, particularly on data availability and mapping software;

 Bonani Madikizela of the Water Research Commission and John Dini for their guidance and patience;

 Faheima Daniels for assistance with creating the web map feature services.

 A special thanks to my family, colleagues for their patience, support and encouragements,

 And, not forgetting to mention Admire Moyo, Heidi van der Venter and Vuyokazi April from ZEN Environmental Consultants whose inputs and interactions assisted in shaping this thesis.

This project would not have been possible without the visionary approach adopted by the magistrate and prosecutor in the matter (case 462/07/2009, Ermelo Regional Court) between the State and a coal mining company charged with contravening the National Water Act and National Environmental Management Act after illegally mining within a wetland and diverting a river, among other activities. In return for agreeing to plead guilty to the charges, the mining company received a fine of R1 million, suspended for five years; agreed to rehabilitate the damage caused; and pay R1 million each to the Water Research Commission, Mpumalanga Department of Economic Development, Environment and Tourism, and the Mpumalanga Tourism and Parks Agency. At the time, this was the largest penalty ever imposed on a mining company in a criminal prosecution for environmental violations. In addition, the restitutive elements of the plea agreement, in the form of the requirement for rehabilitation of the affected ecosystems and payments to government agencies supporting water resource and environmental management of water, enabled these agencies to strengthen their pursuit of their statutory mandates. It is the payment from the mining company to the Water Research Commission that enabled the Commission to procure the project described in this report, with the explicit aim of strengthening decision-making relating to mining and water resource management in the Mpumalanga coalfields.

To God Almighty whose Mercies Endureth Forever, Thank You for Your Blessings. I Am Forever Grateful to You My Father.

(8)

vii

DEDICATION

This thesis is dedicated to my late Parents, may their souls continue to rest in peace. To my siblings who have been my pillar of strength, this work is to show you that the sky is not even the limit bantwana basekhaya, anything is possible!

(9)

viii

ACRONYMS

CDSM: Chief Directorate of Surveys and Mapping

CoCT City of Cape Town

CSIR: Council for Scientific and Industrial Research DEA: Department of Environmental Affairs

DEAT: Department of Environmental Affairs and Tourism DEM: Digital Elevation Model

DWAF: Department of Water Affairs and Forestry DWS Department of Water and Sanitation

EWF Endangered Wildlife Fund

GIS: Geographical Information Systems MTPA: Mpumalanga Tourism and Parks Agency NBSAP: National Biodiversity Strategy and Action Plan NEMA National Environmental Management Act

NEMBA National Environmental Management: Biodiversity Act NEMPAA National Environmental Management: Protected Areas Act NFEPA: National Freshwater Ecosystem Priority Areas Project NGI: National Geospatial Information

NSBA: National Spatial Biodiversity Assessment

NWA: National Water Act

NWM: National Wetland Map

SANBI: South African National Biodiversity Institute

WFW Working for Wetlands

WWF World Wildlife Fund

(10)

1

1. INTRODUCTION

Previously, wetlands were neglected ecosystems in terms of conservation and management (Noble & Hemens 1978; Heeg & Breen 1982; Begg 1986). As ecosystem, wetlands deliver a range of services to human society (Kingsford et al., 2009; Finlayzon & D'Cruz, 2005; Turpie et al., 2008). Until recently, wetlands are recognised for their social, economic and ecological functions. However, due to their negligence, many important wetlands have been lost and severely degraded. They have become some of the world’s most threatened ecosystems (Lemly et al., 2000; Finlayson C.M. et al., 2005), with an estimated 50% loss of wetlands being noted worldwide since the 1900s (Cowardin, 1982; Finlayson & Spiers, 1999).

Topographically, wetlands constitute a relatively small surface area within a region, and serve proportionally large role in ecosystem services provision. In the face of climate change where wetland systems are vulnerable to changes in quantity and quality of their water supply; it is expected that climate change will have a pronounced effects on wetlands through alterations in hydrological regimes (Sieben et al, 2008). Therefore, it makes strategic sense to manage and conserve this ecosystem. Intervention is needed and as a baseline a wetland map or wetland inventory need to be developed for consistent data on wetland conservation planning across entire region as this is currently lacking.

Creating the wetland inventory has its own challenges, temporal variability of wetlands, coupled with their spatial variability (often long and narrow), makes them difficult to detect using multispectral satellite imagery that is more readily available. Due to their seasonal occurrence wetlands are more easily studied at a site level than relying on technical tools that are used for their detection.

It has been recognised that the sustainable management of wetlands in Africa has been hindered by an absence of basic inventory data (i.e. information on wetland distribution, area extent, type and condition) (Taylor et al.

1995; Stevenson and Frazier 1999; Thieme et al. 2005). The importance of wetland inventory for supporting wetland management in southern Africa was clearly outlined by Taylor et al. (1995) when undertaking a review of wetland inventories: “The value of wetlands in Africa is beginning to be realised as countries struggle with natural resource management decisions and new pressures imposed by structural adjustment programmes and National Environment Action Plans. This realisation has led to the development of programmes to manage wetlands, all of which are held back by the need to know the extent and nature of national wetland resources – both to evaluate their worth and to plan their management priorities.”

1.1. DEFINITION OF A WETLAND

In many countries, the term ‘wetland’ is defined more restrictively than the Ramsar definition, usually with specific reference to the presence of saturated soils and/or hydrophytic vegetation. In South Africa, as a case in point, the definition of wetlands in the National Water Act (No 36 of 1998) a wetland is defined as, “land which is transitional between terrestrial and aquatic systems where the water table is usually at or near the surface, or the land is periodically covered with shallow water, and which land in normal circumstances supports or would support vegetation typically adapted to life in saturated soil.” This definition includes all naturally occurring wetlands and pans, but excludes rivers, lakes and artificial wetlands, except for the transition zone from the river/lake and the terrestrial ecosystem. The definition of wetlands as defined in the Convention on Wetlands of International Importance especially as Waterfowl Habitat (Ramsar Convention) has a broader concept. Article 1.1 of the Ramsar Convention includes the following definition (Cowan 1995): “areas of marsh, fern, peatland or water, whether natural or artificial, permanent or temporary, with water that is static or flowing, fresh, brackish or

(11)

2

salt, including areas or marine water the depth of which at low tide does not exceed six metres”. Rivers, lakes and artificial wetlands, as well as tidal zones, are therefore included in this definition.

