Skip to main content
ARS Home » Pacific West Area » Maricopa, Arizona » U.S. Arid Land Agricultural Research Center » Plant Physiology and Genetics Research » Research » Research Project #443512

Research Project: A Satellite-based Crop Monitoring and Forecasting System to Enhance Food Security in Nepal

Location: Plant Physiology and Genetics Research

Project Number: 2020-21600-001-006-I
Project Type: Interagency Reimbursable Agreement

Start Date: Apr 24, 2023
End Date: Apr 23, 2025

The overall goals of this proposed project are to 1) develop an operational web-based Satellite Based Agricultural Monitoring and Forecasting (SAMF) tool to provide in-season qualitative and quantitative yield metrics for rice and wheat systems in Nepal at multiple spatial scales, and integrate this system into the existing ICIMOD’s Food Security Information System and NASA-SERVIR crop monitoring and assessment service, and 2) engage with various stakeholders (e.g. local scientists) to deliver technical knowledge of SAMF tool's usage. Specific objectives of the proposed project are to: 1. Building on existing NASA-SERVIR services (e.g. Regional Land Cover Monitoring System ( and Geofairy) and crop type mapping algorithms developed under previous NASA LCLUC project (Bandaru 2018), we will produce early season rice and wheat masks using combination of optical (Landsat 8&9, sentinel-2) and SAR (Sentinel-1) imagery for 2019-2021 at 30m resolution. These data will lead to crop-specific input and model parameterization for SAMF framework. 2. Bias correct Subseasonal Experiment (SubX) S2S forecasts and downscale them to 5 km spatial resolution using available robust statistical downscaling approaches (e.g. NCAR’s Generalized Analog Regression Downscaling (GARD) tool) by taking advantage of SALDAS downscaling efforts under previous SERVIR project. 3. Develop web interface to implement SAMF at 1km pixel level over Nepal and produce probabilistic qualitative and quantitative yield outlooks for rice and wheat at sub-district (i.e. palika) level on monthly basis, and evaluate its performance for previous years (2019-2021) through quantifying the uncertainty and skill of forecasts at pixel and regional scales. 4. Engage with government and non-government organizations, with the support of ICIMOD, involved in the Nepal food security programs (e.g., NeKSAP) through conducting in-person and virtual workshops and meetings to a) facilitate the use of web-based SEMF to conduct their analyses to determine precise food security indicators (e.g., Cereal Food Sufficiency Ratio), and b) integrate data products into the ICIMOD’s Food Security Information System for Nepal. We will also share our products with NASA Harvest and other NASA application programs to use in their capacity building programs.

Our research comprises four major tasks. Task 1 involves in-season crop type mapping using multi-temporal, multi-source SAR and optical satellite imagery (e.g. Landsat-8, Sentinel-1&2) to develop early season rice and wheat crop map masks at 30m for 2019 and 2021 in the season. Time-series of surface reflectance and backscatter (for SAR) collected during early and mid-season will be used to accurately classify major crop areas. For this, project will leverage efforts under a related, ongoing NASA LCLUC project (Bandaru, 2018) by which we will use in-house machine learning processing workflows; e.g., random forests and neural networks, which are successfully applied in Thailand to map sugarcane and rice crops. This previous research showed that mid and early season SAR imagery alone can be used for relatively accurate mapping of paddy rice. Here, ground truth data to train and validate classifiers will be obtained through field campaigns and expert photo-interpretation. Task 2 assemble reanalysis and S2S forecasts required for SAMF framework. We will use SALDAS developed as part of ongoing NASA-SERVIR project along with IMDAA data to produce ensembles of weather variables required for SAMF. Further, we will bias-correct NMME S2S forecasts and dynamically and statistically downscale them for achieving regional specificity at local scales. Task 3 integrates the spatial databases and develops web-based SAMF tool to estimate the bi-weekly crop condition and monthly crop yields for rice and wheat at 1 km spatial resolution and at sub-district levels. The EPIC model used in SAMF simulates biophysical and biogeochemical ecosystem processes as influenced by climate, landscape, soil and management conditions, and it has been well tested and widely used for simulating crop productivity under different cropping systems, management regimes, locations and at various scales (Gassman et al., 2005; Bandaru et al., 2013). Using field observations, we will calibrate the model for estimating crop yields and biomass, and integrate all spatial databases (e.g., crop cover, weather, soil and irrigated maps) and develop web-based system. Daily biomass estimates will be used to provide crop condition. Task 4 engages government and non-government agencies involved in the Nepal food security programs (e.g.NeKSAP) to assess utility of our products in their existing services and transfer the knowledge through workshop allowing local scientists to use SAMF. We will also engage with NASA Harvest program (Dr. Bandaru, a research partner of NASA harvest) and share the data products to be used in their capacity building programs. Further, we will participate in the NASA SERVIR and SARI capacity building training programs.