Assessing Causal Relationships in Agricultural Soil and Water Management from Observational Satellite Data
Abstract
Over the past decade, advances in computing, satellite, and drone systems have enabled a surge in remote sensing capabilities for agriculture. The resulting geospatial data includes maps of crop types, yields, and management practices such as planting dates, soil tillage type, and irrigation activity. Challenges remain, however, for drawing robust causal inferences about management interventions from these observational datasets. Here, I present two case studies demonstrating how forest-based machine learning, synthetic controls, and agro-hydrological models can help analyze the causal impact of agricultural practices based on satellite observations. The first case study focuses on the yield impacts of conservation tillage in the US Corn Belt, demonstrating that soil conservation practices can be used with minimal and typically positive yield impacts. The second case study examines the effectiveness of stakeholder-driven groundwater management in the central United States, examining water conservation outcomes, crop yield impacts, and energy used for groundwater pumping.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFMGC14B..02D