The core application of AEZ, leading to an assessment of land suitability and productive potential under specified uses, comprises three groups of compound activities:
· inventory of land utilization types and their ecological requirements;
· definition and mapping of agro-ecological zones based on inventories of land resources (including climate, landform and soils);
· evaluation of the land suitability of each agro-ecological zone.
Figure 5 illustrates the relationship of these activities and their component procedures. The final and intermediate outputs can then be applied in a series of more advanced applications which are determined according to the objectives of the study. The present chapter describes how to apply the procedures for the core AEZ application, leading to assessment of land suitability and potential productivity with particular reference to crop-based production systems. Following this description, brief summaries are presented of the following advanced applications:
· land productivity assessment;
· extent of potential arable land;
· land use optimization.
Procedures are described in a step-wise manner, together with the data input requirements and the intermediate output results, and are illustrated by examples drawn from AEZ studies undertaken by FAO. Emphasis is placed on providing the user with an understanding of the procedures so that they can be implemented or adapted according to the objectives of the specific AEZ study and the resources available. Strictly speaking, a computer is not necessary to carry out any of the above procedures (excepting those involved with objective decision making). However, it is assumed that most users will have access to commercially-available database and spreadsheet software, and be familiar with its use. Dedicated software tools, which are available for various groups of procedures, and linkages with GIS, are described in Chapter 4.
The information contained in the land resources inventory is determined to a large extent by the requirements of the land utilization types and their component crops. The procedures for inventory of land utilization types are therefore described first, although the relationship of land-use requirements with the land characteristics contained in the land resource inventory should be noted.
A range of LUTs should be selected to reflect current land use and/or land use under a projected improved situation. All subsequent assessments of land suitability and potential productivity carried out as part of the AEZ study will refer to these specific LUTs as practised in defined agro-ecological zones or cells.
LUTs are defined in terms of a product, or a specified range of products, and the management system, including the operations and inputs, used to produce these products. The socio-economic setting is also usually included in the definition. The level of detail to which LUTs are defined is principally determined by the objectives of the study and the data needs of the land suitability assessment. Most AEZ studies have separated LUTs on the basis of crops, or ranges of crops, and level of inputs, as shown in Table 4. Currently available databases, such as the Land Use Database (de Bie, van Leeuwen and Zuidema, 1995) enable a more quantitative characterization of inputs, operations and outputs.
The following factors should be implicit in LUT definition:
· The description of an existing or anticipated agricultural production system in terms of products, production techniques, and expected type and range of inputs and outputs.
· The identification of the important factors which affect the production potential, such as limits to mechanization on sloping lands, and soil requirements for irrigation.
· The production scenarios to be modelled and the level to which production constraints are assumed to be overcome in each scenario.
· The quantification of input levels (labour, materials, capital, etc.) associated with various production scenarios. This is used for:
· estimation of the likely levels of input which correspond to the anticipated outputs;
· estimation of total input demands in relation to actual or anticipated resource availability at country/province level.
Following the definition of LUTs, the next steps involve the inventory of their requirements in relation to the climatic, soil and landform conditions necessary for the component crops and for the management system. These inventories form the basis of a sequential assessment of climatic suitability, edaphic suitability and potential yield calculation.
A crop climatic inventory is compiled based on crop phenological requirements, thermal ranges and photosynthetic characteristics.
An example of the crop attributes necessary for determination of climatic suitability is given in Table 5. Requirements for day length should also normally be included, but the cultivars considered in this particular case are all day neutral. Similar information regarding other crops is given in FAO (1978) and Kassam (1980).
There are often considerable differences in such factors as the length of crop growth cycle, which are mainly due to the adaptation of different cultivars to different ranges of thermal conditions. Several crop ecotypes are distinguished under days to maturity in Table 5. These ecotypes are treated separately for evaluation of land suitability and potential performance.
Crops should be arranged into climatic adaptability groups based on similar abilities to photosynthesize, assuming their phenological requirements are met. Table 6 summarizes the main characteristics of each group, and gives examples biomass productivity (Step 3.2.1, p. 41).
The agricultural exploitation of the climatic potential of crops depends on the properties of soil, and on how the soil is managed. Constraints imposed by landform or by other features of the land surface, such as susceptibility to flooding, must also be taken into account.
Many soils are a result of climatic action and, as a result, climate and soil in many instances have relationships which may have a mutual enhancing effect on crop productivity. The close interrelation of climate and (zonal) soil and natural plant community, to some extent, aids assessments of land suitability.
The basic soil requirements of crop plants may be summarized under the following headings, related to internal and external soil properties.
· Internal requirements:
· soil temperature regime;
· soil moisture regime;
· soil aeration regime;
· natural soil fertility regime;
· effective soil depth;
· soil texture and stoniness;
· soil toxicity;
· other specific properties, e.g. soil tilth.
· External requirements:
· occurrence and depth of flooding;
· soil accessibility and trafficability.
From the basic soil requirements of crops, ranges of optimal and marginal conditions can be defined. These are subsequently used for matching with relevant land characteristics in the determination of crop edaphic adaptability inventory is presented in Table 7. Detailed complementary information can be found in numerous FAO publications (FAO, 1976; 1978; 1981; 1983; 1985; 1994a).
