**2.2 Data sourcing and analysis**

Data sourcing began with desk study through literature review, computer generation of study sites, and secondary data gathering. Specific secondary data include the following: historical records (1700–2000) on Ahafo District; rainfall, humidity, and temperature for Goaso synoptic station, and satellite images of Asunafo for 1986, 2003, and 2020 [22–24]. The desk research was followed by a 4-day drive across the districts. The drives included trips from:


The next step was the verification of the selected 21 case study communities. The selection process was informed by Ghana Statistical Service's choice of the 20 largest communities in each district in the country ([25], p. 54). However, Goaso, Kukuom, Mim, Kwapong, and Sankore were replaced by nearby communities because of lack of logistics to cater for such large towns.

Subsequently, 21 community meetings were organized early in the morning for farmers in each community. Attendance of the meetings was recorded in a field notebook, and the proceedings were recorded on audio cassettes. About eight (8) farmers attended the smallest meeting, while 81 farmers attended the largest meeting. The total number of farmers who attended the 21 meetings was 774, and they formed the population frame for the study. A sample size of 264 farmers was selected based on a formula: n = N/1 + N(e)2. About 15 research assistants were drawn from third and fourth-year undergraduate students to assist in the data gathering through the administration of questionnaires, key informant interviews, and personal observation. The rest were transect drives, community meetings, field visits, and photography. Archival data on Ahafo District were provided by the Regional Public Records and Archives Administration (PRAAD) of the Brong Ahafo Region. The study categorizes the period for a considerable outcome into three, namely: short-term, medium-term, and long-term. The short-term deals with a period of fewer than 2 years, the medium-term ranges between 2 and 6 years, and the long-term considers more than 6 years. Data analysis involved the use of ENVI 4.3, ArcGIS 9.3, SPSS, and Microsoft Excel.
