Chapter 5
Program AEZCCS04: LP Matrix Generator

Program AEZCCS04 reads the output from the land productivity assessment (program AEZCCS03) and prepares a data file for input to a linear programming package according to the specifications given in the scenario control input file. Output is generated in standard MPS file format (see Appendix E).

5.1 Program Logic and Overview

The scenario control input file, file CTRL$$$, is read (subroutine CTRLIN). Here, the user specifies the mode of operation, several program control switches and, optionally, a subset of constraints of the linear program which is prepared to subsequently arrive at an optimal land use pattern. Aggregation indices, conversion weights and various technical coefficients at different levels of aggregation are read from a data set as specified on the scenario control file (subroutine LOADT1). Then, the livestock systems definition tables are processed (subroutine L0ADT2), and crop specific input requirements in table function format are read into memory (subroutine LOADTM). Livestock distribution parameters are checked for consistency and feasibility (function LVDCHK).

The main program loop starts with reading the cell information record from the unformatted sequential file created by program AEZCCS03 (subroutine CELLIN). Then, basic accounting of cell extents takes place (subroutine ACCALC), the population density relevant to the current cell is retrieved, and the crop combination records relating to the current cell are screened and copied to a temporary working file (subroutine TBGET).

The temporary file is re-read and each available crop combination record is processed (subroutine CMBIN). Various land area counters are updated (subroutine ACCALC), and production from every admissible crop combination is determined (subroutine QCCALC). The input requirements for production in terms of seeds, fertilizer, power and pesticides are derived from the technology matrix (subroutine INPUTS) and the respective weight in the objective function of the LP is calculated (function OBJFUN). The relevant coefficients of the LP constraint matrix are generated (subroutine CMBMAT) and saved on temporary unformatted sequential working files (subroutine OUTREC).

After having processed all the crop combination records available for the current land inventory cell, the program proceeds with reading the next cell information record continuing this operation cycle until all cells have been read and dealt with.

FIGURE 5.1
Information flow in program AEZCCS04

Finally, the program turns to the livestock systems feeding and distribution constraints (subroutine LVSMAT). While processing all the crop combination information, the program also calculates and aggregates data on feed supply by livestock zone. This information is used to generate livestock zone and livestock system specific feed balances and livestock system share constraints (subroutine QLCALC).

The program ends with writing out the LP specification - objective function, constraint matrix, right hand sides and activity bounds - in standard SPECS and MPS data file format (subroutine LPGEN). The information flow in program AEZCCS04 is shown in Figure 5.1.

5.2 Input/Output File Sizes and Connections

^a) ‘V’ indicates a variable, district or country specific number of records.

^b) FS = formatted sequential, UR = unformatted random access, US = unformatted sequential.

^c) R = input, W = output.

^d) The size of the temporary working files SCR104 to SCR404 and the size of the LP specification file LPSPECS largely depend on the complexity of the land inventory and the number of admissible crop combinations that have to be processed. In the Kenya Case Study LPSPECS files of more than 10 Mb have been observed in rare cases only (e.g. Meru district, 25 Mb), with typical sizes in the range of 1 – 5 Mb.

^e) ‘$$$’ in file names BIN.$$$ and OUT.$$$ has to be replaced by the respective three digit district codes.

5.3 Subprogram Description:

TABLE 5.1
Variables read from scenario control input file, file CTRL$$$

*⁾ ‘’ indicates array subscript value calculated from configuration specific parameter constants (see Appendix B).

5.4 Program Control Data, File CTRL$$$

Each AEZ-CCS program is controlled by a short input data file which is usually read from standard input. Table 5.1 provides the list of variables read from the control file used to run LP matrix generator program AEZCCS04 and LP Report Writer program AEZCCS05. In the table only the fixed portion of the control file is explained. In addition, the user can optionally indicate various LP constraints related to target demand, production levels and land use, availability and use of production inputs, investment constraints, and soil conservation. Example 5.1 shows a simple version of the control input data file for Meru district in Kenya at intermediate level of input,

In addition to the fixed portion of the scenario control file CTRL$$$ the user can provide data to generate various LP constraints that must be entered in four categories:

1. net production constraints at aggregate food commodity level, e.g. target production level of pulses

2. acreage and production constraints by agricultural commodity, e.g. wheat acreage, coffee acreage, maize production level

3. input use constraints, e.g. fertilizer use availability

4. miscellaneous constraints: e.g. conservation labor supply, investment limit.

