Climate is known to impact the growth of plantation forests in South Africa. To characterize the conditions, this study developed temperature and rainfall models for specific plantations in the Lowveld Escarpment and Highveld forestry regions of South Africa. Global position, altitude, slope, aspect, and topographic position indices were considered when developing the models. Seasonal minimum (Tmin) and maximum (Tmax) temperature models were developed using observations obtained from 43 weather stations for the period 2013 until 2019. Out of six models which were initially evaluated, two machine learning models, random forests and linear-based, were selected for further scrutiny. R-square values for models predicting Tmin on the Lowveld Escarpment ranged from 0.64 to 0.89 for seasonal models. For the Highveld, seasonal models to predict Tmin had R-square values ranging from 0.47 to 0.84. Seasonal Lowveld Escarpment Tmaxmodels were also significant with R-square values ranging from 0.76 to 0.89. The equivalent models for the Highveld had R-square values ranging from 0.54 to 0.83. Rainfall models were developed from 69 rainfall stations spanning a twenty-year period (1999 – 2019). Multiple median linear regression was used to model mean seasonal rainfall (Pmedian) with R-square values ranging from 0.87 to 0.90. These findings demonstrate that it is possible to develop accurate local climate models using global positioning and local terrain features. Comparing the models to transformed WorldClim 2.1 predictions, it was found that the majority of the newly developed climate models were more accurate than the global models, with the exception of minimum temperature models which was marginally better. However, the tranformed WorldClim 2.1 predictions were surprisingly accurate. The local climate models use a finer scale, and they improve the understanding of how terrain features impact regional climate and historic tree growth. In addition, the finer scale models give insights into the relationship between the local terrain condition and regional climate. It is also likely that these models can be used for other forestry regions within Southern Africa, and similar approaches may be relevant globally.