Product Line 5.2. Spatial analysis for effective technology targeting
In recent years, remote-sensing (RS) technology and geographic information systems (GIS) for spatial analysis have become widespread due to the improved availability of computing hardware and software at lower costs as well as improved access to spatial data due to rapid developments on the Internet and in spatial data infrastructure (SDI). These technologies can be used to monitor and evaluate agricultural systems to determine where and when (spatial and temporal) rice is grown and where crops are performing well or where they are not. The causes for these variations can be analyzed within GIS by combining RS-derived information with thematic layers on the socioeconomic and biophysical characteristics that are obtained from other sources.
Such mapping and monitoring of the biophysical and socioeconomic characteristics of rice-producing areas is key for developing effective targeting strategies for the dissemination of new technologies and sustainable crop management and diversification options. Modeling of the target domains and mega-environments of rice production/consumption and understanding their resilience under economic, demographic, and climate change scenarios will help to guide limited resources to achieve the greatest potential benefits.
Similarly, the identification of regions where there is an opportunity to expand the area of rice production is necessary as an input for policies that will deal with the growing demand for rice. Moreover, by placing agricultural systems in the context of a river basin or at a national or regional level, the impact of existing and potential rice production areas on the environment, such as degrading water quality or water availability, can be assessed.
The key message for this product line is the supportive role of GIS and RS activities to “provide actionable information to decision makers.”
Mapping and characterizing rice-growing areas. A principal activity in the spatial analysis of rice systems is to map them. On a regional or global scale, RS techniques using high-frequency MODIS imagery will allow rice areas and changes in rice areas to be mapped from year 2000 to now. Components of this analysis include (i) ground-truth surveys, (ii) the generation of “ideal” spectral and temporal signatures of rice cultivation under different conditions, (iii) supervised and unsupervised classification of MODIS time series, and (iv) validation against plot-level data and subnational statistics. Semi-automated classification techniques will be developed and run on high-performance computing (HPC) facilities such as the Amazon Elastic Computing Cloud (EC2) service to ensure timely provision of rice area maps across Africa and Asia at high spatial resolution.
Knowledge on rice phenology provides key information for integrated pest and invasive species management. Moreover, it can be used in crop modeling in a move toward spatially variable crop calendars that can vary from year to year instead of static country-level crop calendars. Rice phenology information can be derived from remotely sensed imagery once the imagery has been smoothed to remove artifacts and noise caused by pervasive cloud cover and atmospheric effects. Time-series smoothing and curve-fitting algorithms will be developed to derive these smooth temporal signals. The results will include (i) a library of rice signatures, (ii) key phenological information at high spatial resolution for Africa and Asia for use in crop and pest and disease models, and (iii) relevant syntheses of climate data from key phenological dates. Furthermore, failed seasons or changes in cropping intensity due to climate shocks or policy changes may be identified from this information and, in the longer term, climate change effects may also be observed.
Abiotic and biotic stresses. Monitoring and mapping drought and flood events—that have high spatial and temporal variability—requires a remote-sensing approach that can identify the onset and duration of stress-causing events and validate them against daily weather station data. RS methods to detect surface water at vulnerable stages of rice growth and vegetation stress from drought conditions will be improved and calibrated against subnational time series of rice statistics to develop a spatial database of the frequency, duration, and extent of drought and submergence to permit spatial targeting of new stress-tolerant varieties. The best available information on soil constraints (e.g., iron toxicity, salinity, and sodicity) will be compiled and standardized to assess the extent of these stresses on rice production.
The potential and actual impact of biotic stresses on rice yield will be assessed using a spatial model of potential epidemics in conjunction with existing and proposed pest and disease surveys.
Yield and yield gaps. Spatial crop modeling using daily climate data within a GIS environment is a strong tool for analyzing where crop production is close to its potential and where a significant yield gap exists. By combining these results with subnational yield statistics and the previously mentioned outputs on rice areas and their biotic and abiotic stresses, targeted locally adapted interventions can be formulated to close the yield gap.
Recommendation domains for technology targeting and delivery. A comprehensive database of relevant spatial layers and survey data for South Asia, Southeast Asia, and sub-Saharan Africa will be developed and will form the basis of an agroecological zoning model for rice cultivation and the multiscale modeling of target domains and rice mega-environments based on socioeconomic and biophysical factors and constraints. The resilience of these mega-environments under global change scenarios will be assessed and the results fed back into the decision process for future rice research investments and targeting of new technologies.
An inventory of potentially suitable rice ecologies in Africa will be completed. Activities will include ground-truth surveys, the development of a rice detection algorithm suited to African rice-growing environments, and adaptation of an existing model for mapping potential suitable areas for rice cultivation. Current activities on mapping biotic and abiotic stresses will continue.
5.2.1. Seasonally updated information on rice agroecologies
5.2.2. Maps of major rice-growing areas suffering from abiotic and biotic stresses
5.2.3. Identification and characterization of rice mega-environments for effective technology targeting
A strong network exists from partners in both the public and private sector. Most importantly, researchers from IRRI, AfricaRice, and CIAT will team up to combine their knowledge and develop innovative methodologies that are applicable on the three target continents. The seasonally updated information on rice agroecologies will be gathered in collaboration with partners involving other CGIAR centers, regional and national research centers, agricultural universities, and public- and private-sector research organizations. The research partners for the characterization of abiotic and biotic stresses are IRRI, AfricaRice, and regional centers. Results will be developed by IRRI and AfricaRice, while some of the secondary data collection will be conducted in collaboration with regional partners. The results will be shared with regional partners to identify or target stress-prone areas. Recommendation domains for technology targeting and delivery will be a joint research activity between IRRI and AfricaRice. Examples of national partners include Institut d’Economie Rural (Mali) and Université de Abomey-Calavi (Benin), etc. International partners include Wageningen University (The Netherlands), Cirad (France), FutureWater (The Netherlands), International Water Management Institute (CGIAR), etc. Private-sector partners include HPC providers Amazon and RADAR, and remote-sensing specialist Sarmap.
The short-term outcome is the use of results by GRiSP partners, NARES, ARIs, and donors to enable more effective targeting of rice technologies. The long-term outcome is increased benefits for the poor and the environment from more targeted and better funded investments in production and processing infrastructure and research. Results must be validated and demonstrated to add value over existing similar products. This requires an effective delivery and communication mechanism for these large data sets. This product line provides baseline data to themes 3, 5, and 6 in GRiSP. It will benefit from collaboration with HPC providers (e.g., Amazon) and key satellite imagery providers (ESA or JAXA).
Current funding comes from BMGF-GSR (2009-11), BMGF-STRASA (2008-10), Philippines RSSP (2009-11), and the Japan Rice Breeding project (2010-14). Approximately $600,000 of additional annual investment is needed, starting in 2011.