The prolonged (permanent or periodic) presence of water, on the land surface or in the soil, is a fundamental feature of wetlands, even when narrowly defined on NWA. As such, despite being viewed as transitional systems, wetlands can be considered to be a type of aquatic ecosystem, where ‘aquatic’ implies relating to or consisting of or being in water. An aquatic ecosystem could thus be defined as “an ecosystem that is permanently or periodically inundated by flowing or standing water, or which has soils that are permanently or periodically saturated within 0.5 m of the soil surface” (after Ollis et al. 2013), and this is the definition that has been assumed for this research. Wetlands, whether defined broadly (as per the Ramsar Convention, for example) or more narrowly (to exclude rivers and permanent standing waterbodies, for example), are generally taken to exclude deep marine waters.

1.2. CHARACTERISTICS FOR WETLAND

A) Landscape position

This is a practical index for identifying those parts of the landscape where wetlands are likely to occur (Figure 1-1). Wetlands typically occur in topographically low places or depressions, the most common being “valley bottom”. However, wetlands may also occur on steep to mild slopes, where groundwater discharge or hillslope interflow is to the surface as seeps and on relatively flat “plains”, most commonly depression or pan wetlands.

Figure 1-1: Landscape position where wetlands are likely to occur. From Nel et al 2011 B) Hydrophytic Vegetation

Plants must be specially adapted for life under saturated or anaerobic conditions in order to grow in wetlands (Figure 1-2). Such species are referred to as "hydrophytic" plants, or "hydrophytes."

Developing a set of personal notes on the characteristics of wetland plants can assist in identifying them in the field. Some important aspects to look at include:

 Frequency the plant occurs in wetlands (plants that are always found (>99%) in wetland are called obligate, plants that are found in 67-99% are called facultative wetland plants, while those found in wetlands 34-66% of the time are simply called facultative; Reed, 1987);

 Root form (clump or tuft-forming, rhizomes, stolons or fibrous roots);

(12)

3

 Habitat (permanently wet, seasonally saturated, sand, highly organic soil, salty estuarine area, steep slope, rocky area etc.); and

 Distinguishing features (hairy ligule or stem, opposite or alternate leaves etc.).

For the purposes of this study, areas of relatively homogeneous vegetative composition were characterised by

"dominant" or “indicator” species. The vegetation of an area of relatively homogeneous community composition is considered to be hydrophytic if greater than 50 percent of the plant cover is comprised of hydrophytic plants.

Figure 1-2:: Wetland plant Juncus exsertus taken at Chrissiesmeer wetland in Mpumalanga.

C) Soil Wetness Indicator

Field examination of soil conditions is important to determine if wetland anaerobic conditions exist or existed (i.e.

low oxygen conditions due to soils being saturated for long duration). Due to anaerobic conditions, wetland soils exhibit certain characteristics, collectively known as "redoximorphic features," that can be observed in the field.

Redoximorphic features include: high organic content; accumulation of sulfidic material (rotten egg smell);

greenish‐ or bluish‐gray colour (gley formation); spots or blotches of different colour interspersed with the dominant matrix; colour (mottling); oxidized rhizospheres (root zones); and dark soil colours (low soil chroma).

For the purposes of delineation, soil colours are described both by common colour name (for example, “dark brown”) and by a numerical description of their hue, value, and chroma (for example, 10YR 2/2) as identified on a Munsell soil colour chart (Munsell Color, 1992).

(13)

4

a) b)

c)

Figure 1-3: Tools for assessing wetland soils in the field include a) an auger to dig soil plots, b) a Munsell soil colour chart and c) a datasheet. From DEPARTMENT OF WATER AFFAIRS AND FORESTRY 2005

D). Determining Soil Texture in the field

Soil texture is described by the relative amounts of clay, silt and sand in the soil (Figure 1-4).

(14)

5

Error! Reference source not found.Figure 1-4:: Soil texture triangle From DEPARTMENT OF WATER AFFAIRS AND FORESTRY 2005 .

One way to determine rapidly whether a soil is dominated by sand-, silt- or clay-sized particles is to do a simple experiment with a sample of the soil and a little water (see Table 1.2-1.1). Take a small sample of the soil, add a little bit of water to it, and knead the wet soil between your fingers and thumb. Then answer the following questions:

1. Can the soil be shaped into a ball?

o If the answer is NO, then the texture is sand (Error! Reference source not found.).

o If the answer is YES, then move to question 2.

2. Can the soil be shaped into a ball but not pressed into a ribbon?

o If the answer is YES, then the texture is loamy sand.

o If the soil can be pressed into a ribbon, note the length of the ribbon and take a small sub-sample of the soil. Then excessively wet the sub-sample in the palm of your hand and rub it with your forefinger of your other hand to determine the grittiness of the soil.

3. Can the soil be pressed into a ribbon no longer than 2.5cm in length?

o If the answer is YES, the possible textures are sandy loam, silt loam, silt, or loam.

o If the wet sub-sample feels gritty, the texture is sandy loam.

o If the wet sub-sample feels very smooth with some grittiness, the texture is silt loam.

o If the wet sub-sample feels smooth and silky, with no grittiness detectable to the fingers, the texture is silt.

o If the wet sub-sample does not feel gritty or smooth, the texture is loam.

4. Can the soil be pressed into a ribbon that is 2.5 to 5cm in length?

o If the answer is YES, the possible textures are sandy clay loam, silty clay loam, clay, or loam.

o If the wet sub-sample feels gritty, the texture is sandy clay loam.

(15)

6

o If the wet sub-sample feels very smooth with some grittiness, the texture is silty clay loam.

o If the wet sub-sample does not feel gritty or smooth, the texture is clay loam.

5. Can the soil be pressed into a ribbon that is >5 cm in length?

o If the answer is YES, the possible textures are sandy clay, silty clay, or clay.

o If the wet sub-sample feels gritty, the texture is sandy clay.

o If the wet sub-sample feels very smooth with some grittiness, the texture is silty clay.

o If the wet sub-sample does not feel gritty or smooth, the texture is clay (Error! Reference source not found.).

Table 1.2-1: Different soils have differing capacities to hold water or slow water movement.

SOIL TEXTURE

Peaty soils retain water.