Important note: Information on optimal and marginal ranges of edaphic conditions for certain crops, such as that presented in Table 7, may be unavailable or difficult to obtain. In the absence of published information, an educated guess must be made based on parallels with other crop species with similar physiological requirements. These “guesstimates” are important as the models which match crop requirements with soil and climatic characteristics do not allow for missing data. When more reliable local data are obtained, the databases should be updated and the assessment re-run.
This compound activity comprises the following steps:
2.1 analyse length of growing period (LGP);
2.2 define thermal zones;
2.3 compile climatic resource inventory;
2.4 compile soil and landform resource inventory;
2.5 compile present land use inventory;
2.6 combine above to make land resources inventory based on agro-ecological zones or agro-ecological cells. This inventory also normally includes information on administrative boundaries.
The land resources inventory is based on combining different layers of information to define agro-ecological cells (AECs) with a unique combination of climate, soil and other related land attributes (Figure 5, p.18). Such overlay techniques are most conveniently carried out in a GIS environment. However, alternative methods can be used if a GIS is not available (see step 2.6, p. 33).
Box 4 summarizes the data required to prepare the climatic resources inventory
BOX 4 CLIMATIC DATA REQUIREMENTS
Dataset 1 : Maps
Dataset 2 : For each climatic station
Note The time period over which the data is collected depends on the purpose and level of detail of the AEZ study. Where possible, rainfall should be collected for a historical sequence of years
The growing period is the period of the year when both moisture and temperature conditions are favourable for crop growth (Box 3, p. 13).
In the tropics, where temperature is rarely a limiting factor except at very high altitudes, I GP can be assessed by a simple moisture balance of precipitation (P) and potential evapotranspiration (PET). LGP should be assessed for all valid rainfall stations in the study area with a minimum of 20 years of complete records. Where the synoptic data required for PET calculation are not available, PET can be assessed through locally validated correlations with altitude (e.g. De Pauw, 1987), or, in flatter areas, by linear interpolation from surrounding stations (e.g. Schalk, 1990; Radcliffe, Tersteeg and De Wit, 1992).
Although the original FAO AEZ study at continental scale based LGP calculations on average monthly rainfall and PET data, more detailed studies (e.g. Radcliffe,1981; De Pauw,1987; FAO,1993a) have recognized the value of analysing historical rainfall records and using the results as a basis for statistical analysis of LGP distribution. The approach based on historical data is highly recommended, particularly in semi-arid areas where inter-annual variations in rainfall and resulting LGP are often extreme (FAO, 1993a; Radcliffe, 1993).
Table 8 gives a simple example of LGP calculation over 11 years at Nazreth, Ethiopia, which has a single growing period in most years, determined by moisture availability. This example is taken from a manual intended for field staff who do not necessarily have access to computer facilities. Continuous periods of at least two months when P > PET/2 are taken as the intermediate plus humid phases of the growing period (Figure 2, p. 8). Fifteen days are counted for the first month when rainfall exceeds PET/2, and 30 days are added for each succeeding month with P > PET/2. A further 20 days are added to comprise the soil moisture reserve period.
Statistical analysis of the LGPs in Table 8 gives a dependable growing period, exceeded in 75% of years, of 95 days. The median LGP, exceeded in 50% of years is also 95 days. Assessment of LGP based on average rainfall data gives a value of 155 days, considerably overestimating the actual situation.
The Kenya AEZ study (FAO,1993a) compared PET and moisture balance for historical rainfall records in a way which is similar in principle to the example in Table 8. The computer facilities used in this study enabled a much more detailed analysis of component LGP periods, based on a shorter time period (three days), which is particularly important in areas with multiple growing periods. Based on this analysis, 22 LGP pattern zones, with specified frequencies of occurrence of one, two, three and four growing periods per year (and also of all dry and all humid years), were recognized. These LGP pattern zones are illustrated in Table 2 (p. 11).
The climatic resources inventory listed each individual occurrence of humid, intermediate and dry period and derived statistical correlations, firstly between total lengths of growing period in years with the dominant pattern and years with the associated pattern, and secondly between the lengths of individual component growing periods and the total LGP in years with multiple growing periods. Individual growing periods and total LGP in any one year are used to evaluate the climatic suitability of annual crops and perennial crops respectively (Step 3.2, p. 38).
In temperate regions, temperature is often of equal or greater importance to moisture availability as a determinant of crop growth, and its influence is not adequately catered for by the original AEZ methodology (FAO, 1978). Apart from requiring a more detailed specification of thermal regime (Step 2.2), temperature interacts with moisture availability in determining LGP. Particular modifications to the LGP model developed to deal with the temperate conditions in China are shown in Box 5.
BOX 5: LGP ASSESSMENT FOR TEMPERATE REGIONS
THE CHlNESE EXAMPLE
The AEE study in China identifies four components of a temperature related moisture balance in determining LGP:
Source Zheng Zhenyuan, 1994
The China study demonstrates the necessity for adapting elements of the AEZ methodology when it is applied in a different range of environments from those in which it was developed. Modifications to the moisture balance model, however, go beyond what is required to account for the low seasonal tempera- tures, and some of these have potentially broader application. The use of crop coefficients, albeit in a rather generalized way, represents a step towards a more accurate and crop-specific moisture balance modelling, which is a significant development on existing AEZ methodology.
