At this stage of the TwoRains project we are concentrating our efforts in northwest India, including the modern states of Haryana, Punjab and Rajasthan.

There has been considerable historical interest in the hydrology of the Indus plain and many scholars have attempted to reconstruct the location of ancient, now disappeared, rivers in relation to the distribution of known archaeological sites in the area. However, there are various problems with the approaches used previously: in particular, the enormous size of the study area presents a challenge to the application of remote sensing. No single image with enough spatial resolution can cover a significant portion of the area and most of the previous analysis have employed single images or a limited numbers of images in combination. The second challenge encountered is the high diversity of environmental scenarios in the region, with there being significant differences in seasonality, cultivation patterns and strong landscape modifications, which suggest that single images will not capture the range of variability.

Figure 1: RGB composite of Tasseled Cap indices for brightness, greenness and wetness showing several palaeorivers in the north-east of the study area

We plan to overcome these problems by employing large-scale multi-temporal remote sensing approaches. A long-term temporal research perspective will be invaluable in overcoming problems posed by changing cultivation and land-use patterns. As the availability of multi-spectral imagery for the study area cover the last forty years, this approach is theoretically possible, but hindered by the enormous computational resources necessary for the treatment and analysis of the large amount of data entailed. The use of multi-cluster parallel computing, which is available through the Cambridge High Performance Computing Service, is allowing us to enact such approach.

With the implementation of these methodological premises, we aim to comprehensively re-evaluate the relationship between settlement and the changing hydrological systems of the Indus zone in northwest India. By doing so, we expect to be able to explore the mechanisms involved in the human adaptation to and management of changing water availability conditions.

The preliminary results have allowed us to identify several known, but also hundreds of previously unknown palaeochannels in the study area (fig. 1). Although a degree of correlation between ancient settlements and watercourses is evident, the distribution of settlements does not correspond with the detected palaeorivers exactly. This might be due to the braided morphology of rivers (fig. 2) in the study area (typical of a flat geomorphological setting). However, some tests with GIS-based hydrological modelling of the study area show that known sites might have a closer link to the maximum extent of flooding episodes. However, at present the quality of the site distribution data and the lack of accurate chronologies for these sites prevent the extraction of definitive hypotheses.

Adam Green is currently working on the extraction a more reliable (in terms of location and chronology) database of archaeological sites that can be then employed to compare with the remote sensing identified palaeochannels in the area. Until then the remote sensing aspect of the TwoRains project will continue delineating and modelling water movement and accumulation in the study area.

Figure 2: Digital terrain model showing the braided morphology of rivers in the study area