Fabric peat (which typically has a high hydraulic conductivity) and amorphous peat (which typically has a

low hydraulic

conductivity).

Clay soils greatly slow water movement.

Water can move quickly

through sand.

(16)

7 E) Hydrology

Water must be present in order for wetlands to exist. However, it need not be present throughout the entire year.

Wetland hydrology is considered to be present when there is permanent or prolonged inundation or soil saturation, sufficient to influence the soils and plants. In many wetlands, not all parts may be equally saturated, some may be nearly always saturated (permanent), some saturated for a significant duration (seasonal) and some saturated for only a short period of the year (temporary) sufficient, however, for the formation of hydromorphic soils and the growth of wetland vegetation (Figure 1-5). These three zones may not be present in all wetlands, and in certain parts of South Africa, seasonal and ephemeral wetlands without any zones of permanent wetness, are common. The object of the delineation procedure is to identify the outer edge of the wetland, including the temporary zone. This outer edge marks the boundary between the wetland and adjacent terrestrial areas.

Figure 1-5: Wetness of a wetland can vary from year to year, and season to season (adapted from Nebraska Wetlands).

It is important to emphasise that mappers should be alert in the field for “dry”-looking areas that may, in fact, be wetland. This can be confirmed by investigating the vegetation and soils. In this case, hydrology is a means to identifying the presence of a wetland and to deepening awareness that wetlands comprise a wide range, from very wet, to only occasionally wet. Detailed level mapping projects may go further and describe more specifically the hydrology of a wetland.

1.3. ECOSYSTEM MANAGEMENT BACKGROUND AND CONTEXT

The National Freshwater Ecosystem Priority Areas Project (NFEPA) identifies a set of priority areas which together meet national biodiversity goals for freshwater ecosystems and provide a single, nationally consistent information source for incorporating freshwater ecosystem and biodiversity goals into planning and decision‐

making processes. Much funding and specialist time has already been invested by the Water Research Commission, Council for Scientific and Industrial Research (CSIR), Department of Water and Sanitation and the South African National Biodiversity Institute (SANBI) in the collaboration that generated the Atlas of Freshwater Ecosystem Priority Areas in South Africa, in support of sustainable development of water resources. The NFEPA project was a tremendous step forward in consolidating existing knowledge and generating new knowledge on the distribution, type and condition of freshwater ecosystems.

(17)

8

However, experience in using the maps has shown that there is room to improve the underlying data (particularly the wetland layers) used to identify the NFEPAs, and that this has implications for the confidence that can be attached to the information on ecosystem typing, condition and threat status generated by the project. In an area under great development pressure, such as the coal‐mining area of the Mpumalanga Highveld Grasslands, the accuracy of the spatial information is critically important. This research thus focuses on verifying and refining the current data layers of the extent, distribution and type of wetlands in the Mpumalanga Highveld coal belt.

The project is also an opportunity to develop standardised methods for ground‐truthing that can be applied in the province and in other provinces, for the long term goal of strengthening of the National Wetland Inventory of South Africa.

The project study area is extensively focused on the coal belt areas of Mpumalanga, an estimated two thirds of Mpumalanga province. The primary project goal is to deliver the highest confidence possible wetland spatial (boundary) layer, and wetland types systematically across the full study area. This means that the greatest attention will be paid to verifying the presence, spatial extent and type of all wetlands across the study area, rather than securing more in‐depth information about certain wetlands only.

It should be noted that the project outcomes will not replace the need for further input in assessing land use impacts and wetland edge verification during an environmental impact assessment or similar intensive site‐

specific assessments usually done by wetland specialists.

1.4. APPROACH IN USING GEOGRAPHICAL INFORMATION SYSTEM (GIS) AS A TOOL FOR WETLAND MAPPING AND TYPING

Geographical Information System and Remote Sensing are becoming an important tool for natural resource research and management. GIS can be used for spatial mapping of the location and extent of wetlands. Using Global Positioning Systems, (GPS) coordinates can be collected to map a point representing a wetland.

Wetlands can also be mapped by use of onscreen digitising of polygons using base imagery.

One of the advantages of using Geographical Information Systems (GIS) is the ability to automate classification of large data sets in a relatively short period of time. Processing of large data sets may provide the opportunity for more consistency of classification at this scale, as opposed to the creation of local wetland data sets and classification systems by various wetland specialists and merging these in a bottom-up approach to one data set.

The disadvantage of the using GIS, is that few national data sets are available that represent the extent of wetlands accurately at a scale of 1:50 000 (though this scale is not considered a national scale, many geospatial data sets allow for national data sets to be available in regular 1:50k-sheets for a country, for example); is consistently captured at a national scale; results can be easily generated with a great variety of results, difficult to motivate the best one; and though results can be easily generated in a short period of time, many errors may result (e.g. slivers) which may take a considerable time to “clean up”.

There is an array of different techniques used to map wetlands. The choice of technique is dependent on the scale at which the wetlands are being mapped. The most accurate technique to map wetlands is the field survey technique, where the researcher literally walks the perimeter of the wetland with a GPS or scaled field map, plotting numerous points along the walk and then logging them into a geographic information system. This traditional technique is costly and time consuming, and is not feasible for mapping wetlands over large landscape areas. The introduction of aerial photographs and launching of satellites have resulted in rapid advancements in wetland mapping techniques over the past century. Wetlands exhibit distinct light-reflectance characteristics in

(18)

9

the visible or infrared portions of the electromagnetic spectrum. Wetland soils and water have distinctive reflectance characteristics that can be used to identify the presence of a wetland (Lyon, 1993: 72).

These characteristics are found in both aerial photographs and satellite imagery. With satellite imagery, large spatial domains can be rapidly mapped through various classification methods; therefore, the delineation of wetlands from satellite imagery is an internationally accepted method of creating regional inventories of wetlands (Thompson, 1994).

