ICAR NADRES v2 Forecasting Input Dataset

District-level climatic, ecological, and outbreak input matrix for NADRES v2 livestock disease risk forecasting across 755 districts, 36 states and 15 diseases.

About Dataset

This sample data was generated in April 2026 for June 2026 disease risk prediction. This dataset is a structured, district-level input matrix used by NADRES v2 for forecasting livestock disease risk across India, spanning all 36 states and UTs, 755 districts, and 15 major livestock diseases at national scale (sample data set shared here covers Karnataka with different disease). The 127 variables include 19 meteorological parameters, 5 remote-sensing indicators (NDVI, EVI, LAI, LST, PET), Delta variables capturing long-term climatic trends, climatic event flags, ecological parameters (soil pH, elevation, forest fire frequency), one-month lagged versions of all variables, and livestock population data, integrated across four temporal layers: static average (2011-2022), dynamic averages for 2023-24 and 2025, and predicted values for 2026. Outbreak data from the NADRES database covering a rolling 10-year period is aggregated at district level and converted to a binary variable: 0 for no outbreak and 1 for one or more outbreaks, which serves as the response variable for model building. A monthly score system further scales 10-year attack data for all 15 diseases to a range of 1-10, standardising comparability across diseases and months. Data is available under ICAR Data Use License. This data set belongs to ICAR-NIVEDI, Bengaluru and creator for this data is Dr KP Suresh and colleagues. For access to whole data, a request may be made to ICAR-National Institute of Veterinary Epidemiology and Disease Informatics.

Purpose of Dataset

This Dataset Supports The Training, Validation, And Operational Deployment Of Nadres V2, An Ai-based Livestock Disease Forewarning System Developed By Icar-nivedi. It Is Used To Build And Evaluate A 25-model Ensemble Of Machine Learning, Bayesian, And Deep-learning Models That Classify District-level Disease Risk Into Six Categories From Very High Risk To No Risk, Up To Two Months In Advance, Across 755 Districts And 15 Major Livestock Diseases At National Scale. The Dataset Is Intended For Researchers, Veterinary Epidemiologists, Agri-tech Developers, And Public Health Institutions Working On Ai-enabled Animal Disease Surveillance, Climate-driven Disease Modelling, Early Warning Systems, And Decision-support Tools For Livestock Health Management Across India.