An alternative method of area measurement for large agricultural surveys

The unsuitability of current ‘objective’ methods used in agricultural surveys of smallholder farming in Africa has often been stressed but scientifically tested alternative methods that may be used to replace them are yet to be developed. The present paper reports on the ‘Longest Diagonal Method’ of area estimation, with which encouraging results have been obtained in Malawi.

The field work consists of measuring the longest distance within the area (farm plots) from one end to another: in the testing stage, area estimates based on this variable, designated LD, were compared with the ‘known’ area as measured with tape and compass. At an early stage of our experimentation, an approach involving the computation of the mean of a factor F relating LD to the area A was rejected, and this was replaced by regression analysis in which a second diagonal, designated as the perpendicular or P, is also playing a role in combination with LD. This second diagonal P runs at a 90° angle from the midpoint of LD to the edges of the plot, and since the introduction of P the field work consisted of measuring both LD and P. After the introduction of shape classification based on the ratio R = P/LD as an indicator of a field's degree of compactness, valid aggregate area estimates were obtained and as regards farm size classification the results were also quite acceptable.

Hitherto, efforts to find alternative, more cost-effective methods that would enable agricultural surveys to provide reliable information on crop production at project level of disaggregation have focused almost exclusively on the possibility of using farmers' own estimates. However, this method taken in isolation has its limitations and to be fully effective a breakthrough in this direction, one way or another, needs to be complemented by a less cumbersome approach towards area estimation. While improved area estimation has merits in its own right as well, the most significant effect of the present contribution will be found in this combination.