1.4.1. Wetland Mapping

Some mapping methods used GIS landscape modeling techniques, with a Digital Elevation Model (DEM). In using DEM, four layers are produced namely (i) occurrence of sinks or depressions,(ii) slope steepness, (iii) flow accumulation, and (iv)relative slope position or topographic index (TPI). The generated layers are then further reclassified on a scale of 1 to 5 (1-5) ranking from better to poor water retaining capacity. The reclassification was based on researching on the ability of water to hold when flowing fast or slow (i.e. 5 was given on average flow rather than high flow since modelling for wetlands or wetness avoids run-off`s). In addition, the highest value is divided by 5 to find threshold values (Thompson et al. (2002). The threshold value value 5 means good and 1 is considered bad based on water research. The four layers were combined into one integrated model through the use of a weighted overlay function. Several configurations of input parameters, reclassifications, evaluation scales, weightings and influence factors were tested before suitable model parameters were found

A map of South Africa wetlands has been produced under the National Wetland Inventory project. The project has followed different ways including remote sensing, field and desktop approach, using Geographical Information Systems (GIS), to map and classify wetlands according to ecosystem types (hereafter “wetland types”) that were used as coarse filter surrogates of wetland biodiversity across South Africa (sensu Groves, 2003). The layer was derived from the National Land Cover 2000 GIS layer (Thompson et al. (2002) Van den Berg et al. 2008), in which wetland polygons are described as ‘Wetland’ or ‘Waterbody’. The waterbody category does not distinguish between natural or artificial waterbodies. To overcome this problem, National Wetlands Map 1 was combined with the 1:50 000 inland water features (DLA-CDSM 2006), to derive National Wetland Map 2 that was divided into three GIS layers: (i) wetland, (ii) natural waterbody and (iii) artificial waterbody. To derive National Wetland Map 3, the wetland and natural waterbody GIS layers were combined to produce a natural waterbody GIS layer. This was then combined with the artificial waterbody GIS layer in which wetland polygons have been described as either ‘natural’ or ‘artificial’ waterbodies. Finally, existing sub-national wetland locality maps from other biodiversity planning initiatives were added to the National Wetland Map 3 to derive the final National Wetland Map 4 (see Figure 1-6; Nel et al 2011).

(19)

10

Figure 1-6: Flow diagram of the input data used to derive the National wetland Map 4.

While the accuracy of this GIS layer varies within regions of South Africa (e.g. lower success in densely vegetation regions), it represents an enormous advance for the country, resulting in the detection and mapping of approximately 3 million hectares (ha) of wetlands (<2.5% of the surface area of RSA), including natural, artificial and estuarine wetlands (SANBI, 2010).

(20)

11 Figure 1-7: SA National wetland map. From Nel et al 2011

This data was also used for South Africa’s first national freshwater conservation plan which entailed the systematic conservation planning for rivers, wetlands and estuaries (see Figure 1-7; Nel et al., 2011) National Freshwater Priority Areas (NFEPA) project.

1.4.2. Wetland Typing

Wetlands can be classified according to different classification methods, based on the hydrological functioning, fauna and / or flora composition or diversity, soil, or a combination of these features. The classification method used mostly depends on the reason for classification. Botanists would mostly use vegetation composition to classify wetlands, whereas an ornithologist will mostly use bird habitat features as a defining criterion. Ecologists are, however, more interested in the interaction of different biotic and abiotic features within a system and will therefore use the complete system functioning as part of the classification (Ward & Lambie 1999, Allan et al 1995, Cowardin 1982). In addition, there are also various classification methods using remote sensing, depending on the percentage open water, type of vegetation cover and density of the vegetation cover (Dely et al 1999, Shanmugam et al in press). One of the most popular methods for the classification of wetlands is the fundamental hydrological functioning of the wetland. The wetlands are therefore classified into different hydro- geomorphic units. In recent years, the development and use of classification systems based on the so-called hydrogeomorphic (HGM) approach to classification (after Brinson 1993) has become widespread in South Africa, particularly for inland wetlands (defined in the “narrow” sense of the word). This classification is a hierarchical classification system based on the hydrological characteristics to determine the ecological character and functions of the wetlands

(21)

12

Palmer et al (2002) classified the wetlands into hydro-geomorphic units during the previous wetland inventory.

Six broad hydro-geomorphic units were identified. The six units are Non-flood-plain riparian, floodplain riparian, hillslope seepage, pans, and other non-riparian and artificial wetlands.

The classification used by Kotze et al (2005) in WET-EcoServices is a classification using hydro-geomorphic units (Figure 1-8). This is a useful classification method for assessment of the functions and services of wetlands, since specific functions and services can be associated with each unit type, depending on site specific conditions. The level 1 assessment of Wet-EcoServices is a desktop study focussed on the hydro-geomorphic type of the wetland. The hydro-geomorphic types specified in Wet-EcoServices include floodplain, valley bottom with channel, valley bottom without channel, hillslope seepage feeding a water course, hillslope seepage not feeding a water course and depressions (pans).

Figure 1-8: The hydro-geomorphic wetland types according to Wet-EcoService guideline developed by Kotze et al., 2005.

The national wetland classification system (Ollis et al 2013) was used to classify wetland ecosystem types on the National wetland map 4 (Figure 1-9). It is a hierarchical classification framework consisting of six levels, with each level requiring increasing levels of detail about the wetland. Level 1 separates wetlands into inland, marine and estuarine systems. Levels 2 to 4 identify broad groups of wetlands sharing similar regional context, landform and broad hydrology. Levels 5 and 6 describe site characteristics such as hydro period, geology, vegetation, substratum, salinity, pH and naturalness.

This research will use the HGM approach following the published system (Ollis et al2013).

(22)

13

Figure 1-9:The hydro-geomorphic wetland types according the National wetland classification system (Ollis etal 2013).

For wetland typing, landform elements using researched relative classification model built on surface shape and altitude position, documented by Biasi (2001). The classes are ridge tops, valleys, mid-slopes, foot-slopes, upper-slopes, and flat surfaces.

1.5. GLOBAL, NATIONAL AND REGIONAL IMPLEMENTATION OF WETLAND MAPPING AND TYPING

National Landcover database and Wetland probability map (Thompson et al. (2002) Van den Berg et al.

2008.): The wetland class in the National Landcover was extracted as a separate layer to serve as wetland map Beta Version. It was released in 2006 for public use. The scale of the wetland cover was approximately 1:250 000. The layer has undergone a lot of improvements over time. Currently, the South African national wetland inventory, National Wetland Map v.3(NWM3), is a country-wide GIS layer that was constructed using a combination of remotely-sensed aerial photography, orthophoto and Landsat data of waterbodies (Van den Berg et al., 2008; SANBI, 2008 and 2010), combined with GIS data from desktop and field mapping at sub-national scales (Nel et al., 2011)

(23)

14

Endorheic pans (Allan et al 1995): A synthesis of all literature on the pans in the Transvaal and Free State was compiled by Allan et al (1995). The report included the definition, distribution, density and the size of pans.

