Developing the Necessary Data

In order to find valid tradeoffs between agriculture and conservation, and to gain a proper understanding of how agriculture responds to global change, we need agricultural (e.g. productivity, crop choice, field boundaries) and environmental (e.g. daily weather, carbon stocks, species range maps) data that are regional to continental in extent, span the last two to three decades, and are accurate at fine field- to landscape-scales. Why do we need data that have such exacting standards? For two important reasons:

  1. To make agricultural development more sustainable, we are essentially trying to find and direct development to those areas that are both a) agriculturally productive and b) will result in unusually low environmental damage when developed. This means that we are looking to delineate specific areas, based on the values of two or more variables, which makes this type of investigation unusually sensitive to data error.
  2. To understand how agriculture responds to socioeconomic and environmental changes, it is ultimately necessary to study how farmers change their agricultural practices and land use over time and in space. That cannot be done with provincial or national-level statistics.

Unfortunately, the agricultural and environmental datasets that we need to conduct such analyses do not yet exist (with some rare exceptions) for the regions that will change most rapidly this century, particularly sub-Saharan Africa. Furthermore, the characteristics of agricultural systems in this region (small fields that often look indistinguishable from savanna vegetation in satellite image analyses) complicate the task of developing the necessary data.

I therefore devote a substantial amount of research effort to developing the tools and methods needed to create these datasets, utilizing some of the latest advances in Earth Observing technology and data analytics.

Related Work



  • Debats, S., Luo, D., Estes, L.D., Fuchs, T., Caylor, K. 2016. A generalized computer vision approach to mapping crop fields in heterogeneous agricultural landscapes. Remote Sensing of Environment, 179, 210-221.
  • Estes, L.D., McRitchie, D., Choi, J., Debats, S., Evans, T., Guthe, W., Luo, D., Ragazzo, G., Zempleni,R., Caylor, K. 2016. A platform for crowdsourcing the creation of representative, accurate landcover maps. Environmental Modelling and Software, 80, 41-53.
  • Sweeney, S., Ruseva, T., Estes, L.D., Evans, T. 2015. Mapping cropland in smallholder-dominated savannas: integrating remote sensing techniques and probabilistic modeling. Remote Sensing, 7, 15295-15317.
  • McDermid, S.P., Ruane, A., Hudson, N.I., Rosenzweig, C., Morales, M.D., Agalawatte, P., Ahmad, S., Ahuja, L.R., Amien, I., Anapalli, S.S., Anothai, J., Asseng, S., Biggs, J., Bert, F., Bertuzzi, P., Bhatia, V.S., Bindi, M., Broad, I., Cammarano, D., Carretero, R., Chattha, A.A., Chung, U., Debats, S., Deligios, P., De Sanctis, G., Dhliwayo, T., Dumont, B., Estes, L.D., et al. 2015. The AgMIP Coordinated Climate-Crop Modeling Project (C3MP): Methods and Protocols. In Hillel, D., and C. Rosenzweig (Eds.), ICP Series on Climate Change Impacts, Adaptation, and Mitigation, Vol. 4., Handbook of Climate Change and Agroecosystems: The Agricultural Model Intercomparison and Improvement Project Integrated Crop and Economic Assessments, Part 1. (UK: Imperial College Press).
  • Estes, L.D., Bradley, B.A., Beukes, H., Hole. D.G., Lau, M., Oppenheimer, M., Schulze, R., Tadross, M., Turner, W. Comparing mechanistic and empirical model projections of crop suitability and productivity: Implications for projecting ecological change. 2013. Global Ecology and Biogeography 22, 1007-1018.
  • Estes, L.D., Reillo, P.R., Mwangi, A.G., Okin, G.S., and Shugart, H.H. (2010). Remote sensing of structural complexity indices for habitat and species distribution modeling. Remote Sensing of Environment 114, 792-804.