77 CHAPTER 8 Regional Characterization of Inland Valley Agroecosystems in West and Central Africa Using High- Resolution Remotely Sensed Data Prasad S. Thenkabail and Christian Nolte BACKGROUND Inland valleys (IVs) (locally also known as fadamas, bas-fonds, and dambos) have the potential to become agroecosystems with a substantial impact on African food production (Izac et al., 1991). These agroecosystems are favorable for rice cultivation and dry season cropping and have the potential to increase acreage and yields in Africa if careful attention is paid to technical, envi- ronmental, and socioeconomic constraints (Juo and Lowe, 1986). Inventorying, mapping and characterizing inland valleys at regional level (meso/semidetailed level, typically mapped on scales of 1:50,000 through 1:250,000) in sample areas of macrolevel agroecological zones (Figure 8.6, color section) is crucial for the selection of representative re- search sites, which then allows for the development of appropriate technologies that can be reli- ably tested and transferred to larger regions (regionalization or technology transfer). Recognition of the importance of inland valley agroecosystems led the International Institute of Tropical Agriculture (IITA) to adopt a new research agenda, as put forth in Izac et al. (1991). This strategy involved a combination of biophysical and socioeconomic research issues to be addressed at three spatial scales: a. level I (macro or continental or subcontinental, typically mapped in scales of greater than or equal to 1:5,000,000); b. level II (meso or regional or semidetailed mapped in scales of 1:100,000 to 1:250,000); and c. level III (micro or research site/watershed related, mapped in scales of less than or equal to 1:50,000). This was conceived to facilitate the design of appropriate technology, able to sustain the highly varying resource base and at the same time to be acceptable to smallholder farmers in the diverse socioeconomic and ethnic environment of West and Central Africa. The first step to characterize parameters critical to land use of IVs constituted the (macro) level I map based on secondary agroecological and soil data using a Geographic Information System (see Thenkabail and Nolte, 1995a for details). The objective was to map on a subcontinental © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 78 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT (macro) scale broad agroecological and soil zones in the mandate area of IITA in West and Central Africa (Figure 8.1b). The level I map was the result of combining two parameters: a. IITA’s agroecological zones; and b. major soil groupings according to the FAO classification (FAO/UNESCO 1974, 1977). The five agroecological zones of IITA’s map, namely northern Guinea savanna, southern Guinea savanna, derived and coastal savanna, humid forest, and midaltitude savanna, were over- laid with the 23 zones of major soil groupings. This resulted in 18 zones of more than 10 million ha each in West and Central Africa, for a total of 492 million hectares (Figure 8.6, color section). Regional (or level II) characterization of inland valley agroecosystems were planned within the “windows” of macro (level I) zones. For rapid characterization and mapping at a regional scale of such a large area, which is spread across a subcontinent, it was only feasible using high-resolution Figure 8.1 Spatial distribution of level 1 agroecological and soil zones (AEZ) in Landsat 192/54. Of a total area about 3.12 Mha, 49% fall into AEZ 2 and 46% into AEZ 7. The remaining 5% area is out- side the 18 zones mapped in Figure 8.1. © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing satellite images (Thenkabail and Nolte, 1995a; Thenkabail and Nolte, 1996). The location of the acquired Landsat TM and SPOT HRV satellite images for regional (level II) characterization study in relation to the IITA level I (macroscale) map are displayed in Figure 8.6 in color section. Four international research centers (IITA, WARDA, CIRAD, Winand Staring Centre and Wage- ningen Agricultural University) and several national research systems (from Republic of Benin, Burkina Faso, Côte d’Ivoire, Ghana, Mali and Nigeria) acknowledged this holistic research ap- proach to characterization of IVs at a workshop at WARDA headquarters, Bouaké, Côte d’Ivoire, June 1993, since it fit well with the ideas developed by these institutes. RATIONALE AND JUSTIFICATION Characterization of the agroecosystems in which IVs occur is