Participatory spatial data analysis to assess environmental degradation


Dr. Isaac Agyemang
Department of Environmental and Resource Studies of the University for Development Studies
Tamale, Ghana
Email: issacagyemang@yahoo.com

Abstract
While geographic information system (GIS) is now widely accepted as an important tool in research involving natural resources assessment and management, others are of the view that qualitative approaches have been and continue to be the main methodological technique for natural resource-based projects where greater emphasis is required on societal concerns and responses to environment-related problems. In this article, participatory spatial data analysis and interpretation in GIS involving environmental degradation and assessment in the Bolgatanga and Talensi-Nabdam districts of northern Ghana is presented. Environmental resources in the area are characterised by a dry savannah climate and vegetation, poor soils and irregular rainfall patterns, leading to poor yields and levels of production. The area is one of the worst degraded regions in Ghana with high rate of illiteracy, poverty, complicated land tenure system and high population growth rate. The research was conducted with the aim of involving the affected communities in the analysis and interpretation of spatial data of land use changes in some select communities. It is argued in this article that proper environmental policy can be formulated by the state government through district assemblies if participation of local people in the analysis and interpretation of spatial data, which hitherto has been done by experts using geo-spatial techniques, is encouraged.

GIS and participatory spatial analysis
Participatory GIS has received great attention in recent years as a tool for involving local people in spatial data analysis (Alcorn, 2000; King, 2002). The technique has, in recent years, moved from using basic sketch maps to include a range of technologies including GPS, compasses, three-dimensional modelling, photo-mapping, GIS based maps, remote sensing image interpretation, mobile GIS and utilising all kinds of multi-media graphics software in visualization (Alcorn, 2000; Gonzalez, 2000; McCall, 2002). Of the elementary significance of PGIS implementations are the questions of access, control and ownership of geographical information and output (Dunn, 2007). In the process of representing different geographical understandings and in attempting to reveal contradictions and similarities in spatial thinking and activity (Williams and Dunn, 2003), PGIS seeks not to privilege any one type of information but to grant equal validity to all (Dunn, 2007).

Participatory GIS (PGIS)
PGIS has received great attention in recent years as a tool for involving local people in spatial data analysis (Alcorn, 2000; King, 2002). The technique constitutes a process in which local people create representations based on local ecological knowledge and engage in the analysis of objects, relationships and issues. The technique has, in recent years, moved from using basic sketch maps to include a range of technologies including GPS, compasses, three-dimensional modelling, photo-mapping, GIS-based maps, remote sensing image interpretation, mobile GIS and utilising all kinds of multi-media graphics software in visualisation (Alcorn, 2000; Gonzalez, 2000; McCall, 2002).

In the process of representing different geographical understandings and in attempting to reveal contradictions and similarities in spatial thinking and activity (Williams and Dunn, 2003), PGIS seeks not to privilege any one type of information but to grant equal validity to all (Dunn, 2007). In this way, local technical knowledge can grant poorer groups an equivalent standing to outsiders (McCall, 2003) and in participatory spatial planning such knowledge may be the only resource that the poorest groups control while their land resources, property or labour are rapidly appropriate (McCall, 2003). Of critical relevance is the extent to which local knowledge can be portrayed in a spatial way and through the use of GIS. According to Grenier (1998) and Dunn, (2007), through generations, local knowledge is expressed through stories, songs, folklore, proverbs, cultural values and agricultural practices and usually communicated orally. McCall and Minang (2005) commented that local knowledge is normally more reliable, and may be more accurate as it embodies generations of practical essential knowledge and operates in interactive and holistic systems. While not geographical, much local knowledge has both embedded geographical context in which the natural environment is central and specific spatial associations-for example, knowledge related to location of resources, environmental hazards, ecosystems and spatial correlations between groups and resources (McCall, 2003). The thematic data layering properties of GIS facilitate representation of multiple perspectives and offer potential of a holistic view of local communities. Some local ecological knowledge can be mapped through the use of GIS for decision making purposes (Dunn, 2007). Participatory spatial data analysis and interpretation (PSDAI) as a subsidiary of PGIS as adopted in this article is reviewed in the next sub section.

