Data-driven reconstructions of the Indian palaeomonsoon (1500–1995 CE)

These datasets are fully described in the paper A novel explainable deep learning framework for reconstructing South Asian palaeomonsoons. 

Paper abstract:
We present novel explainable deep learning techniques for reconstructing South Asian palaeomonsoon rainfall over the last 500 years, leveraging long instrumental precipitation records and palaeoenvironmental datasets from South and East Asia to  build two types of model: dense neural networks (`timeline models') and convolutional neural networks (CNNs). The timeline models are trained individually on seven regional rainfall datasets and while they capture decadal-scale variability and significant droughts, they underestimate interannual variability. The CNNs, designed to account for spatial relationships in both predictor and target, demonstrate higher skill in reconstructing rainfall patterns and produce robust spatiotemporal reconstructions. The 19th and 20th centuries were characterised by marked inter-annual variability in the monsoon, but earlier periods were characterised by more decadal- to centennial-scale oscillations. Multidecadal droughts occurred  in the mid-seventeenth and nineteenth centuries, while much of the eighteenth century (particularly the early part of the century) was characterised by above-average monsoon precipitation. Extreme droughts tend to be concentrated in south and west India and often coincide with recorded famines. Our findings offer insights into the historical variability of the Indian summer monsoon and highlight the potential of deep learning techniques in palaeoclimate reconstruction.

Dataset description:
regional-famines.csv: Derived from https://en.wikipedia.org/wiki/Timeline_of_major_famines_in_India_prior_to_1765 and https://en.wikipedia.org/wiki/Timeline_of_major_famines_in_India_during_British_rule. Columns are the seven homogeneous rainfall regions of India, rows are years. Variable is 1 is a significant famine occurred somewhere in that region in that year, else 0.

timeline-model-regional-prcp.csv: Reconstructed seasonal (Jun–Sep) monsoon precipitation anomalies for each of the seven homogeneous rainfall regions of India, as well as all-India. Period covered is 1501–1995.

cnn-ensemble-prcp.nc:  NetCDF file containing spatial maps of reconstructed seasonal monsoon precipitation anomalies from the CNN model. Available for each year from 1501–1995, at a resolution of 0.25°x0.25°. Two variables are included, the full ten-member ensemble and its mean.

cnn-ensemble-mean-individual-years.zip: A ZIP archive containing maps showing the CNN ensemble mean for each year.

cnn-ensemble-mean-all-India.csv: The all-India average of the CNN ensemble mean for each year, giving a yearly monsoon index from 1501–1995.