available at subconti- nental/macrolevel of West and Central Africa (Hekstra et al., 1983; Andriesse and Fresco, 1991; Izac et al., 1991; Windmeijer and Andriesse, 1993). Country/mesolevel studies of inland valley agroecosystems were published for the Republic of Benin by Kilian (1972), for Senegal by Bertrand (1973), for Burkina Faso and Mali by Albergel et al. (1993), and for Côte d’Ivoire by Becker and Diallo (1992). Turner (1985, 1977) gives mesolevel details upon biophysical data of fadamas in Central and Northern Nigeria, which were completed by the studies of Kolawole (1991) on fadama economics and management. Many studies upon dambos and their cultivation pattern have been done in South and Southeastern Africa. For example, Rattray et al. (1953) de- scribed vlei cultivation in Rhodesia/Zimbabwe. Other regional studies in this part of sub-Saharan Africa are reviewed by Ingram (1991). Given the fact that the characteristics of IVs are known to vary dramatically within and across agroecosystems, it is inadequate to characterize only a few IVs as most conventional studies often do. Lack of representativeness of a few study sites in the context of a regional agroecological zone leads to limited extrapolation or regionalization of the results of key sites to other areas within the same agroecological zone. Lack of an appropriate approach to characterization constitutes a major constraint to research activities in developing technologies that are able to sustain the resource base of large regions and are adoptable by diverse groups of farmers with highly differing socioeconomic and ethnic back- grounds. Therefore, IITA attempted to systematically characterize agroecosystems of IVs in West and Central Africa. This three-tier methodology from macro- to microlevel is intended to lead to the development of technologies at benchmark sites. OBJECTIVES AND APPROACH Level-II (regional) characterization work involves establishing detailed characteristics of IV agroecosystems in sample areas of level I. The specific objectives envisaged in level-II characterization fall within the overall objectives of Izac et al. (1991) and were: 1. inventory the area of inland valley systems and their uplands; 2. map the spatial distribution of IVs, and study their spatial variability; 3. study the existing land-use pattern of IVs and uplands, and establish their interactions; 4. explore the potential of IVs for dry-season cropping; 5. determine the existing crop types and cropping pattern in IVs, and on uplands; 6. map the major road systems and significant settlements; 7. establish the cultivation pattern with respect to distance from road network and settlements; 8. study the watershed, and establish morphometric characteristics of IV watersheds; REGIONAL CHARACTERISTICS OF INLAND VALLEY AGROECOSYSTEMS 79 © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 80 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT 9. describe morphological IV characteristics such as shape, size, and slope; 10. make it possible to compare and contrast IVs within and across agroecological and socioeconomic zones in West and Central Africa; 11. rationalize the selection of representative benchmark site/s or benchmark IV watershed/s for the main phase research (i.e., technology development) activity; and 12. reveal the socioeconomic conditions of IV use. (Note that this is not part of the ongoing study). IITA adopted a specific approach to achieve the above specific level-II objectives through inte- gration of high-resolution satellite data with Global Positioning System (GPS) data, and ground- truth data in a Geographic Information Systems (GIS) framework. A comprehensive methodology Figure 8.2 Spatial distribution of level 1 agroecological and soil zones in SPOT K:J of 47/338. Of the total area of ).49 Mha, the entire 100% are falls in AEZ 16. © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing REGIONAL CHARACTERISTICS OF INLAND VALLEY AGROECOSYSTEMS 81 for rapid characterization and mapping of inland valley systems at regional level using satellite imagery was developed and reported by Thenkabail and Nolte (1996, 1995a). In this chapter two Landsat TM images (path 192 and row 54 and path 197 and row 52, [(see Figure 8.1and 8.2; and Table 8.1 for details and spatial location)], and one SPOT HRV image (Fig- ure 8.3, 8.6 in the color section and Table 8.1)] were was used for the study. These images were se- lected, based on a set of criteria that included availability of cloud-free (0% cloud cover) satellite scenes, location of images in different agroecological and socioeconomic zones, most recently available dates of satellite overpasses (images were of early 1990’s-see Table 8.1), accessibility of Figure 8.3 Spatial distribution of the level I agroecological and soil zones in Landstat 197/52. Of the total area of about 3.14 Mha., 45% is in AEZ 1 and 12% in AEZ 2. As for the rest of the area, 43% is outside the 18 AEZs mentioned in Figure 8.1. © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing the region for ground-truthing, and coverage during wet and dry seasons. The images covered four agroecological and soil zones (AEZ) of level I—AEZ 1, 2, 7, and 16. Inland valleys, their fringes, and uplands distributed in each of the above AEZs were be re- ported along with their land-uses, cultivation intensities , and their spatial distribution. The level II work offers : a. the capability to zoom in on IV agroecosystems from macro- to microlevels (“top-down” ap- proach) leading to the rationalization of benchmark site selection for technology develop- ment research activities; and b. the capability to extrapolate research results to larger regions or the regionalization of re- search results (“bottom-up” approach) for the transfer of technology. DEFINITION USED FOR MAPPING INLAND VALLEYS It is obvious from the various accounts in the literature that there is no single, widely accepted definition of IVs. They (IVs) are one of the many forms of wetlands that are characterized in their bottom by hydromorphic soils. Beyond this, there are significant variations in the definitions and characteristics of IVs even within the same region as perceived by different people (e.g., Savvides, 1981; Raunet, 1982; Hekstra et al., 1983; Acres et al., 1985; Mäckel, 1985; Turner, 1985; An- driesse, 1986; Oosterbaan et al., 1987; Andriesse and Fresco, 1991; Izac et al., 1991; Mokadem, 1992). For example, Mäckel (1985) defines a uniform zonation of dambos in Southeast Africa de- pending on vegetation, soil type, moisture content, and morphodynamics of the dambo. Lack of a consistent definition of IVs has been a constraint in comparing or synthesizing dif- ferent studies. To most researchers, characterization of IVs meant valley bottom characterization. Others included fringes also in their studies. Andriesse and Fresco (1991) proposed a physiohy- drographic model of an inland valley. The working definition adopted in this study will be as follows: Inland valleys (IVs) comprise valley bottoms and valley fringes (Figure 8.5). Valley bottoms are characterized by hydromorphic soils that constitute a relatively flat surface with or without a central stream. Valley fringes refer to areas along the slopes of the valley; rainfall either runs off above the surface of these areas or in- terflows horizontally on impervious subsurface layers toward the valley bottom and the central stream. Valley fringes, typically, have two distinct characteristic zones (Figure 8.5): 1. the lower part of the valley fringe immediately adjoining the bottoms that may have a high likelihood of a seasonal hydromorphic zone with significant potential for dry-season crop- ping; and 2. the upper part of the valley fringe with steeper slopes, in zones with less than 1400 mm rain- fall (Guinea savanna) characterized predominantly by impervious layers (ironstones or cara- paces) from which rainwater quickly runs off to the valley bottom). Soils in these upper portions of valley fringes dry out rapidly once the rains have ceased, and therefore, have no potential for dry-season cropping. METHODOLOGY An overview of the methodology has been shown in Figure 8.4 and described in detail in Thenkabail and Nolte (1995) and Thenkabail and Nolte (1996). The methodology permits a rapid characterization of large areas on a regional scale. The methodology includes the description of techniques for: 82 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing Figure 8.4 Methodology for mapping and characterizing Inland Valley (valley bottoms plus valley fringes) agr oecosystems using Remote Sensing, Global Positioning System (GPS), and Ground-Truth Data. © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 84 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT 1. valley bottom mapping; 2. valley fringe mapping; 3. mapping roads and settlements; 4. establishing land-use land-cover characteristics across the IV toposequence (uplands, valley fringes, and valley bottoms); and 