A double sampling technique is combined with remotely sensed and field data to estimate fuelwood rapidly. Fuelwood is an essential energy resource for cooking and heating in many developing countries, but due to population increase and intensification of agriculture, the capacity of this resource to meet present and future needs is being exhausted. In extreme cases this leads to famine, but more insidiously, a shortage of fuelwood increases rural poverty as households must spend meagre financial resources on alternative (fossil) fuels. Techniques to rapidly estimate fuelwood in an efficient manner are therefore critical to rural planning and development.
The technique presented in this paper uses a novel double sampling approach for rapid fuelwood appraisal. A multistage sampling design was used to select field plots and the basal area of important fuelwood species were calculated for these plots. The basal area values were then regressed against vegetation indices calculated from Landsat TM for three test areas (Gatshe Gatshe, Kanyati and Hurungwe) with different fragmentation patterns in northwestern Zimbabwe. NDVI yielded the highest correlation coefficients. The relationship between NDVI (auxiliary variable) and fuelwood (variable of interest) was then inverted using double sampling, in order to estimate fuelwood over the three rural (communal) test sites (correlation coefficients for the three areas were r2 = 0.91, r2 = 0.80, r2 = 0.76). Double sampling estimates for different vegetation types showed high fuelwood quantity in mopane woodlands as compared to miombo and remnant woodland in an agricultural area. The implications for management of fuelwood in northern Zimbabwe are discussed.