In order to make the analysis comprehensive all of the criteria have to be included in the model. Moreover, they have to be weighted according to their measured or estimated importance. This section lays out the procedure.
It is convenient to consider land and water together in terms of their joint availability in each district. Also, within the water criterion, irrigation schemes and “effective rainfall” have to be accommodated. Within the land criterion, “good” and “fair” soil polygons have to be accommodated with the alluvial soils adjacent to rivers. These criteria have been considered together to make a Land and Water Index, KL:
KL = 1.5 (G*R+I*3) + (S*R)4
G=km2 of soils rated “good” for ponds plus km2 of alluvial soils adjacent to rivers
R=“effective rainfall” taking values 0.15, 1.15, 2.15 and 3.15 corresponding, respectively to “insufficient, marginal, adequate and ample”
4 Statistical Footnote. All of the index values that were used to build the integrated models were normalized. In the initial runs of the models it was found that the Land and Water Index (KL) was overwhelming the other indices because of its relatively high values. Normalization overcomes this problem.
Likewise, in order to make the results of the integrated models directly comparable, the values generated by each model were normalized.
I=Km2 of irrigated area
S=km2 of soils rated “suitable” for ponds
Irrigated area, I, has an R value of 3 because, by definition, in an irrigation scheme, water should always be “ample”. Furthermore, irrigated area is added to “good” soils because with ample water, texture or permeability problems can be overcome (although at some cost).
Soils rated “suitable” area, of course, second in quality to “good” soils. This is taken into account by using a weighting coefficient of 1.5 for the “good” soils.
The spatial distribution of the Land and Water Index values is shown in Figure 20 and values by district are in Table 6.
4.1.1 Integration of Agglomeration, Extension and Development
In combining these three criteria, we reasoned that, because extension capabilities are at present so limited, agglomeration is a more important criterion than extension in stimulating fish farming. Similarly, we judged that the general level of development in a district is less important than agglomeration in stimulating fish farming. Thus, the criteria were combined to make an “Other Factors” index, KF, in which agglomeration was given twice the weight of development and extension.
The spatial distribution of the indices of agglomeration, extension and development, taken together, is shown in Figure 21 and the index for individual districts is in Table 5.
There was no preliminary integration of inputs, markets and welfare. This was done at the final stage, the integrated models.