The endorheic pans distribution map is on a very small scale (approximately 1:9 000 000), and is therefore, difficult to identify the locations of pans. The classification of pans and the difficulties experienced in classifying pans are discussed, as well as, the possible origin and maintenance of the pans. The report includes a section on the abiotic and biotic characteristics of pans, as well as, a section on threats to the pans (Allan et al 1995).

Numerous pans occur in the study area, but account for only a small percentage of the total wetland area (5%).

The non-permanent pans are more numerous than the permanent pans, but account for a smaller total area (less than 4%). The permanent pans in the Olifants catchment are larger than the non-permanent pans

Wetland inventory (Palmer et al 2002): A wetland inventory conducted by Palmer et al (2002) mostly focused on Upper Olifants river catchment including the area up to the confluence of the Olifants; Klein Olifants Rivers and Wilge River catchment. The most extensive wetland types in the catchment are the riparian wetlands. Since riparian wetlands are associated with rivers and streams, the total length of the rivers in the catchment is the most important contributing factor towards making the riparian wetlands the most extensive wetland type. The seasonally inundated channelled valley bottom floodplains with footslope seepage wetlands are the wetland types with the largest individual wetlands (Palmer et al 2002).

Although wetlands in this type are few, they account for 18% of the wetland area indicated in Palmer et al (2002).

Artificial wetlands (dams and weirs) are the most common wetlands in the study area, accounting for almost 50%

of the wetlands recorded in Palmer et al (2002).

Overberg Municipal District (Snaddon and Day, 2009) and the City of Cape Town (CoCT) Metropolitan Municipality (Job, 2010): both located in the Western Cape Province of South Africa. Both of these studies captured natural wetlands from 1:10 000 colour orthophotographic photographs and the mosaic of the SPOT 5 satellite imagery (SANSA, 2011). These wetlands were subsequently desktop classified according to Level 4A of SANBI (2009) by rapid visual interpretation of wetland types.

A wetland inventory was compiled by the Mpumalanga Parks Board: All the wetlands in the Mpumalanga province, floodplain, seepage and pans were distinguished in this layer. Wetland database of palustrine wetlands mapped for the Witbank Area, Usutu Catchment, Blyde River Catchment Reserve, Ohrigstad River, Nkomati River Catchment and Verloren Valei Nature Reserve. Additional wetlands and pans digitised and classified accordingly. Within the Usutu catchment additional wetlands were mapped.

1.6. MOTIVATION

The information collected through wetland inventory is considered to be necessary for effective wetland conservation and wise use especially as the pressure on wetlands increases from, for example, expanding agriculture and water regulation (Dugan 1990; Finlayson & van der Valk 1995a; Finlayson et al. 2005a).

Contracting Parties to the Ramsar Convention on Wetlands have undertaken to compile an inventory as part of the process of developing and implementing a national wetland policy for the wise use of all wetlands on their territory and to address key knowledge gaps. However, whilst there has been increased attention in Africa directed towards the threats at individual wetlands there are still large knowledge gaps about the location, extent and type of wetlands; extent of wetland loss; degradation; inter-related loss of wetland species and ecosystem services (Thieme et al. 2005).

(24)

15

This research will assessed the use of GIS in mapping (location and extent) and typing the wetlands in the Mpumalanga Highveld grassland region. GIS has been used to model a wetland classification system (Ollis et al 2013). Therefore, the empirical hypothesis of this research is to ascertain if GIS could be used as predictive tool in modelling and typing wetlands.

The research had the following aims:

1. To refine the current data layer of the location and extent of wetlands in the Mpumalanga Highveld coal belt

2. To assign wetland type using a GIS model

3. To collect and ground-truth data for some areas about the extent, location and type of wetlands ecosystems in the Mpumalanga Highveld coal belt; and

4. Produce a new database for wetlands to support informed and consistent decision-making by regulators in relation to the water-biodiversity-energy nexus.

1.7. STUDY AREA

The Mpumalanga Highveld comprises a wide, flat open area above the eastern escarpment It is a temperate region, with undulating hills dominated by grasslands and wetlands. Significantly, the Mpumalanga Highveld is a source of many of the country’s major rivers (Figure 1-10). The Vaal component of the vast Vaal-Orange system originates in the region and flows westwards across South Africa. The watersheds of the Komati, Pongola and Crocodile rivers drain eastwards. These water catchments provide much of South Africa’s water and are, thus, important to water security. The headwaters of these rivers are characterised by areas rich in wetlands. These wetlands are vitally important for a range of ecosystem services, including regulating services (such as streamflow regulation, water purification, and flood attenuation) that contribute to the sustainable functioning of the river basins, as well as provisioning (including grazing and water supply) and cultural services.

(25)

16

Figure 1-10: The study area of the project within the Mpumalanga Highveld region.

The unique grassland and wetland association within the area is also host to numerous threatened and conservation-worthy species and ecosystems. The Eastern Highveld Grassland vegetation type, which covers much of the study area, is classified as Vulnerable and has only minimal formal protection. The Wakkerstroom / Luneberg Grasslands and Chrissiesmeer Panveld are classified as Endangered. The grasslands are species rich, containing a number of endemic plants, such as the Barberton Daisy. They are famous as an exceptional birding site, the Mpumalanga grasslands host a number of endemic bird species, with specials including Rudd’s Lark, Botha’s Lark and the Yellow-breasted Pipit. The area also has one of the highest concentrations of Freshwater Ecosystem Priority Areas (FEPAs) in the country.

The Mpumalanga Highveld is also a mineral rich area and centre of large industry in South Africa. The vast coal reserves that underlie the grasslands support some of the largest coalfields in the world. Several large, international coal-mining companies and an increasing number of smaller independent mines operate extensively throughout the study area. Coal accounts for 70% of primary energy consumption, 93% of electricity generation and 30% of petroleum liquid fuels in South Africa (Eberhard, 2011). The coal is primarily used within South Africa for power generation, but the Mpumalanga coalfields contribute extensively to exports from the coal terminal at Richards Bay (GCIS, 2013). The majority of the country’s electricity originates from coal power, and as many as 11 operational coal-fired power stations are found in Mpumalanga. They are positioned there for close proximity to the coal reserves, but also for the year-round supply of water to satisfy the enormous demands of the power stations. Also located in the area are two coal liquefaction plants. Aboveground, much of the Mpumalanga Highveld grasslands are used for agriculture, particularly rangeland for sheep (wool) and cattle (dairy). Large tracts of commercial forestry are also found, which provide timber for a number of large pulp mills ( Mbona et al 2015) .