5. establishing other IV characteristics such as IV stream densities and stream frequencies. Figures 8.7 to 8.11 in the color section demonstrate the application of the methodology (Fig- ure 8.4) for certain subareas of the study. Using the methodology, valley bottoms were high- lighted (Figure 8.7, see color section) and delineated (Figure 8.8, see color section) for a subscene in Landsat TM path and row 192/54 that is part of agroecological and soil zone (AEZ) 7. The fringes adjoining the bottoms were mapped as illustrated in Figure 8.9 in the color sec- tion. A similar procedure was used to delineate valley bottoms from a SPOT subscene (Figure 10, see color section) which is in AEZ 16. Figure 8.11 in the color section (in AEZ 2) illustrates the large widths of the valley bottoms that are seasonally flooded and are characteristically dis- similar to valley bottoms in other AEZs. Sixteen land-use classes (Table 8.2) were mapped con- sistently across each study area based on the percentage distribution of 10 different land-cover types (Table 8.3). Table 8.1 Parameters Describing the Level I Agroecological and Soil Zones a Agroecological zone Level I According to IITA’s LGP c Major FAO Soil Area e AEZ b Definition (days) Grouping d (million ha) 1 Northern Guinea savanna 151–180 Luvisols 25.2 2 Southern Guinea savanna 181–210 Luvisols 18.4 3 Southern Guinea savanna 181–210 Acrisols 12.4 4 Southern Guinea savanna 181–210 Ferralsols 11.9 5 Southern Guinea savanna 181–210 Lithosols 10.7 6 Derived savanna 211–270 Ferralsols 47.2 7 Derived savanna 211–270 Luvisols 24.9 8 Derived savanna 211–270 Nitosols 14.2 9 Derived savanna 211–270 Arenosols 14.0 10 Derived savanna 211–270 Acrisols 11.7 11 Derived savanna 211–270 Lithosols 10.8 12 Humid forest > 270 Ferralsols 150.1 13 Humid forest > 270 Nitosols 27.2 14 Humid forest > 270 Gleysols 19.2 15 Humid forest > 270 Arenosols 18.9 16 Humid forest > 270 Acrisols 18.0 17 Mid-altitude savanna b Ferralsols 45.4 18 Mid-altitude savanna g Nitosols 12.3 a AEZs in bold have been investigated in this study at next level (level II) using high resolution satellite imagery, and the results are reported here b AEZ: level I agroecological and soil zones. c LGP: length of growing period. d Names refer to the FAO soil classification scheme of 1974 (FAO/UNESCO 1974). e The area figures are for West and Central Africa. f Area distribution of LGP in AEZ 17 is: 151–180 days 11%, 181–210 days 9%, 211–270 days 59%, > 270 days 21%. g Area distribution of LGP in AEZ 18 is: 151–180 days 2%, 181–210 days 5%, 211–270 days 53%, > 270 days 40%. © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing Figure 8.5 Cross-section showing model inland valley as defined in this study. © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 86 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT STUDY AREAS AND GROUND-TRUTHING In this chapter results of the three study areas (each satellite image comprising a study area) have been discussed (Table 8.4 and 8.1; Figures 8.1, 8.2, and 8.3; Figure 8.6, see color section). These study areas cover four distinct agroecological and soil zones (AEZ) in: 1. AEZ 2 (49% of the total area) and AEZ 7 (46%) by Landsat path 192 and row 54 (having an total area of 3.12 million hectares) (Figures 8.1, 8.6, see color section, Table 8.1 and Table 8.4); 2. AEZ 16 (100% of the total area) by SPOT HRV K 47 and J 338 (having an total area of 0.39 million hectares); (Figures 8.2, 8.6, see color section, Table 8.1 and Table 8.4) and 3. AEZ 1 (45% of the total area) and AEZ 2 (12%) by Landsat path 197 and row 52 (having an total area of 3.14 million hectares) (Figure 8.3, 8.6, see color section, Table 8.1 and Table 8.4). Table 8.2 Land-Use classes Mapped in Level-II Characterization of IV Agroecosystems of West and Central Africa Code Land-Use Class Description Designated Color Derived Vegetation Indices Upland 1 significant farmlands gray 2 scattered farmlands seafoam 3 insignificant farmlands violet 4 wetland/marshland mocha 5 dense forest rose 6 very dense forest red-orange Valley Fringe 7 significant farmlands white 8 scattered farmlands pine-green 9 insignificant farmlands red Valley Bottom 10 significant farmlands cyan 11 scattered farmlands yellow 12 insignificant farmlands magenta Others 13 water blue 14 built-up area/settlements tan 15 roads navy 16 barren land or desert land sand Table 8.3 Land-Cover Types Identified in Level-II Characterization of IV Agroecosystems of West and Central Africa Code Land-Cover Type Description Code Land-Cover Type Description 1 water 6 barren farms 2 tree 7 barren lands 3 shrub 8 built-up area/settlement 4 grass 9 roads 5 cultivated farms 10 others © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing [...]... Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 90 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT Table 8. 6 Distribution of Land-Use Classes in the Different Agroecological and Soil Zones of the Save (Republic of Benin) Study Areaa AEZ 2 Land-Use Class 1 2 3 4 5 6 7 8 9 Uplands significant... Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 92 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT Table 8. 8 Land-Use Distribution in the Gagnoa (Côte d’Ivoire) Study Areaa Full Study Area No Land-Use Category Area (ha) % of Total Study Area 1 2 3 4 5 6 Uplands significant farmlands scattered farmlands savanna vegetationb wetlands/marshland dense... settlements and major road networks (areas around Gagnoa and Guibéroua in Figure 8. 4) The forest-cropland mosaic land-use classes (2, 8, and 11) contain a significantly higher percentage of farmlands (land-cover types 5 and 6) and a lower percentage of trees and shrubs (landcover types 2 and 3), but they still have a significant amount of forest vegetation The overall area under forest-cropland mosaic... and Bobo-Dioulasso (Burkina Faso) Study Area AEZ 2 AEZ 7 Area (ha) % of Total Area 8. 3 36.4 16.3 0.9 3.1 0.3 280 ,990 901,130 727, 987 85 ,202 79,955 18, 175 9.0 28. 7 23.2 2.7 2.6 0.6 17, 789 52,260 29,363 4.6 13.4 7.5 104 ,87 9 282 ,455 254, 088 3.3 9.0 8. 1 0.5 4.2 4.4 4 ,86 3 16,792 8, 726 1.2 4.3 2.2 21729 139,7 58 107,519 0.7 4.5 3.4 0.7 0.0 0.2 2.5 544 82 4 98 4,600 0.1 0.0 0.1 1.2 12,570 5 ,84 9 6, 585 106, 985 ... Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 88 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT color section)] AEZ 2 is representative of 18. 4 million ha and AEZ 7 of 24.9 million ha in West and Central Africa (see spatial distribution of these zones in Figure 8. 6 in the color section and Table 8. 1) Both these agroecological and soil... farmlands scattered farmlands savanna vegetationb wetlands/marshland dense forest very dense forest Valley fringes significant farmlands scattered farmlands insignificant farmlandsc AEZ 7 Entire Study Area Area (ha) % of Total AEZ 2 Area (ha) % of Total AEZ 7 Area (ha) % of Total Area 57 ,85 3 547,694 233,120 129 ,80 8 126,096 30,4 98 3 .8 35.7 15.2 8. 5 8. 2 2.0 1 48, 609 492,119 155,091 78, 665 63 ,87 6 28, 445... is, inland valleys with potential for dry season cropping) At the time of groundtruthing 69% of the inland valleys were wet, 21% were moist, and 10% were dry A mapping of © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 © American Society for Photogrammetry and Remote Sensing 96 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT. .. (Table 8. 8) The total area covered by humid forest vegetation with insignificant farmlands (land-use classes 5, 6, 9, and 12) was 58. 3%, whereas the area with humid forest-cropland mosaic (land-use classes 2, 8, and 11) was 23.0% of the total geographic area The intensity of cultivation was significantly higher for valley bottoms (20.6%) compared with valley fringes (16.9%), and uplands (15%) (Table 8. 7)... third-order inland valley streams, and about 0.5 degrees for the fourth-order inland valley streams The study showed a strong relationship between upland cultivation and inland valley cultivation (Table 8. 9) proving one of the hypotheses of this study SUMMARY This study used high-resolution satellite images from Landsat TM and SPOT HRV in three separate study areas [Figure 8. 1, 8. 2, 8. 3 and 8. 6 (see color... bottoms) for each study area (Tables 8. 6, 8. 8, and 8. 10; and Figures 8. 15, 8. 16 and 8. 17 in the color section) The toposequence oriented land-use mapping has been a unique feature of this study It is indeed clear from this study that the inland valley bottoms that constituted © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter 14 . 192 and row 54 and path 197 and row 52, [(see Figure 8. 1and 8. 2; and Table 8. 1 for details and spatial location)], and one SPOT HRV image (Fig- ure 8. 3, 8. 6 in the color section and Table 8. 1)]. fringes, and up- 88 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter. of techniques for: 82 GIS FOR WATER RESOURCES AND WATERSHED MANAGEMENT © 2003 Taylor & Francis Chapters 1, 3, 5 & 6 © American Water Resources Association; Chapter 13 © Elsevier Science; Chapter