Growing period zones are plotted on a map, and may be based on fixed intervals of mean LGP, or on the dependable LGP exceeded at a given level of probability (0.75 or 0.8). Figure 7 gives an example of growing period zones in Bangladesh (Brammer et al., 1988).
Thermal zones describe the temperature regime available for crop growth during the growing period. They are usually defined based on ranges of mean temperature. In tropical highland areas, mean temperature is usually strongly correlated with altitude. Table 9 gives the mean temperature ranges and corresponding altitude for reference thermal zones in Kenya.
Reference thermal zones in Kenya
|Thermal zone code||Mean daily temperature range (°C)||Altitude range (masl)|
Such a simplistic treatment of thermal regime may be inadequate in temperate regions. The AEZ study in China (Zheng Zhenyuan,1994) uses a combination of the duration of time and the accumulated degree days above several critical temperature thresholds, and the mean monthly temperatures in January and July in the definition of thermal zones (Table 10).
A recent revision of the thermal regime concepts has led to the following definitions:
· Thermal growing period zones (LGPt)
Period during the year when Tmean $ 5°C. This period is inventoried at 30-day intervals. The winter break (Tmean < 5°C) is defined as (i) dormancy period when hibernating crops can survive, or (ii) cold break when killing tempera- tures for hibernating crops occur (killing temperatures are adjusted according to depth of snow cover (the killing temperature model variable is set at -8°C for 0 cm snow cover, is increasing to -22°C for snow cover heights of 65 cm or more and should not exceed a total duration of 200 days).
· Frost-free period zones
The frost-free period is assumed to coincide with the period Tmean > lO°C. This period is also inventoried at 30-day intervals.
· Reference permafrost zones
The reference permafrost zones refer to climatic conditions assumed to be conducive to the formation and maintenance of permafrost. As an approxima- tion for reference permafrost zones, Tmean < -5°C is assumed for areas with potentially continuous permafrost and Tmean ranging from 0 to -5°C with potentially discontinuous (intermittent) permafrost.
The inventory of climatic resources is prepared as follows:
· plot the individual station data of temperature, LGP-pattern and mean total dominant LGP derived as described above onto a map;
· construct boundaries of thermal zones, LGP pattern zones, growing period zones and isolines of mean total dominant LGPs.
In addition to normal extrapolation techniques, extensive use is usually made of Landsat images, climatic maps, vegetation maps, land-use maps, topographic maps, and soil maps to guide the delineation of boundaries and isolines. If a GIS is used, the inventory maps should be subsequently digitized. Given the necessary base maps, point data and knowledge on the interpolation of climatic variables between these points, the user can prepare climatic maps in the GIS environment.
Information on soil type and landform is normally derived from existing soil maps, legends and reports. National soil maps at a scale of 1:1 000 000 or larger are excellent sources from which the required input data can be derived. At more detailed levels of investigation, provincial soil maps may be used or additional data may have to be collected. For purposes of correlation, soils should preferably be classified in the FAO-Unesco Soil Map of the World classification system (FAO,1974; FAO,1990b) although national classification systems can also be used provided the essential characteristics needed for evaluation are included in soil type definitions.
On small-scale maps, the mapping unit consists generally of associations of individual soil types occurring within the limits of a mappable physiographic unit (Figure 4, p. 14). The mapping units reflect as precisely as possible the soil pattern of large regions. The information available for each soil type should include those parameters required for matching with land-use requirements. Although it is possible to define a minimum data set necessary for virtually all applications, the range of parameters required may vary according to the geographical region and the level of detail of the investigation. For example, it may be necessary to include such factors as exchangeable aluminium in soil type characterization in humid tropical regions, whereas other factors, such as soluble salt concentrations, are usually more important in arid areas. Box 6 lists the soil parameters required for most AEZ studies.
Soil phases Soil phases indicate land characteristics which are not considered in the definition of the soil units but are significant to the use and management of land. Soil phases are defined in the FAO-Unesco Legend (FAO,1974; 1990b) and can be grouped as follows:
BOX 6 : SOIL DATA REQUIREMENTS
Data Set 1 : Maps
Data Set 2 : For each soil/landform mapping unit
· indicating a mechanical hindrance or limitation
· Rocky, bouldery, stony, gravelly;
· indicating an effective soil depth limitation
· Lithic, paralithic, petrocalcic, petroferric;
· indicating a physico-chemical limitation
· Saline, sodic.
The mapping unit composition table shows the distribution of soil types, and of their key properties, within each soil mapping unit. An example is given in Table 11.
Present land use and land cover are particularly important when the results of AEZ are applied to land use planning. Classes of land use and land cover should therefore be systematically recorded during the land resource inventory, and can be regarded as attributes of AECs. This inventory is quite distinct from the inventory of land use types (Compound Activity 1), which defines potential land use and lists its requirements for land evaluation.
Compile land resources inventory The land resource inventory is the result of overlaying of thermal zones, LGP zones and soil resources inventories. Additional information on administrative boundaries, land use and other constraints, such as tsetse fly incidence, may also be overlaid as shown in the example in Figure 8. The output of this procedure is a number of agro-ecological cells: approximately 91 000 were defined in the Kenya AEZ study. Table 12 gives an example of land resource mapping units, soil mapping units and AECs in such a land resource inventory.
