Agricultural production and groundwater depletion under climate variability in India - Results from a regional scale crop modeling approach
Abstract
In India, recent declines in national food security may point to systemic deficiencies of agricultural production. Over the past decade and in the face of declining public investments in irrigation projects, the growth of production has increasingly become reliant on the allocation of large volumes of groundwater in an unsustainable manner. As a result, shallow as well as deep fossil groundwater resources are increasingly depleted and the buffer that mitigates negative impacts on production in case of Monsoonal dry-spells / drought conditions is lost. In the face of future climate and food supply uncertainty, it is vital that the connections between climate variability, unsustainable irrigation practices and their impacts on regional scale agricultural production be quantified and better understood. In our analysis, we focus on rice production in the Telengana region in Andhra Pradesh, which is characterized by a semi-arid tropical climate that is driven by the bimodal seasonality of the south-western monsoon. Traditionally, agricultural production of rice was constrained by precipitation variations during the wet season (Kharif). However, the advent of inexpensive pump technology in the 1970's, coupled with governmentally subsidized electricity has allowed year-round rice production. Thus, the Monsoon rains must not only drive wet season production but must also sufficiently recharge groundwater in order to support dry season production. Observed Production time series are characterized by non-stationarity and heteroscedasticity. Using a subset of eight districts, a non-linear Gaussian Process regression model is developed and yearly crop production is modeled at the district level over 48 years. We show that interannual climate variations, in the form of the monsoon rains, play a significant role in determining the area of land set aside for dry season planting and thus affect total yearly production. The results suggest that a non-linear Bayesian regression approach combined with more accurate monsoon forecast may lead to predictive skill for crop forecasts at regional / national scales. Additionally, coupled climate and agricultural production models that explicitly account for uncertainty in forecasts may be useful to decision makers and stakeholders as they attempt to use water resources more efficiently and sustainably.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2009
- Bibcode:
- 2009AGUFM.H11D0824S
- Keywords:
-
- 0402 BIOGEOSCIENCES / Agricultural systems;
- 1807 HYDROLOGY / Climate impacts;
- 1842 HYDROLOGY / Irrigation;
- 1880 HYDROLOGY / Water management