Participatory spatial data analysis and interpretation
The term was first proposed by Agyemang et al. (2007) to mean the empowerment and involvement of local people to review, comment and give meanings to spatial data of relevance to them. As a synonym for ground truthing, participatory spatial data analysis and interpretation is the validation of spatial data by local people in terms of its accuracy and meanings to any observed changes. The term originated from the GIS and Society concept (Sheppards, 1995; Harris and Weiner, 1996) to mean the recognition of the fundamental importance of involving local people in the evaluation of spatial data for decision making purposes. Through a wide range of participatory rural appraisal techniques (Agyemang et al, 2007), selected people from a community are given the chance to critically review spatial data of interest to them, identify issues of relevance and criticise and/or add meanings in terms of any observed inconsistencies on the spatial data as presented. The approach is based on participants’ own deep knowledge on the ground as related to the spatial data presented, their ability to visualise and interpret spatial data (Williams and Dunn, 2003) and less on participants’ mapping in the analysis and interpretation of image maps as originally proposed by (McCall, 2003).

PSDAI, as used in this article, is an attempt to make use of conventional GIS results, within local ecological knowledge, to assess the spatio-temporal state of the natural environment. It is based on an interpretative philosophy of examination of the meaning and symbol content of quantitative data (Seidel and Kelle, 1995) and the use of reality probability statements instead of real accurate description statements by participants, meaning that the views, opinions and comments of participants are taken seriously, despite sometimes being triangulated with theories, secondary evidence and statistical data sources (Glaser and Strauss, 1967; Gadamer, 1989). In reviewing environmental degradation in northern Ghana it is suggested in this article that local ecological knowledge is of vital importance in the interpretation and analysis of spatial data of relevance to environmental degradation. In order to assess the complex interplay of environmental degradation in Bolgatanga and Talensi-Nabdam districts of northern Ghana, this literature suggests that local people should be given the meaningful opportunity to participate; local ecological knowledge held by local people should be given a prominent role in environmental degradation assessment; and different people and interest groups should work together to reach consensus about critical issues on environmental degradation. It is envisaged that a GIS which is vested in the interests of people through an approach based on GIS in participatory research may be more successful and achievable than a truly participatory GIS (Williams and Dunn, 2003). The aim of this article is to use a PSDAI to assess environmental degradation in the Bolgatanga and Talensi-Nabdam districts of northern Ghana.

Study Area
The area of this study (Figure 1) consists of the Bolgatanga and Talensi-Nabdam districts of northern Ghana, one of the most deprived and degraded areas of the country. They are located on the north-eastern corridor of Ghana between longitude 1°W and 0°E and 10°N and 11°N and cover an area of 1,509 km˛ or 16.7% of the 8,842 km˛ of the Upper East Region of Ghana. It is part of the tropical continental climatic zone characterised by pronounced dry and wet seasons (Benneh et al., 1990). Rainfall is usually infrequent, discrete and largely unpredictable.


Figure 1: Map of Study Area
Bolgatanga, which is about 120 km˛ of the 1509 km˛ of the study area, is the largest town in the study area and district capital of the Bolgatanga municipality which is approximately 420 km˛. During the period of 1990 through 2004, and as part of the economic recovery programme, the study area has seen major infrastructural developments in terms of expansion of towns and settlements. Small-scale mining communities such as Accra, World Bank, Tarkwa, Kejetia, Bantama and Obuasi were created by small-scale miners in the late 1990s. Other well known communities in the study area and of interest include Sheaga, Nangodi, Kongo, Winkogo, Pwalugu, Pelungu, Yikine, Sekoti, and Zuarungu, Gororo Yikpemeni, Tindongobulagu Duusi, Sheaga, Pelungu and Sherigu (Ghana Statistical Service, 2000).

Spatial data collection and analysis
To assess the current and recent state of the environment study area, 30m resolution spatial data of three different scenes, 1990, 2000 and 2004, taken during the dry season, were utilised. The acquired spatial data were geo-referenced to the UTM projection of the World Geodetic System of 1984 datum and with insignificant cloud coverage. The acquired images were sub-setted to include only the area of interest (Bolgatanga, Talensi-Nabdam Districts of northern Ghana).

