CLOSING THE LOOP FROM SATELLITE EMBEDDINGS TO WATER ALLOCATION: A GEOSPATIAL FOUNDATION MODEL FRAMEWORK FOR CROP SPECIFIC DROUGHT RISK ACCOUNTING IN THE WATER STRESSED IRRIGATED DRYLANDS OF THE ARAL SEA BASIN

Authors

  • Ismoilov Avazbek Abdusamat oʻgʻli BSc Agrologistics, 2nd year student International Agriculture University Author

Keywords:

Keywords: geospatial foundation models; satellite embeddings; crop type mapping; agricultural water accounting; evapotranspiration; probabilistic drought risk; irrigation; Aral Sea Basin; Central Asia; transfer learning; decision support

Abstract

Irrigated agriculture in the Aral Sea Basin consumes roughly ninety percent of all withdrawn water, yet the maps that should inform that allocation and the models that should govern it remain disconnected. Earth observation now produces accurate crop maps, and hydrological tools now produce probabilistic drought risk profiles, but the two communities rarely meet inside a single operational pipeline, and almost never in a setting that is simultaneously transboundary, dominated by fragmented and salinity affected fields, and chronically short of ground truth labels. This paper proposes and argues for a coupled framework that treats the modern geospatial foundation model not as an end in itself but as the connective tissue between the pixel and the water allocation decision. The central idea is to use analysis ready satellite embeddings, in particular the ten meter annual embedding fields produced by foundation models such as AlphaEarth Foundations, together with multitemporal radar and optical features, to derive crop specific classifications and evapotranspiration estimates with very few labels, and then to feed those quantities into a probabilistic water balance and drought risk accounting layer adapted from the D Risk family of decision tools. We specify the components, justify each design choice against the alternatives, and lay out a rigorous validation design with baselines, ablations, cross region transfer tests, and uncertainty propagation. We are deliberately ruthless about failure modes, including phenological information loss in annual embeddings, domain shift across the post Soviet irrigation mosaic, salinity confounding of vegetation signals, the political economy of transboundary data, and the equity consequences of acting on uncertain maps. We close with prioritized, justified recommendations for researchers, water authorities, and the wider Central Asian agricultural statistics enterprise. The contribution is not another crop map and not another water tool, but a defensible architecture for turning self supervised representations of the planet into accountable decisions about a scarce and contested resource.

References

1.Becker-Reshef, I., Vermote, E., Lindeman, M., Justice, C. (2010). A generalized regression based model for forecasting winter wheat yields in Kansas and Ukraine using MODIS data. Remote Sensing of Environment, 114(6), 1312 to 1323.

2.Brown, C. F., Kazmierski, M. R., Pasquarella, V. J., Rucklidge, W. J., Samsikova, M., Zhang, C., Shelhamer, E., Lahera, E., Wiles, O., Ilyushchenko, S., Gorelick, N., Zhang, L. L., Alj, S., Schechter, E., Askay, S., Guinan, O., Moore, R., Boukouvalas, A., Kohli, P. (2025). AlphaEarth Foundations: a geospatial embedding model and the Satellite Embedding dataset. Google and Google DeepMind. Earth Engine dataset GOOGLE/SATELLITE_EMBEDDING/V1/ANNUAL.

3.Chengot, R., Knox, J. W., Coxon, G., Cojocaru, G., Holman, I. P. (2023). An enhanced version of the D Risk decision support webtool for multiscale management of water abstraction and drought risks in irrigated agriculture. Computers and Electronics in Agriculture, 204, 107516.

4.Defourny, P., Bontemps, S., Bellemans, N., Cara, C., Dedieu, G., Guzzonato, E., Hagolle, O., et al. (2019). Near real time agriculture monitoring at national scale at parcel resolution: performance assessment of the Sen2 Agri automated system in various cropping systems around the world. Remote Sensing of Environment, 221, 551 to 568.

5.Dimou, S., Stamou, A., Kaimaris, D., Patias, P. (2021). Mapping and monitoring of agricultural land with time series satellite data. EARSeL Symposium proceedings.

6.Gallego, J. (2017). Use of remote sensing for the design of sampling frames. In Handbook on Remote Sensing for Agricultural Statistics. Global Strategy to improve Agricultural and Rural Statistics (GSARS).

7.Guzinski, R., Nieto, H., Ramo Sanchez, R., Sanchez, J. M., Jomaa, I., Zitouna Chebbi, R., Roupsard, O., Lopez Urrea, R. (2023). Improving field scale crop actual evapotranspiration monitoring with Sentinel 3, Sentinel 2, and Landsat data fusion. International Journal of Applied Earth Observation and Geoinformation, 125, 103587.

8.Jakubik, J., Roy, S., Phillips, C. E., Fraccaro, P., Godwin, D., Zadrozny, B., Szwarcman, D., et al. (2024). Prithvi EO 2.0: a versatile multitemporal foundation model for Earth observation applications. arXiv:2412.02732.

9.Raza, D., Shu, H., Nazeer, M., Aslam, H., Mirza, S., Xiao, X., Sardar, A., Aeman, H. (2024). Improved method for cropland extraction of seasonal crops from multisensor satellite data. International Journal of Remote Sensing.

10.TerraMind (2025). A multimodal any to any geospatial foundation model. IBM, European Space Agency, and Julich Supercomputing Centre.

11.Wang, Y., Braham, N. A. A., Xiong, Z., Liu, C., Albrecht, C. M., Zhu, X. X. (2023). SSL4EO S12: a large scale multimodal multitemporal dataset for self supervised learning in Earth observation.

12.Waldner, F., et al. (2015). Field size and the suitability of satellite sensors for agricultural monitoring, as discussed within the GEOGLAM framework.

13.International Water Management Institute (2025). Uzbekistan country profile: water scarcity and agricultural water use.

14.Nurbekov, A., et al. (2025). Water scarcity, crop diversification, and the cotton and winter wheat rotation in Uzbekistan. Frontiers in Sustainable Food Systems.

15.Carnegie Endowment for International Peace (2025). The Qosh Tepa canal and the prospect of a Central Asian water crisis.

Downloads

Published

2026-06-09