For the overlay of such large amounts of information a GIS is strongly recommended. If a GIS is not available, however, it is sometimes possible to assign information from one inventory (e.g. climate) to mapping units defined in a separate inventory (e.g. soils), and to use the boundaries of these mapping units as the sole spatial framework for the land resource inventory. For example, the national land suitability assessment of Botswana (Radcliffe et al, 1992) used the 1:1000 000 national soil map (De Wit and Nachtergaele,1990) to define the spatial distribution of units to be evaluated. The boundaries of, these units had been determined by satellite image interpretation and extensive field work and were relatively reliable. The boundaries between climatic zones, based on data collected from a number of reference stations, were not reliable, and in the relatively flat terrain of Botswana, no relationship between altitudinal and climatic factors could be established. Rather than attempting to overlay unreliable climatic boundaries over reliable soil boundaries, each soil mapping unit was assigned a set of climatic information which was used as an input to the land suitability evaluation. This procedure led to 846 land suitability units, which are analogous to AECs.
Even if a GIS is used, digitization of data from different sources may lead to poor coordination of boundaries, and a number of land mapping units may result which do not actually occur in practice. Such problems were encountered in mountainous areas of China (Zheng Zhenyuan,1994), where it was decided to adjust soil association boundaries to boundaries of climatic zones (essentially the reverse of the procedure used in Botswana where climatic zone boundaries were defined by soil mapping units).
Irrespective of whether a random overlay technique is used or whether a single map is used as a spatial framework for the land resources inventory, the AECs must be precisely defined in terms of their land and climatic features. Typical outputs of the land resources inventory are:
total extents of all soil units, broken down by texture class, slope class and phase as they occur in each thermal zone, in each pattern of growing period zone on a country/province basis;
· a tabulated summary of the inventory showing the distribution of individual soil units (combined for all slopes, textures and phases) by length of growing period zone (combined for all thermal zones and pattern of growing period zones);
· a tabulated summary showing the distribution of individual soil units (combined for all slopes, textures and phases) by length of growing period zones for each thermal zone (combined for all pattern of growing period zones);
· a tabulated summary showing the distribution of individual soil units by texture, slope, phase and by length of growing period zones for each thermal zone and each pattern of growing period zone;
· maps and tabulated information on agro-ecological zones.
Assessment of land suitability is carried out by a combination of matching constraints with crop requirements, and by modelling of potential biomass production and yield under constraint free conditions. This activity is normally carried out in two main stages, in which firstly the agro-climatic suitability is assessed, and secondly the suitability classes are adjusted according to edaphic or soil constraints. Each stage comprises a number of steps which are listed as follows:
3.1 Matching the attributes of temperature regimes to crop requirements for photosynthesis and phenology as reflected by the crop groups, to determine which crops qualify for further consideration in the evaluation.
3.2 Computation of constraint-free yields of all the qualifying crops taking account of the prevailing temperature and radiation regimes in each LGP zone.
3.3 Computation of agronomically attainable yields by estimating yield reductions due to agro-climatic constraints of moisture stress, pests and diseases, and workability for each crop in each length of growing period zone.
3.4 Comparison of the soil requirements of crops with the soil conditions of the soil units described in the soil inventory, at different levels of inputs.
3.5 Modification of the soil unit evaluation by limitations imposed by slope, texture and phase conditions.
Apart from step 3.2, which involves a mechanistic model of biomass production and crop yield, all the above procedures involve the application of rules which are based on the underlying assumptions which relate land suitability classes to each other, and to estimates of potential yields under different input levels. Many of these rules were derived from expert knowledge available when the first FAO AEZ study was undertaken (FAO, 1978), and they should be regarded as flexible rather than rigid. The number of suitability classes, the definition of management and input levels, and the relationships between them can be modified according to increasing availability of information and the scope and objectives of each particular AEZ investigation. Box 7 gives an example of rules applied in the Kenya AEZ study.
The initial step in the matching process is comparison of the temperature requirements of individual crops with the identified thermal zones of the climatic resource inventory. This step is essentially a screening exercise which excludes crops which are unsuitable in the specified temperature regimes from further analysis.
An example of matching crop temperature requirements with thermal zone is presented in Table 13. Where requirements are fully met, the zone is rated S1, where requirements are sub-optimal, the zone is rated either S2, S3 or S4, and where the requirements are not met, the zone is rated N (not suitable). Expected yield reductions resulting from sub-optimal conditions are given in Box 7.
Matching of crops to growing period zones is according to the following steps:
3.2.1 computation of net biomass and constraint-free crop yield by individual lengths of growing period zones;
3.2.2 inventory of agro-climatic constraints for each length of growing period zone by crop and by input level;
3.2.3 application of the agro-climatic constraints to the constraint-free yields to determine agro-climatically attainable crop yields by individual lengths of growing period zones;
3.2.4 computation of agro-climatically attainable crop yields as affected by year-to-year variability in moisture conditions;
3.2.5 agro-climatic suitability classification of each mean total dominant growing period zone (inventoried) for each crop according to agro-climatically attainable yields by thermal zones and by pattern of growing period zone.
FAO AEZ studies have derived figures on potential maximum biomass and crop yield by using a model, the essential features of which are:
1. calculation of gross dry matter production for standard crop;
2. application of correction factor for crop species and temperature;
3. application of correction factor for crop development over time and leaf area;
4. application of correction factor for net dry matter production;
5. application of correction factor for harvested part.
