...Map it a little at a time!

The last post showed the study by Lebel et al (2015) which carried out modelling on a continental scale across Africa. As stated in the last post, I think modelling should be carried out on a smaller, more detailed scale to really capture the how well rainwater harvesting (RWH) works specific to each area. This post aims to shw that modelling RWH on a smaller scale and in more detail, can be more effective than when modelled at a continental scale.

Nthuni et al (2014) used spatial modelling techniques in ArcGIA to model the benefits of RWH in Kakamega, western Kenya. They created four conceptual models at three different levels of detail to show the potential of RWH as a source of safe water for domestic use. Geospatial data and GIS were used to map the areas where RWH has most potential. RWH is not commonly used here and so this study investigates the potential it has. The regional climate is characterised by heavy rainfall every afternoon, but climate change means there are more frequent and extended periods of drought. This needs to be taken into account when assessing RWH potential.

Modelling at this scale when determining the potential of RWH is, in my opinion, more effective than continental scale modelling for many reasons. Firstly, this model considers the rainwater received by the catchment over a given period of time (Nthuni et al 2014). This is more accurate than at a continental scale because climate can change considerably across such a large land mass like Africa. Furthermore, at this scale, one is able to assess the social factors impacting the potential of RWH in the area. For example it can look at how densely populated the region is and even the wealth of the region to see if they have the infrastructure for, or can afford, RWH (Nthuni et al 2014). At a continental scale this is not possible and so saying an area has high potential for RWH would be inaccurate since it would not have considered the socioeconomic factors of the local people. For example, in this study they assessed the potential RWH has with reference to demand from that area. They didn’t just look at how many people would use the water, but also the domestic use of each individual or household over a period of time (Nthuni et al 2014).

The land in Kakamega is uneven terrain. The study found that up to 16000m asl there is an increase in precipitation, but beyond this the relationship is inverted (Nthuni et al 2014). This is another factor important in determining RWH potential that would not be taken into account if using modelling at a continental scale. Another key conclusion from this study is that there are three sub locations within Kakamega that would not be suitable for RWH use. This is because these areas have the highest population densities within the region. Again, this is another factor what would not be considered by continental scale modelling. The detail when modelling at a continental scale is not fine enough to analyse population densities of areas in this much detail. Furthermore, this is an essential point to consider because in the light of climate change and growing populations it is likely this could become an issue for many areas of Africa and so needs to be taken into account. By looking at locations in this much detail one is able to tell if the potential of RWH is restricted by population densities being too high, a lack of sufficient rainfall, or a mixture of both.

Therefore, as we can see from this example, modelling in more detail than at a continental scale is much more effective. Continental scale modelling is not detailed enough to show many things that we can decipher from more detailed modelling at a local scale. The study by Nthuni et al (2014) looked at the potential of RWH for domestic purposes. This blog focuses on agriculture in Africa so the methods and concepts used in this study can be applied to the look at potential of RWH for agricultural purposes. Continental scale modelling may be appropriate to research other factors but when researching the potential of RWH I think local scale modelling is better.