Using computer software (ERDAS Imagine 8.7), the raw spatial data was displayed on a computer and various enhancement techniques performed until the most appropriate band combination was arrived at. Unsupervised classification technique was performed on the raw images using ISODATA (Iterative Self-Organising Data Analysis Technique) algorithm (Jensen, 1996). The ISODATA technique is a modified version of K-means clustering that aids in the categorisation of pixels based on the spectral differences in each band, thus allowing for the development of unique class signatures and diminishes the spectral distances between classes (Tou and Gonzalez, 1974). The objective of the K-means algorithm is to minimise the within cluster variability. Using the technique, the number of classes desired and a confidence threshold were input into the computer. An initial set of 20 random classes of 100 iterations and at 95% confident threshold were input to the computer for it to build the classes iteratively until the confidence threshold of 95% was reached. During the process, each pixel of spectral characteristics was examined and similar pixels aggregated into classes of 20 cover types. The result was unsatisfactory because of the large number of overlaps of the land-cover classes. To refine the classes, manual recoding was done to further aggregate the initial spectral classes into information classes containing categories that represent land-cover features of interest. This was made possible through the use of areas of known land cover, visual interpretation and spectral profiles compiled by CERSGIS of the University of Ghana. Seven classes were obtained after merging inseparable classes in an initial set of 20. Class names and colours were identified and assigned to the seven identified classes based on reference classification scheme designed by Agyepong et al. (1996) modified from Anderson et al. (1976) land-cover classification scheme.

Initial spatial data results
Results of the spatial extent of the various land-cover types, for the three periods of 1990; 2000 and 2004 as shown in Figure 2, Table 1 and Figure 3 demonstrate extensive conversion of savannah vegetative woodland types to built-up and barren land.


Figure 2 (a): Spatial data for 1990; 2000 and 2004

Figure 2 (b): Extent of land cover changes: graphical presentation
Table 1: Extent of land cover changes

PSDAI results and discussions
Table 2 contains summarised responses of participants concerning the spatial distribution of the various land-cover types in the study area for the 14 year period of the study. Most participants were able to discuss, to the best of their knowledge, the spatio-temporal distribution of the land-cover types in the study area with reference to the spatial data provided. The general observation of most of the participants was that in communities such as Bolgatanga, Zuarungu and Kombosigo, the original land-cover type of open savannah woodland interspersed with some closed savannah woodland has recently been replaced by grasses and barren environment. Participants attributed this to human activities. The popular assertion by Conteras-Hermorilla (2000) that the degradation of savannah vegetation is attributed to human actions seem to be applicable in the study area as observed by the research participants.

Table 2: Participants contribution of observed changes on the environment

In the Nangodi, Kongo, Pelungu and Sekote communities of the Talensi Nabdam District, participants noted that the original closed savannah woodland was recently replaced by dominant herbs and savannah grasses, few patches of open savannah woodland and built-up and barren environment. Participants attributed the observed changes to social processes such as population growth and migration and human activities such as surface and underground mining in Nangodi and the immediate environment.

In the south-eastern portion of the study area such as Duusi, Accra, World Bank, Kejetia, Bantama, Tarkwa and Obuasi of the Talensi-Nabdam district, participants noted that the original savannah woodland (both closed and open savannah woodland) interspersed by few dense herb and savannah grasses was recently cleared and subsequently replaced by dominant savannah grasses and built-up and barren environment. Most of the research participants were of the view that human activities such as mining, grazing and sand winning might have initiated such environmental changes.

Most of the original closed savannah vegetative type at Sherigu community, in the north-western portion of Bolgatanga, had also been degraded to barren environment with few patches of scattered dense herb and savannah grasses as noted by the participants. Those who responded attributed this to various human activities such as farming, grazing, human settlement and mining activities. In the Pwalugu, Tongo and Winkogo communities, in the south-western portion of the study area and part of Talensi-Nabdam district, the nature of degradation was not different from other locations as observed and noted by the research participants. The original open-savannah woodland type, interspersed with a few closed savannah types had recently been degraded as a result of human activities and various government infrastructural developments initiated by the district assembly under the Local Government Act 462 of 1993 which, among other things, empowered district assemblies to set developmental activities as their priority to improve living standards. Figures 3 and Figure 4 show local knowledge on the spatial distribution of observed land cover changes converted into geo-spatial data via GIS.