The detailed application of the biomass and yield model is described by Kassam (1977) and FAO (1978). The model is also included in the Agricultural Planning Toolkit (APT) and the AEZ country study (AEZCCS) software developed by FAO (FAO, 1990a; Fischer and Antoine, 1994).
Potential maximum biomass and yield are calculated for all annual crops rated as at least marginally suitable (based on thermal zone) for each individual length of growing period in defined LGP zones. In areas with significant altitudinal variation, the increasing length of crop growth cycle associated with cooler temperatures needs to be accounted for in the assessment. Perennial crops are assessed on the basis of total growing period in areas with more than one LGP per year.
Table 14 gives an example of constraint-free yields based on the effect of the prevailing temperature and radiation regimes on crop photosynthesis and growth within the lengths of growing periods.
Some recent AEZ studies carried out in Asia (FAO, 1994a) have indicated discrepancies between potential maximum yields calculated by the standard AEZ model and best yields achieved on research stations and even on farmers’ fields. In some cases this could be attributed to recent advances in plant breeding, particularly of paddy rice, which have made some of the originally published input parameters to the model redundant. Other discrepancies may simply be the result of knowledge gaps in the actual physiological responses of certain crops to environmental variables. In China, maximum yield figures of wheat, maize, rice and soybean obtained from agricultural research sites were used in preference to those calculated by the biomass yield model (Zheng Zhenyuan, 1994).
In the agro-climatic suitability assessment, yield losses likely to occur due to agro-climatic constraints must be taken into account. Yield losses in a rainfed crop due to agro-climatic constraints are governed by the following major conditions:
· How well the length of the normal growth cycle of the crop in question fits into the available length of the growing period.
· The degree of water stress during the growing period.
· The yield and quality reducing factors of pests, disease and weeds.
· The climatic factors, operating directly or indirectly, that reduce yield and quality of produce mainly through their effects on yield components and their formation.
· Climatic factors which affect the efficiency of farming operations and the cost of production.
All these agro-climatic constraints can be rearranged into a set of four, as follows:
· Constraints resulting from moisture stress during the growing period (e.g. unreliability of rainfall).
· Constraints due to pests, diseases and weeds, directly affecting the physical growth of the crop (e.g. stem-borers, leaf blights and virus diseases).
· Constraints due to various factors affecting yield formation and quality (e.g. cotton stainers, pod borers and silk drying).
· Constraints arising from difficulties of workability and produce handling (e.g. excessive wetness of the land or the produce).
The severity of the four groups of constraints, by crop, length of growing period zone and level of inputs can be presented in a table form as shown in the example in Table 15.
Ratings of 0,1 and 2 correspond to nil, moderate and severe constraints respectively. The agro-climatic constraint-free yields are reduced according to acting constraints in accordance with the rules in Box 6.
This step is only carried out if the LGP has been assessed for individual years. Anticipated yields of annual crops are computed for each crop by each individual component LGP in each thermal zone for each level of inputs.
Each AEC is evaluated with respect to LGP pattern by taking into account all the constituent component lengths of LGP in each pattern. As the frequency of occurrence of numbers of LGPs within LGP patterns is known (Table 2, p.11), a profile of the variability in potential yields over time is constructed. Yields can then be expressed in terms of averages, maxima and minima.
Perennial crops are matched to total LGP, with potential yields being downgraded for LGPs which indicate moisture stress. For example, in total LGPs which include an occurrence of intermediate lengths of growing period in their make-up, yield losses due to such occurrences can be quantified according to yield reduction rules (Box 7).
The results of the above-described computations are the attainable yields for each crop by each mean total length of growing period zone by each pattern of growing period zone and by each thermal zone. These attainable yields form the basis of the agro-climatic suitability classification presented below.
Classes of agro-climatic suitability are derived by relating the agro-climatic yields (reduced according to the constraints in Table 15) to the potential maximum yield determined from radiation and temperature considerations. Normally between four and six classes of suitability are defined based on different ranges of attainable yield relative to the potential maximum. Rules, such as those in Box 7, are used to establish the limits between suitability classes. Table 16 gives a diagrammatic presentation of potential yields and agro-climatic suitability classes associated with different LGP zones.
The soil unit evaluation is expressed in terms of ratings based on how far the properties of a soil type meet crop requirements under specified level of inputs. Ratings may be made in five basic classes for each crop and level of input, i.e., very suitable (S1), suitable (S2), moderately suitable (S3), marginally suitable (S4), and not suitable (N). These ratings correspond to percentage reductions in potential maximum yield as indicated in Box 6.
Table 7 (p. 24) gives some examples of optimal and marginal ranges of crop edaphic requirements. Suitability ratings are assigned to each combination of crop and soil type by comparing such ranges with the characteristics listed in the soil inventory. Soil type ratings should be based on as much local expertise and knowledge as possible, and site-specific conditions not necessarily reflected in the soil type nomenclature should be taken into account. As an example, soil ratings for selected crops at two levels of input are given in Table 17. These ratings may be further modified according to limitations of soil texture, phase or slope.
Modify c/asses based on texture and phase limitations and slope Limitations imposed by soil texture and phase should be evaluated based on local expertise or expert knowledge. Appropriate rules should be drawn up to account for any additional constraints due to coarse textures or particular phases. An example of such a rule is given in Box 6 (p. 33).