Each of the constraints requires specification of a constraint type indicator and a constraint value. The constraint type indicator determines the kind of constraint that will be generated: 0 = unconstrained, 1 = equality constraint, 2 = lower bound inequality, 3 = upper bound inequality. Example 5.1, showing a simple scenario control file for Meru district in Kenya, illustrates the use of optional acreage constraints.

5.5 Commodity Aggregation Data, File COMMDAT

This text file contains three kinds of record types, CF, CQ, and AG, which refer to three different levels of commodity aggregation, namely, crop list, agricultural production, and aggregate food commodity list, respectively. Each record type contains commodity name information and technical coefficients such as conversion weights and extraction coefficients. The format of each record type is described below.

5.5.1 CF-Card: Conversion from Crop -> Agricultural Production List

The CF cards describe the first aggregation step, i.e. the conversion from individual crops to an agricultural production commodity list. Apart from crop name text labels this card type contains aggregation indices, respective conversion weights and other crop specific information.

5.5.2 CQ-Card: Conversion from Agricultural Production -> Food List

The CQ cards describe the second aggregation step from the agricultural production commodity list to an aggregate food product list. Each record contains commodity name, aggregation index and conversion weight. Waste factors account for losses incurred during harvesting and marketing.

5.5.3 AG-Card: Food List Cards

The AG cards describe the third aggregation level, the aggregate food commodity list. Demand constraints specified in the LP control input data file refer to this level of aggregation. There are two ways of specifying this type of constraints:

1. direct specification of quantities demanded

2. automatic generation of demand constraint from target per capita consumption levels contained in control file.

If per caput consumption levels are non-zero in file CTRL$$$ and the automatic demand generation flag (FLCCP) is set, then demand constraints will be imposed according to the priority level specified on the AG card. Priority level 01 results in inequality constraints, priority level 02 results in equality constraints. Note also, that direct specification of demand constraints overrides automatic demand constraint generation.

5.6 Livestock System Definition Data, File LVSDAT

This text file contains three kinds of record types, LZ, LS, and LQ, which refer to livestock zone definition records, livestock system definition, and livestock product definition, respectively. The format of each record type is described below.

5.6.1 LZ-Card: Livestock Zone Definition Records

The LZ cards map agro-ecological zones to livestock zones. Each combination of thermal zone and LGP codes is related to a livestock system zone. This relation is important as all the livestock feed balances are imposed at the level of a livestock zone.

5.6.2 LS-Cand Livestock System Definition Records

The LS cards describe each livestock system in terms of feed requirements and productivity. The records contain a suitability rating per livestock zone, feed requirements per TLU and relevant production of livestock products per TLU (up to four joint products per livestock system).

* In the Kenya Case Study MLVZ=16 livestock zones have been distinguished. In the computer program the number of livestock zone codes is defined by parameter constant MLVZ (see Appendix B).

5.6.3 LQ-Card: Livestock Product Definition Record

The LQ cards describe livestock products in terms of name, aggregation index and conversion weight, i.e. their relation to the aggregate food list, and their value in terms of nutritional contents and unit price.

5.7 Technology Matrix, File TECHMAT

The TECHMAT file is used to describe the input to output relationship of the crop production activities in the study. The technology matrix uses a table function format. Each record refers to a specific crop under specific agro-climatic conditions at a specific level of technology and indicates the required input levels in terms of seeds, fertilizer, power, and pesticides in relation to the respective yield level. Access to this table is achieved using the respective look-up index specified on the LC-cards described in the previous section.