(26)

17

The area is, thus, a centre of water-biodiversity-energy nexus within the country. Each of these competing land uses are necessary priority within the context of socio-economic development. The trade-offs required in this region are, thus, more visible and contested than anywhere else in the country. Hence, focussing the project within this area will help to mitigate the land-use conflicts and encourage integrated development that accounts for the significant role played by biodiversity and ecosystem services in supporting sustainable development.

1.8. THESIS STRUCTURE

This thesis has five chapters inclusive of the introductory chapter which is Chapter One. Chapter Two sets out to review the existing literature on the key theory informing this research. The chapter begins with a section on the definition, importance and conservation of wetlands, and the associated legislation. Chapter Three explores the methods and approaches used to achieve the aims and objectives of this study. This chapter outlines the building of the wetland map for the study area. The use of GIS model to type (automate wetland classification system) wetlands is outlined and also discuss the tools used to collect field data for ground truth process.

Chapter Four outlines the results of the wetland layer for this area. The success of mapping and typing wetlands using GIS tools is discussed. The new layer is compared to the existing wetland layer from the national dataset.

Furthermore, the results obtained through visual assessment and accuracy assessment are assessed in this chapter. Chapter Five provides a discussion of the results obtained and critically examines the techniques used to obtain the layer. The accuracy of the output using this method and approach and the applicability of such an output is also discussed. Results obtained in this study will be related to previous research. Recommendations for future research are discussed.

(27)

18

2. LITERATURE REVIEW

Since this study focuses on the identification of wetlands, the literature review begins with an overview that covers the definition of wetlands, their formation, legislation and regulations pertaining to them their conservation and how to classify wetlands. The literature review revisits past research on wetland mapping approaches, both internationally and locally, and identifies certain limitations and gaps that this study aims to address, the idea of using GIS modelling for applying a wetland classification system.

 Introduction

Globally, the conservation of wetlands began in the mid-1970s. Prior to this, the destruction and drainage of these systems was common practice, and wetlands were seen as a nuisance, as health hazards, and as good potential agricultural and commercial development areas once drained (Mitsch and Gosselink, 2000). Wetlands were seen as flat open areas that would, once drained, provide land that could easily be modified for agriculture, and the perceived benefits included accessibility, good soils and the possibility of a water source for irrigation (Heimlich et al., 1998). These perceptions led to the destruction of wetlands and the loss of these valuable ecosystems. The disappearance of wetlands led to undesirable consequences, such as the loss of groundwater reserves and the consequent need for irrigation, flash floods, shoreline destruction, and the accumulation of pollutants (Ramsar, 1993). These undesirable consequences resulted in the loss of a great deal of wetland fauna and flora that relied on those systems.

 Importance of Wetlands

Wetlands provide important provisioning services (e.g. food, freshwater, fibre and fuel, biochemical products, genetic materials), regulating services (e.g. climate regulation, water regulation, water purification and waste treatment, erosion regulation, natural hazard regulation, pollination), cultural services (e.g. spiritual and inspirational, recreational, aesthetic, educational) and supporting services (e.g. soil formation and nutrient cycling) (MEA, 2005). It is beyond the scope of this thesis to provide an in-depth review of these goods and services, but they deserve some attention as they provide a strong foundation for why wetlands are important and why mapping these systems is necessary. The transboundary nature of wetland systems and the realisation of their importance resulted in the need for international agreements and arrangements to use wetlands wisely and preserve them. In 1971 this gave rise to the Convention on Wetlands of International Importance (more commonly known as the Ramsar Convention). The Ramsar Convention recognised the importance of wetlands as valuable ecosystems, and the importance of compiling national wetland inventories as a key tool for informing policies and other actions to achieve the conservation and wise use of wetlands (Ramsar, 2010).

 Definition of wetlands

Wetlands are areas that are seasonally or permanently waterlogged or saturated, and are characterised by vegetation that has adapted to survive in saturated soil conditions. Heimlich et al. (1998: 10) define a wetland as

“land that (1) has a predominance of hydric soils; (2) is inundated or saturated by surface or groundwater at a frequency and duration sufficient to support a prevalence of hydrophilic vegetation typically adapted for life in saturated soil conditions; and (3) under normal circumstances does support a prevalence of such vegetation.” A complicating factor in defining wetlands is that they occur in a wide variety of hydrologic conditions and also vary in geographical extent, making it difficult to delineate or spatially map and define them. Scientists and researchers realise the complex nature of wetlands and have even suggested treating wetland ecology as a

(28)

19

distinct field of study (Mitsch and Gosselink, 2000). The Ramsar Convention (Ramsar, 2006: 7) defined wetlands as “areas of marsh, fen, peat land or water, whether natural or artificial, permanent or temporary, with water that is static or flowing, fresh, brackish or salt, including areas of marine water the depth of which at low tide does not exceed six meters”. For the purposes of this study, wetlands will be defined as per the Ramsar Convention with slightly adjustment of removing fen as it is the same as peatland for SA and also the low tides are usually at ten meters (Ollis et al 2013)

2.1. LEGISLATION AND PROTECTION OF WETLANDS IN SOUTH AFRICA

As a consequence of a growing awareness for the importance of wetlands to society, and in response to their unchecked destruction, there has been considerable progress made over the last 20 years in the protection of wetland areas globally. The recognition of the value of wetlands has resulted in an emphasis on wetland protection in many laws and international agreements (Mitsch and Gosselink, 2000). The Ramsar Convention has been at the forefront of international intergovernmental cooperation on the protection of wetlands and/or wetland conservation. Developed and adopted in 1971 in the town of Ramsar, Iran, this global treaty provides the framework for the international protection of wetlands as habitats for migratory fauna that do not observe international borders and that benefit human populations dependent on wetlands (Mitsch and Gosselink, 2000).