Limitations imposed by slope affect both ease of cultivation and susceptibility to erosion. Table 18 gives an example of slope limits for various cultivation types at specified input levels.
Slope limits (%) for land use types
Land utilization type
Level of Inputs
|Dryland crops without soil conservation measures||
|Dryland crops with soil conservation measures||
|Wetland crops without soil conservation measures||
|Wetland crops with soil conservation measures(terracing)||
|Coffee, tea, fuelwood and pasture, with and without soil conservation measures||
Source: FAO ( 1993a).
If a land utilization type is matched to a land unit with a slope greater than the above limits, the land suitability is rated as N, not suitable.
If sufficiently detailed information is available, projected soil erosion loss can be calculated and related to decreases in productivity. This is regarded as an advanced application of AEZ, which has been developed during the Kenya study. The model is described in outline in the section describing land productivity (p. 51). More detailed accounts of approaches and methodology can be found in Mitchell (1984), Stocking (1984) and FAO (1993a).
The basis of these advanced AEZ applications is a set of GIS-based AEZ land resource inventories of individual districts in Kenya. The AEZ land resource inventories combine digitized map overlays that relate to climatic conditions, soil inventory, administrative units and selected properties of present land use, i.e. cash crop zones, forest areas, irrigation schemes, tsetse infestation areas and game parks. The digitized data were converted to a grid cell or raster database. Each pixel represents one square kilometre (100 ha) (Figure 8). AEZ computer programs are applied to the district land inventories to analyse land suitability. This application builds on the Kenya land productivity assessment which includes cropping patterns, linkage to livestock and forestry production systems and soil erosion considerations. A land productivity database is generated which contains quantified information on the productivity of all feasible land utilization types for each agro-ecological cell in the districts. The land productivity assessment involves 64 types of food and cash crops, pastures, 31 fuelwood species and nine livestock systems which are grouped into 26 production commodities, including 26 crop and ten livestock production commodities. This database provides the input to the Optimal Spatial Resource Allocation Model. It has been developed for integrating crop, livestock and fuelwood production within the framework of AEZ land productivity assessment and its application to various land-use planning scenarios at national and district levels. The model accepts user-specifiable scenario parameters from a control file, reads crop, grassland and fuelwood production potentials by agro- ecological cells from the land productivity database, reads livestock system related data derived from herd structure models, and determines simultaneously land use by agro-ecological cell as well as supported levels of different livestock systems, feed supplies and utilization by livestock zone and season. The model provides a framework for specifying different types of objective functions and kinds of constraints.
The planning scenarios are specified by selecting and quantifying objectives and various constraints related to aspects such as demand preferences, production targets, nutritional requirements, input constraints, cash flow constraints, feed balances, crop-mix constraints and tolerable environmental impacts. Given the potentially large number of agro-ecological cells and number of activities to be taken into consideration, standard linear programming techniques have been used to analyse the multitude of possible solutions and select optimal ones. For instance the linear programming techniques have been used in order to examine alternative regional or district level land-use patterns. Such models suggest feasible land-use allocation patterns that best satisfy specified single develop- ment objectives e.g, target food consumption patterns, population supporting capacities or rural employment levels. One typical application is the determina- tion of potential supporting capacities using various scenarios within defined single or multiple objectives.
The results of the land suitability assessment are a set of land suitability classes for crops grown on different land units or AECs with specified level of inputs. Each land suitability class for each crop under each input level reflects a range of anticipated yields. Knowing the area of each AEC or land unit, estimates of production can be drawn up for more broadly defined agro-ecological zones, or, provided administrative boundaries can be related to AEC or land unit boundaries, by province or’ district. Table 19 gives an example of areas suitable for cultivation of specified crops in Chanthaburi Province, Thailand.
A number of advanced applications of AEZ can be developed from the results of land suitability assessment. These applications are based on sets of rules derived from basic assumptions on the interaction of product yield with the agro-environment, and on the management and conservation requirements of production systems. A conceptually similar set of rules employed in the core application of AEZ is given in Box 7 (p. 38). It must always be borne in mind that rules based on current expert knowledge should be regularly reviewed and updated as more information becomes available.
The need for further analysis of the results on land suitability is determined by the goals and objectives of the AEZ study. The availability of expert knowledge and the reliability of the assumptions on which the analysis is based should be taken into account in applying the results in planning and policy making.
The most extensive set of advanced AEZ applications developed to date is that resulting from the FAO study in Kenya (FAO, 1993a), the prime objective of which was to support land use planning and decision making at district level. Meeting this objective required assessment of the yields and potential productivity of diverse production systems (involving crops, livestock and fuelwood) and the construction of a model to optimize land use, allowing for trade-offs between the benefits of competing production systems. Figure 9 illustrates the overall model used in the Kenya study. Advanced applications which comprise components of this model are described below.
Land suitability assessment enables the selection of single crops to be made for each AEC or land unit according to their yield potential in particular cells. The land productivity is a measure of the potential total annual productivity calculated by fitting the most suitable crops to the available lengths of growing period. Determination of land productivity requires the following steps:
4.1 formulation and quantification of the cropping pattern options;
4.2 formulation and quantification of crop rotations;
4.3 assessment of the impact of soil erosion on productivity.