Figure 3: Local knowledge of recent land cover changes via GIS

Figure 4: Local knowledge of current land cover changes via GIS
Replacements of savannah tree and grass types
As part of the spatial data analysis and interpretation, participants were asked whether there had been any changes in the abundance of recognised indigenous and exotic trees during the 14 year period of study. Table 3 demonstrate research participants’ responses concerning such changes and the perceived direct social causes.

Table 3: Participants’ responses concerning changes in abundance of trees

Notwithstanding the diverse opinions by the research participants, the above findings demonstrate that local people’s ideas and views are not based on mere assumptions but on real issues about their environment as has also been observed by Doolittle, (2003). The findings also relate to assertion by Kessler (1992); Moleele et al. (1996); Nangula and Oba (2004); Blench (2006) that the degradation of savannah vegetative forest in most arid and semi-arid regions is spatio-temporal in nature through extreme human exploitation of tree types and dominant grasses.

Pattern of land cover changes
The suggestion that most land-cover type changes in many savannah environments are spatio-temporal in nature and are trapped in a spiral of irreversible and uncontrollable worsening degradation usually from savannah woodland to built-up and barren environment (Barrow, 1991; Rees, 1992; Rees and Wackernagel, 1994; Dreshsel et al., 2001; Nsiah-Gyabaah, 2004 and Reed et al., 2007), was contrasted by deductions made from participants observations concerning spatio-temporal patterns of change and to verify some of the issues raised during the analysis of the GIS and remote sensing results.

Using the hermeneutic approach which is grounded in the philosophy of understanding and interpretation of qualitative data (Gadamer, 1989), most of the participants, basing their comments on the spatial data presented, commented that the degraded environment, as evident in communities such as Bolgatana, Tongo, Nangodi, Duusi, Sherigu and Pwalugu, where many developmental projects and human activities have, over the year taken place, can stimulate environmental innovations necessary to overcome the drastic conversion of savannah woodland to barren, degraded environment and maintain a quality environment. Most of them believed in the evidence of ecological regeneration under good management practices from closed savannah woodland to barren environment and back to closed savannah woodland through grasses of various types. This supports earlier assertion by Reid et al. (2000) that barren environment, when left unattended for a period of time, may revert initially to dense herbs, shrubs and savannah grasses and finally to the original savannah vegetative cover.

Conclusion
In this article, attempts have been made to examine and analyse the state of the environment using participatory spatial data analysis and interpretation in northern Ghana. Even though the article demonstrated that conventional GIS offers a platform on which different sets of spatially referenced data are assembled, analysed and represented (Burrough, 1986; Aronoff, 1989; Goodchild and Gopal, 1989; Laurini and Thompson, 1992), there still remained unanswered important spatial issues that needed further clarification or truthing. The integration of local ecological knowledge in the latter sections of the article put more emphasis on some of the issues raised and added more social content to the observed environmental changes as captured on the spatial data. The assumption that neither conventional GIS nor local ecological knowledge can stand alone to assess environmental degradation in developing countries was thus tested to be true in this article.

It becomes evident from the article that there had been land-cover changes in the study area during the period of the study and this had manifested itself in the decline of savannah woodland, with a total cover of 634 km˛, over the years with corresponding increase in grasses of various types (208 km˛) and built-up and barren environment (392 km˛). These changes, as commented through the participatory spatial data analysis and interpretation, were due to the existence of activities such as surface mining, underground mining, sand winning, bush burning and quarrying that has taken roots in the study area. The paper thus demonstrates the importance of the adoption of participatory spatial data analysis and interpretation as an effective instrument in assessing the state of the environment of a degraded community such as the Bolgatanga and Talensi-Nabdam districts of northern Ghana.