The mission of the convention is the conservation and wise use of wetlands through national action and international cooperation, with sustainable development being the overarching goal. Subsequent to its initial objectives for protecting waterfowl, the aims have been broadened to include: preventing the loss of wetlands to preserve their fundamental ecological functions as well as their economic, cultural, recreational, scientific and educational value (Ramsar, 2011).

Countries that have ratified the Ramsar Convention are obliged to formulate and implement their planning to promote the wise use of all wetlands and develop national wetland policies. Member countries are required to designate at least one wetland as a Ramsar site and to establish nature reserves at these and other wetlands.

As of October 2012, 163 countries had joined the Ramsar Convention, and there were 2 059 Ramsar wetland sites, totalling an area of approximately 197 million hectares worldwide¹. South Africa has 20 Ramsar sites, totalling an area of approximately 553 000 ha.

Alongside the Ramsar Convention, many countries have formulated their own legislation and regulations relating to the protection of wetlands against degradation and destruction. “Other mechanisms for wetland protection include acquisition, planning, mitigation, disincentives for conversion of wetlands to other land uses, technical assistance, education, and research” (Votteler and Muir, 1996:57).

South Africa’s strong environmental legislative framework and its status as a contracting party of the Ramsar Convention provide the ideal platform to comply with numerous international conservation and environmental agreements, creating a legal framework for enabling the conservation of important natural assets, both terrestrial and aquatic (Table 2.1). In spite of all the laws related to wetland management, it is important to note that there is no specific national wetland legislation. South Africa’s Department of Water Affairs and Forestry (DWAF) outlines how the legislation applicable to wetlands and wetland rehabilitation arises from various other legal instruments currently in place (DWAF, 2007).

The Ramsar Convention covers all aspects of wetland conservation and wise use, recognising wetlands as ecosystems that are extremely important for biodiversity conservation and for the well-being of human communities. As a signatory to the Ramsar Convention, South Africa is obliged to comply with all aspects of the agreement.

1According to www.ramsar.org /cda/en/Ramsar-about-parties-parties/main (date accessed 09/10/2012)

(29)

20

Each contracting party of the Convention on Biological Diversity agrees to “rehabilitate and restore degraded ecosystems”, “regulate or manage biological resources important for the conservation of biological diversity” and

“help develop and maintain necessary legislation for the protection of threatened species or populations”. The Convention on Biological Diversity observes wetlands as important ecosystems, critical for human well-being, in terms of the goods and services they provide (SCBD, 2006).

Table 2.1: Conventions, legislation, policies and plans directly or indirectly related to the conservation and protection of wetlands in South Africa.

International Contribution

Ramsar Convention (1971) - Inter-governmental co-operation on the protection and conservation of wetlands.

- International cooperation in the wise use of wetlands.

- Strong legislative framework to build policies and

agreements in South Africa.

Convention on Biological Diversity (1994)

- Led to important South African legislation (National Spatial Biodiversity Assessment, National Biodiversity Strategy and Action Plan (2004), (Driver et al., 2005).

- Provides the platform for the rehabilitation of wetlands as

stated in its agreement.

South Africa

Constitution of the Republic of South Africa (Act No. 108 of 1996) (RSA, 1996)

- Creates a duty on the state to conserve and rehabilitate wetlands, because “everyone has the right to have the environment protected, for the benefit of present and future generations.”

National Environmental Management Act (Act No. 107 of 1998) (RSA, 1998b)

- Produced the Biodiversity Act (2004) (see below) and the National Environmental Management: Protected Areas Act (Act 57 of 2003).

- Important legislation leading to the conservation of

terrestrial biodiversity and then freshwater biodiversity conservation.

- Emphasised the avoidance and minimisation of

disturbance of ecosystems and loss of biological diversity.

National Environmental Management:

Biodiversity Act (Act No. 10 of 2004) (RSA, 2004)

- Provides the management and conservation of South Africa’s biodiversity within the framework of the National Environmental Management Act 1 998; which protects species and ecosystems that warrant national protection

(30)

21 National Water Act (Act No. 36 of

1998) (RSA, 1998a)

- Legal framework for the effective and sustainable management of our water resources. It defines water resources as rivers, streams, wetlands, estuaries and groundwater (RSA, 1998a).

- It contains rules about the way that the water resource is protected, used, developed, conserved, managed and controlled in an integrated manner.

- Enforced the identification and prioritisation of catchments and wetlands.

- Enforced the authorisation of any water use.

Conservation of Agricultural Resources Act (Act 43 of 1983) (RSA, 1983)

- It prevents the unauthorised cultivation of virgin soil, regulates the utilisation and protection of vleis, marshes, water sponges and water courses, and prevents the alteration of flow patterns of runoff water.

National Spatial Biodiversity Assessment (2004, 2011)

- Spatially assesses the state of South Africa’s biodiversity by incorporating four fundamental components: terrestrial, freshwater, estuarine and marine environments. The reports include the state of wetland ecosystems. This information is useful to policymakers, decision makers and practitioners. Revise and update key policies and strategies, including the National Biodiversity Strategy and Action Plan (NBSAP).

National Biodiversity Strategy and Action Plan (2005)

- Preparation of national biodiversity strategy aimed at mainstreaming planning and activities that can impact (positively and negatively) on biodiversity i.e. wetlands (DEAT, 2006).

Other legislation pertaining indirectly to wetland conservation are laws against poor forestry and mining practices (the National Forests Act 84 of 1998 and the Mineral and Petroleum Resources Development Act 28 of 2002), and legislation protecting the environment against destruction caused by urbanisation and infrastructure (the Physical Planning Act 125 of 1991, the Development Facilitation Act 67 of 1995, the South African National Roads Agency Limited and National Roads Act 7 of 1998, and various Environmental Impact Assessment regulations) (Winstanley, 2001). However, government legislation, policies, and programmes are worthless in protecting wetlands if the regulations are not enforced. Perhaps a combination of educating the public on the benefits of wetlands and enforcement of wetland legislation/policies would be a better approach. If the public does not recognise the benefits of wetland preservation, wetlands will not be preserved (Votteler and Muir, 1996).

