Climatic Change

, Volume 126, Issue 1–2, pp 61–76

Winter crop sensitivity to inter-annual climate variability in central India

  • Pinki Mondal
  • Meha Jain
  • Andrew W. Robertson
  • Gillian L. Galford
  • Christopher Small
  • Ruth S. DeFries

DOI: 10.1007/s10584-014-1216-y

Cite this article as:
Mondal, P., Jain, M., Robertson, A.W. et al. Climatic Change (2014) 126: 61. doi:10.1007/s10584-014-1216-y


India is predicted to be one of the most vulnerable agricultural regions to future climate changes. Here, we examined the sensitivity of winter cropping systems to inter-annual climate variability in a local market and subsistence-based agricultural system in central India, a data-rich validation site, in order to identify the climate parameters to which winter crops – mainly wheat and pulses in this region – might be sensitive in the future. We used satellite time-series data to quantify inter-annual variability in multiple climate parameters and in winter crop cover, agricultural census data to quantify irrigation, and field observations to identify locations for specific crop types. We developed three mixed-effect models (250 m to 1 km scale) to identify correlations between crop cover (wheat and pulses) and twenty-two climate and environmental parameters for 2001-2013. We find that winter daytime mean temperature (November–January) is the most significant factor affecting winter crops, irrespective of crop type, and is negatively associated with winter crop cover. With pronounced winter warming projected in the coming decades, effective adaptation by smallholder farmers in similar landscapes would require additional strategies, such as access to fine-scale temperature forecasts and heat-tolerant winter crop varieties.

Supplementary material

10584_2014_1216_MOESM1_ESM.pptx (68 kb)
SI Fig. 1Seasonal precipitation in the study region from 2000 to 2013. Monsoon precipitation was calculated for May to October during 2000-2012. Winter precipitation was calculated for November to January from 2000-2001 to 2012-2013. (PPTX 68 kb)
10584_2014_1216_Fig7_ESM.gif (188 kb)
SI Fig. 2

Spatial distribution of winter crop peak date in 2010 (a) and phenology time-series of sample pixel representing a double-cropped pixel during 2000-2001 to 2012-2013 (b). Vertical red lines in panel b indicate peak date span (Jan 17 – Feb 1) considered in this study. The year 2010 was selected to represent the spatial distribution of winter crop cover in a highly productive year. (GIF 187 kb)

10584_2014_1216_MOESM2_ESM.tif (2.3 mb)
High resolution image(TIFF 2403 kb)
10584_2014_1216_MOESM3_ESM.pptx (742 kb)
SI Fig. 3Comparison of two precipitation datasets (spatial resolution: 0.25 degrees) from different sources – India Meteorological Department (IMD) gridded dataset interpolated from station data, and satellite-derived Tropical Rainfall Measuring Mission (TRMM) dataset used in this study. To the best of our knowledge, no consistent station data were available for the region and time-period considered in this study. However, we obtained gridded (0.25 degree) June-September daily precipitation data for 2000-2010 from the India Meteorological Department (IMD) to examine the accuracy of the satellite data. The IMD dataset is developed by interpolating data from >3,000 rain-gauge stations located all over India (Rajeevan M, Bhate J (2009) A high resolution daily gridded rainfall dataset (1971–2005) for mesoscale meteorological studies. Curr Sci 96:558–562). The degree of correlation between IMD and TRMM data can vary depending on the year considered with Pearson’r r ranging between 0.06 for 2006 and 0.75 for 2001. The two data sets are in moderate agreement (Pearson’r r = 0.43) when all the years are considered together for our study region (bottom right figure). (PPTX 741 kb)
10584_2014_1216_MOESM4_ESM.pptx (98 kb)
SI Fig. 4Relative importance of predictor variables based on the Akaike Information Criterion (AIC) values calculated from the mixed-effect models for all winter crops (a), wheat (b) and pulses (c). Multiple runs were conducted for each model. The AIC values for each of these runs were subtracted from the lowest AIC value to calculate the relative importance of the predictor variables for each model. (PPTX 97 kb)
10584_2014_1216_MOESM5_ESM.pptx (195 kb)
SI Fig. 5Correlation coefficients between seasonal mean temperature and corresponding peak EVI for each of the wheat and pulse field locations (n=45). Pixels with strong correlation (Pearson’s r >0.6) are highlighted in the table. The bottom panel shows the correlations for representative wheat and pulse pixels. (PPTX 195 kb)
10584_2014_1216_Fig8_ESM.gif (121 kb)
SI Fig. 6

Spatial distribution of coefficient of variation for MODIS surface temperature (spatial resolution=1 km). (GIF 121 kb)

10584_2014_1216_MOESM6_ESM.tif (815 kb)
High resolution image(TIFF 814 kb)
10584_2014_1216_MOESM7_ESM.docx (15 kb)
ESM 7(DOCX 14 kb)
10584_2014_1216_MOESM8_ESM.docx (26 kb)
ESM 8(DOCX 26 kb)
10584_2014_1216_MOESM9_ESM.docx (27 kb)
ESM 9(DOCX 27 kb)
10584_2014_1216_MOESM10_ESM.docx (17 kb)
ESM 10(DOCX 17 kb)

Copyright information

© Springer Science+Business Media Dordrecht 2014

Authors and Affiliations

  • Pinki Mondal
    • 1
  • Meha Jain
    • 1
  • Andrew W. Robertson
    • 2
  • Gillian L. Galford
    • 3
  • Christopher Small
    • 4
  • Ruth S. DeFries
    • 1
  1. 1.Department of Ecology, Evolution and Environmental BiologyColumbia UniversityNew YorkUSA
  2. 2.International Research Institute for Climate and Society (IRI)PalisadesUSA
  3. 3.Gund Institute for Ecological Economics, Rubenstein School of Environment and Natural ResourcesUniversity of VermontVermontUSA
  4. 4.Lamont-Doherty Earth ObservatoryPalisadesUSA

Personalised recommendations