Under favourable climatic conditions, increased land productivity can be achieved through multiple cropping. Crops may be grown either sequentially or in mixtures, as defined in Box 8. Sequential cropping is only possible when the available growing period (either single or multiple) extends beyond the duration of the growth cycle of a single crop.
In the frost-free areas in Kenya, the restriction to sequential cropping is one of availability of soil moisture. In the areas with a longer growing period, as in the moist sub-humid (growing period 210-270 days) and humid ( > 270 days) areas, crop growth is possible throughout much of the year. It is in such areas that a strong association with sequential cropping emerges, and sequential crops in both monoculture and multiculture are involved (Table 20). However, because of the cool temperatures in thermal zones T6 and T7 (Table 9, p. 30) sequential cropping is of minor importance because the annual crops that are adapted to the prevailing conditions are generally slow to reach maturity.
In areas with LGP < 120 days, sole cropping of short duration annual crops is dominant in all thermal zones. Some simultaneous cropping is practised with crops with similar maturation periods, but its status in thermal zones T1, T2,T3, T4 and TS is a minor one. In thermal zones T6 and T7, growing conditions only permit a moderate to marginal production from sole cropping of single crops.
In areas with LGPs between 120 and 210 days, crop mixtures, including those involving crops of different maturation periods, are common in thermal zones T1, T2, T3, T4 and T5. Because of the cool temperatures in T6 and T7, crop mixtures involving crops of similar maturation periods are common.
In areas with LGPs > 270 days, crop mixtures, especially those involving crops with different maturation periods, are common. In such areas, the slow-growing and later-maturing components generally tend to mature under better end-of sea- son moisture conditions. In these areas, multiple cropping, both simultaneous and sequential, is practised.
Cropping pattern options are formulated in three steps as follows:
i. fit crop growth cycles into prevailing component LGPs for each AEC;
ii. incorporate the turn-around time between crops, within sequential cropping patterns, needed to harvest the first crop, prepare the land and sow the subsequent crop;
iii. decide for which crops and levels of inputs intercropping is acceptable.
In the model as applied to Kenya, intercropping was considered only at the low and intermediate input levels for all crops except wetland rice, sugar cane, banana and oil palm.
This is done by taking account of the restrictions of space and time, and the fallow requirements, of the selected annual cropping pattern options. Restric- tions are imposed by agro-ecological conditions. For example, only mono- cropping is possible in the semi-arid areas.
The fallow requirement is calculated on the basis of maintenance of humus levels (for details see FAO, 1993a; Annex 4, p. 28). This fallow requirement, expressed as the percentage of time the land is under fallow as opposed to cropping, is built into the cropping patterns. At intermediate input levels, when some fertilizer is assumed to be used, fallow requirements are 33% of those at low input level. With high inputs, fallow requirements are 10% of those at low input levels (specific rules apply to Fluvisols and Gleysols).
In the Kenya study, the basic length of fallow period was taken as that needed for LGPs between 120 and 269 days. For LGPs) 270 days the reference fallow period is 50% greater than the basic, due to additional problems with weeds, pests and diseases, and leaching and erosion. Similarly, for LGP 90-119 days, fallow requirements are greater than the basic by 25% due to additional problems with fallow establishment from dry conditions, and degradation hazards, and for LGP 60-89 days, 50% greater due to problems with fallow establishment, degradation hazards and the need to conserve moisture.
Crop rotation options are formulated for each agro-ecological cell for each cropping pattern option generated. This is accomplished in two steps. Firstly the appropriate crop combination restrictions are applied to rule out risky or undesired crop combinations on space or time grounds, and secondly to incorporate the appropriate fallow requirements for each suitable cropping pattern.
With cropping patterns comprising more than one crop, average fallow requirements for the crops concerned are applied to define the rotations.
The impact of soil erosion on productivity is assessed in three stages. Firstly the potential soil erosion is calculated using a modified Universal Soil Loss Equation (USLE), which takes account of rainfall erosivity, soil erodibility, slope gradient and length, crop cover and conservation practices. The net soil loss is then calculated by comparing the calculated soil erosion with an estimate of the rate of soil formation, which is determined by thermal and LGP zone. Thirdly, loss of soil depth is related to productivity loss by adjusting land suitability classes within a critical soil depth range. Such calculations can be used to estimate limits of tolerable soil loss under defined cropping pattern options and to derive specifications for the required soil conservation measures.
The overall land productivity model, as applied in the Kenya study, quantifies productivity potentials of land by AEC for each crop rotation option, selected according to the rules outlined in Steps 4.1 and 4.2, in three stages:
· quantification of sequential crop yields;
· incorporating intercropping yield increments;
· applying production stability constraints and any other constraints as criteria for selecting optimum crop rotations and productivities.
The model can be applied using different sets of assumptions to govern the selection of crop combinations. Table 21 summarizes the aggregated results for Kenya, based on monocropping, including sequential monocropping where and when suitable growing periods occur. Thus the figures refer to total annual productivity for single crops based on addition of figures for individual AECs.
The determination of the extent and quality of arable land is one of the end results of the calculation of land productivity. Table 21 summarizes the extent of arable land in various productivity classes, based on assumption set B.