(31)

22

2.2. MAPPING WETLANDS

The value in identifying the sizes, locations and types of wetlands in terms of conservation and protection of freshwater ecosystems lies in providing great input into the prioritisation, planning, and rehabilitation needed for the conservation of freshwater ecosystems globally. As seen in the NFEPA project, a wetland inventory forms the baseline spatial data and could be used for many purposes, including comprehensive resource management plans, environmental impact assessments, natural resource inventories, habitat surveys, and the analysis of trends in wetland status (Wilen et al., 2002). Begg (1986: 48) stated that “one cannot hope to formulate a policy for the utilization (or management) of wetlands without knowing where these areas lie, or how much of each catchment comprises of wetlands”.

This section reviews techniques used to map wetlands both internationally and in southern Africa.

 International wetland mapping approaches

Techniques to map wetlands using satellite imagery are extensive and -reviewed (see Kulawardhana et al., 2007; Ozesmi and Bauer, 2002). Currently, methods and approaches to map wetlands using satellite imagery range from supervised, to semi-automated, to unsupervised approaches in a range of temporal, spectral and spatial resolutions. Li and Chen (2005) contend that integrating spatial environmental data with satellite imagery has improved the accuracy of mapping wetlands. Using rule-based approaches where classification is based on spatial data themes (i.e. land cover, soil texture, terrain) in addition to satellite imagery, has raised the overall classification accuracy from 69% (traditional supervised classifier) to 83%. Ozesmi and Bauer (2002) summarised the literature on satellite remote sensing of wetlands and assessed the success of various methods in classifying wetlands and distinguishing them from other land cover classes. The main conclusions drawn from their review are as follows:

1. Classification of satellite imagery may not match the information gained through field surveys but can provide complementary information on wetlands over a large survey area. It can identify areas where change is occurring and more detailed information is needed.

2. Rule-based classification methods generally provide better results than conventional statistical classification methods, often because of their use of ancillary data.

3. Multi-temporal imagery and ancillary data allow for the highest accuracy in wetland identification and discrimination from other land cover types.

4. Wetlands should be separated from other land cover types prior to their classification using ancillary data.

There are a number of different automated and semi-automated approaches to mapping wetlands using satellite imagery, and Kulawardhana et al. (2007) evaluated these methods. The methods included the tasselled cap wetness index (TCWI), normalised difference water index (NDWI), multi-band vegetation indices (MBVI), two band vegetation indices (TBVIs), normalised difference vegetation index (NDVI), and data fusion involving Enhanced Thematic Mapper (ETM+) and SRTM (Shuttle Radar Topography Mission) data and then classifying the same. In this review Kulawardhana et al. (2007) concluded that many of the automated methods resulted in very low accuracies in the wetlands that were delineated and/or very high errors, and are therefore inappropriate in mapping wetlands at larger spatial scales. Automated approaches are unsupervised techniques that rely purely on spectrally pixel-based statistics and incorporate no prior knowledge of the characteristics of the themes being studied (Xie et al., 2008). This study was done using single-date imagery, and multi-temporal imagery may have boosted the overall accuracy. However, the major limitation in the automated process was the spectral

(32)

23

similarity between the vegetation canopies of wetlands and the vegetation or agricultural crop cover found outside wetlands. Semi-automated approaches on the other hand include the enhancement of satellite imagery to obtain a better contrast between wetland and non-wetland land cover types, digitising, and using ancillary data to supplement Enhanced Thematic imagery. Semi-automated approaches involve supervised classification methods where the established classification is learnt from a training dataset, which contains predictor variables measured in each sampling unit and assigns prior classes to the sampling units. Kulawardhana et al. (2007) stated that semi-automated approaches increased the accuracy by at least 30% in comparison with purely automated approaches when mapping wetlands.

Knight et al. (2009) used Landsat TM and Enhanced Thematic Mapper (ETM+) imagery over a 16-year period to develop comprehensive wetland maps for the state of Queensland, Australia. The approach was a classification method called the Standing Water Body (SWB) method. The SWB method separated the main spectral and land cover elements of wetlands (vegetation, standing water, and shadow cast by vegetation and topographic relief) and used rules to combine spectral classes to provide multi-temporal information for mapping of wetlands. They concluded that the SWB method requires enhancement through the inclusion of an NDWI and ancillary data such as vegetation mapping and drainage networks to improve the accuracy in mapping wetlands even further.

Landmann et al. (2010) mapped wetlands over a wide area in semi-arid Africa using 250-metre Moderate- resolution Imaging Spectroradiometer (MODIS) metrics and topographic variables by using a multi-temporal approach. They used ancillary data to supplement the multi-temporal imagery, and this included data such as sinks and streamline areas. The ancillary data was used to mask potential wetland areas, minimising spectral confusion. This study method of mapping wetlands over a large spatial area was largely successful, producing the most detailed wide area wetland dataset available for West Africa. Landmann et al. (2010) mention that this study “demonstrates the need for earth observation methods to not only base their assessments on satellite observations, but also utilise linkages to other higher resolution in situ point data sets”. The resolution of MODIS imagery was problematic and resulted in inaccuracies in pixel variability and mapping errors, and it was challenging to assess even using high-resolution datasets.

The success of these wetland mapping techniques is a result of a combination of approaches that are simple, statistically significant, or highly complex. The combination of high spatial and temporal resolution imagery is essential in mapping wetland ecosystems, but such single satellite imagery is currently unavailable (Ryo et al., 2012). To address this need, Ryo et al. (2012) researched the applicability of multiple end member spectral mixture analysis (MESMA) using single bi-sensor imagery with Landsat Thematic Mapper and MODIS data, across different orbiting periods. This complex research approach demonstrated how MESMA can be applied for multi-scale mapping of wetland ecosystems; however, spectral similarity between dark water and shade still affected the agreement in land cover classes (spectral confusion).

 Southern African wetland mapping approaches

The climate over southern Africa is extremely variable, and therefore saturated soil is often an unreliable indicator of wetland conditions or boundaries, as during certain years wetlands may be much larger or wetter than others. Using field surveys when mapping wetlands to confirm this variability would be ideal, but it is expensive and time consuming. It is clear that at a national scale the mapping of wetlands and the creation of an NWI is no simple task; however, these processes are of critical importance in the conservation and management of wetland systems. An NWI is important for identifying where wetlands are located, in order to prioritise sites according to the functions and values of each wetland site, including ecological, social and cultural values. The NWI provides a tool to inform the planning and management related to wetlands at all levels, and is a baseline for measuring future changes in a wetland’s area, function and value (DEAT, 1997).