We used satellite imagery and aerial photography to map landscapes in the study area. A series of training sites are used, where remotely sensed data were 'ground truthed' during a field campain. We use archaeological data as a proxi to identify historical time intervals when irrigated farming was practiced in the region, and the timing of irrigation cessation.
This project was inspired by the following observation made by the authors. In the study area, landscapes (or simply, land units) that were put under irrigated farming, after several years under production experienced various degrees of salinization, often severe. After irrigation cessation due to high levels of soil salinity (largely caused by improper irrigation techniques, i.e. field flooding with irrigation waters), there appears to be a natual sequence of stages of evolution of those landscapes. We refer to this process as post-irrigational desertification, and postulate that it usually leads to the progression from higly saline solonchak soils, to clayey takyrs, to sandy deserts (with intermediate stages in between). We also suggest that for a given landscape, if ever under irrigation, there is a correspondence between the current state of landscape and the timing of irrigation cessation.
Having given consideration to the above, we have attempted to isolate the human forcing (i.e., land use) from the natural (i.e., climate) forcing as landscapes in the study area evolved over the last five thousand years. We found that climate fluctuations were such that they resulted in vegetation shifts from steppe to semi-deserts and deserts. We suggest our dating of major environmental shifts in the region and discuss causes for uncertainty. Humans tended to adjust to these natural climate and environmental fluctuations by migrating to better drained areas and by adjusting their agricultural practices. Thus human impact being superimposed over the natural trend led to an increased spatial variability in reconstructed and present-day landscapes.
Our results can be useful for performing comparisons over the study region with the output from climate and land surface models, provided that the models are driven with the forcing consistent with that over the last five thousand years.