Assumption set B refers to potential crop productivity on all land which is not indicated as forest zone, game park, or belonging to an irrigation scheme. Whenever possible or appropriate, sequential monocrop combinations of two or three consecutive crops from a crop species have been constructed to ensure highest possible estimates in sub-humid and humid zones.
Six suitability classes have been defined relating average single crop suitability in a cell to maximum attainable yield. The classes C1 to C5 relate to average attainable yields of > 80%, 60-80%, 40-60%, 20-40% and 5-20% of maximum agro-climatic yields. Note that extents in suitability class C5 are usually not considered among the viable crop options, but have been included here to indicate the scope of production in very marginal areas. A sixth suitability class accounts for areas that are entirely unsuitable or allow for only < 5% of maximum yield. Data for this non-suitable class are not included in the results table.
Production potential is calculated from land extents in suitability classes C1 to C4 only. Average, minimum and maximum production potential and yields are determined according to LGP pattern and associated probabilities. The columns are labelled AVG, MIN and MAX respectively. Multiple land use in time, sequential cropping, is indicated by a multi-cropping index (MCI).
Table 21 gives estimates of arable land by productivity class. The algorithm used to determine rainfed arable extents in an AEC works in two stages. Firstly, the crop or monocrop combination which performs best under the worst climatic (according to LGP pattern) is determined. Then all crop combinations which meet the production stability constraint (i.e. fall within a tolerable yield range of the best performing crop) are considered in the final selection. Finally, among all qualifying crops, the combination that maximizes the weighted sum of extents in land suitability classes C1 to C4 is selected as describing the cell’s arable land potential. Suitable extents of the primary crop type in the chosen crop combination (i.e. the first crop to be grown in the sequential cropping pattern) are recorded in the relevant totals of arable land resources.
Population supporting capacity, as defined here, relates to the maximum potential of soil and climatic resources to produce food energy and protein, at a given level of technology. An intermediate level of input/technology is considered in this example (Fischer et al., 1996). The question is simply how much food can be produced on the potentially suitable land under optimal resource use?
An example is given for Bungoma district in Kenya and Figure 10 presents the distribution of harvested area obtained from optimal land allocation to achieve the maximum food production in the district. The scenario used in the optimization specified that all suitable lands are to be considered, including forest and game parks. Since the land resource map of Bungoma district is available in digitized form, a map can also be created showing where in Bungoma what cereals should be grown to achieve the single objective of maximum food production.
The above example shows the application of linear optimization techniques to the analysis of land-use scenarios according to a single objective function which is to maximize food production. Often the specification of a single objective function does not adequately reflect the preferences of decision-makers ,which are of a multi-objective nature in many practical problems of land resources optimization. Multi-objective optimization approaches address problem definitions and solutions in a more realistic way.
In the Kenya study the main issue was to analyse potential population supporting capacity of the district under various land-use scenarios, considering simultaneously several objectives such as maximizing revenues from crop and livestock production, maximizing district self reliance in agricultural produc- tion, minimizing costs of production and environmental damages from erosion. Multi-objective optimization coupled with multi-criteria decision analysis (MCDA) techniques, using the Aspiration Reservation Based Decision Support (ARBDS) approach, was used in the analysis.
The multiple objective programme includes the following objective functions:
1. maximize food output (Food-val) (average yield/production);
2. maximize net revenue (Net_rev);
3. minimize production costs:
4. maximize gross value of output;
5. minimize arable land use (weight of 1 assigned to crop and 0 to grassland) (arable);
6. minimize area harvested;
7. maximize food output (Food-min) (minimum yield in bad years);
8. minimize total erosion (Eros_tot) (sum of all cell erosions);
9. maximize self sufficiency ratio (SSR_v) (minimum of the individual commodity group self sufficiency ratios);
10. minimize maximum erosion (Eros-max) (largest occurring erosion per ha in a cell is small).
The results of a sample analysis for Bungoma district are given in Table 22. The first seven rows of the table contain the criteria values obtained from solutions for which each criterion is optimized in successive single-criterion optimization runs. The diagonal elements of the matrix represent the Utopia or “best” values for the seven criteria (i.e.1197.2, 1316.6, 96.2, 1010.5, 1164.9,1337.8, 12.2). The Nadir or “worst” values are found by taking the lowest values in the columns of the criteria to be maximized (i.e. Food_val = 742.6, Net_rev = 783.0, Food_min = 548.4, SSR_v=1000.0) and the highest values of the columns of the criteria to be minimized (i.e. Arable = 165.4, Eros_ tot = 3527.0, Eros-max = 227.8).
The last five rows of Table 22 contain the criteria values resulting from a session of interactive multicriteria analysis involving five iterations. The user interacts with the software tool through successive screens displaying graphs of the decision variables, using mouse clicks to make the desired changes in values of decision variables.
The results shows an irregular pattern of variation of the decision variables within the sequence MCD-B … MCD-E. Generally the increase in arable land use required to achieve higher food production and self sufficiency ratios appears to be associated with increased total erosion; food production, economic return and food security in terms of guaranteed minimum production in bad years and maximum erosion vary within narrow ranges and seem to stabilize.
Given that the solutions produce self sufficiency rates above the 80% minimum limit which was established for the scenarios, the MCD-C solution appears to be a good choice as it represents the relatively “best” optimal combination of values of the decision variables.