Printer Friendly

X-treme Weather.

Forecasting effects on crops can help countries avoid disasters

Most mid-latitude tropical countries are prone to extreme weather that causes crop failure and leads to negative impacts on the economy.

An example is Mexico where 85 percent of cropland is in arid and semi-arid regions. Droughts there commonly cause severe crop yield reductions and puts national food security at risk. When extreme climate events such as El Nino Southern Oscillation (ENSO) phenomena are likely to occur, predicting crop yield changes can help determine how to modify national production of basic crops. Yield prediction data can also be used to implement emergency programs for alleviating food scarcity.

Today in Mexico, crop yield can be estimated under ENSO conditions four months in advance with 90 percent accuracy by applying process-based crop modeling.

Crop modeling has long been recognized by the Mexican government as beneficial for agriculture. However, uncertainties arise because model forecasts for ENSO phenomena El Nino and La Nina effects vary throughout the country for rainfall and temperatures that depart from mean values.

To remedy this situation, researchers at the National Laboratory for Crop Prediction, Agricultural and Forestry National Research Institute of Mexico (INIFAP) are collaborating with Texas Agricultural Experiment Station (TAES) and USDA Agricultural Research Service scientists. They are developing approaches for crop prediction and assessment at a national scale using remote sensing technology and modeling. The goal of this inter-institutional, multi-disciplinary research is to predict and assess crop yield for major staple crops -- maize and sorghum. Tools include satellite imagery based on the Normalized Difference Vegetation Index (NDVI) and data on crop variables measured in farm fields.

The project uses satellite information and field data taken from major agricultural valleys in Mexico. Research began with gathering information during the winter/spring 1999/2000 and spring/summer 2000 growing seasons. Samples came from roughly 3.7 million acres (1.5 million hectares) of corn under rain-fed and irrigated conditions. Researchers also sampled 2.1 million acres (850,000 hectares) of sorghum under rain-fed conditions in 10 states in northern, central and southern Mexico. A nationwide network of scientists monitored cropland areas at least 742.5 acres (300 hectares) each. They measured crop leaf area index (LAI), intercepting photosynthetic active radiation (PAR), phenological stage, plant height and canopy cover every 15 days. They also used global positioning systems (GPS) to identify study sites in latitude and longitude.

Remote sensing

Satellite remote sensing studies of natural resources usually rely on sensors such as Landsat Thematic Mapper (TM), Multispectral Scanner (MSS), System Probatoire pour l'Observation de Terre (SPOT) and Advanced Very High Resolution Radiometer (AVHRR). Users requiring high spectral and spatial resolutions tend to access TM, MSS or SPOT, while those needing daily access and large regional coverage use AVHRR. In the Mexican study, AVHRR was used because of the large area involved.

Studies have applied historical AVHRR data to detect natural environment changes, map land use and land cover, and monitor crop growth. AVHRR data gives researchers access to near real-time information. The AVHRR data used for the Mexican study were downloaded daily from the NOAA-14 satellite at the receiving station at Blackland Research and Education Center, Texas A&M University in Temple, Texas. The ground station received eight scenes per day from different satellites carrying AVHRR sensors.

Research applications using AVHRR data for natural environment monitoring conducted at Texas A&M's Spatial Sciences Laboratory of Forest Science Department were adopted in this study. INIFAP scientists used the developed algorithms to process satellite images and generate NDVI. The NDVI derived from satellite image data have been linked to vegetation condition and plant biomass on the land surface. NDVI values range from -1 to 1. Larger NDVI values indicate the land surface is covered with dense, healthy vegetation. Negative values indicate clouds, snow, water or a bright non-vegetated surface. A typical NDVI temporal profile for healthy green vegetation increases as plant cover increases in spring. It reaches a peak or plateau during summer and decreases with plant senescence in fall.

Cloud contamination in nearly every AVHRR scene decreases NDVI values so daily NDVI images in a time series do not always depict vegetation conditions during the growing season. To minimize cloud contamination effects, the maximum value compositing (MVC) procedure is used. However, 10-day composites are not cloud free and two- or three-week NDVI composites are cloud contaminated. This study's results helped to develop an alternative to solve this problem. The solution is to retain high-temporal resolution by detecting and removing cloudy pixels from daily AVHRR scenes and creating seven- or 15-day NDVI composites using only cloud-free data.

Another challenge involves daily data preprocessing, which provides near real-time information. The automated procedure converts raw digital counts of visible and near infrared channels to solar reflectance for application to crop yield prediction. The data preprocessing for each AVHRR scene involves radiometric calibration, atmospheric correction, geometric correction and cloud removal.

Crop growth modeling

Mathematical modeling used in this study to estimate crop yield is a strategy for learning and understanding how a crop behaves in real agro-systems defined by climate and soil characteristics. The Erosion Productivity Impact Calculator model is used by INIFAP scientists to simulate crop growth. LAI measured in the field is the main driving physiological variable determining crop yield at harvest. For model application, information on soil characteristics, climate and crop management are gathered from the field.

INIFAP scientists also linked satellite image data with crop growth modeling to assess yields in real time. Net primary productivity (NPP) is calculated from NDVT value and PAR. These data simulate weight increase in each plant organ considering the partitioning factor of biomass driven by the reported development stage of the crop and growth days.

Implications

This project's outcomes could be promising for technology application by agricultural agencies in Mexico, especially the Ministry of Agriculture, Rural Development, Food and Fishery (SAGARPA). In the past, SAGARPA received information on national crop production after harvest. Today, an early yield forecast is possible using satellite imagery and modeling.

The project's impact first became apparent when corn volume in northwestern Mexico was estimated for the 1999/2000 winter. The area comprising 493,000 acres (200,000 hectares) produces nearly 15 percent of the nation's corn. Applying techniques from this study provided corn yield forecasts three months before the official announcement of actual total regional corn production for the 1999/2000 growing season. Crop modeling and satellite image analysis identified an exceptional year for corn production. These estimates mirrored later official reports of actual regional corn yield based on regional harvest.

The study also identified a technological gap in corn production. Nearly 50 percent of crop areas monitored presented LAI values of 5 at silking stage, yielding less than the corn fields that reached an LAT of 6.5. This outcome illustrates crop yield prediction improvements the study made compared with traditional visual yield estimates by agricultural scientists.

Another achievement has been in the combined use of satellite data and crop growth modeling in assessing yield on a wide scale in real time. Results indicate that the study approach can assess yield with 89 percent accuracy in irrigated areas and 76 percent accuracy in rain-fed fields. The approach benefits assessments in large rain-fed areas with variation in size of farm fields and planting dates, genetic variability of corn germplasm and diversity in crop management practices. Compared with current SAGARPA methods, this approach can be applied to any rain-fed com valley regardless of size, planting data and applied management practices.

Modeling offers low-cost technology for a range of agricultural applications given large areas and crop diversity. It offers benefits for agricultural regions beleaguered by extreme climate events.

Raghavan Srinivasan is director and associate professor, Spatial Sciences Laboratory, department of forest science.
COPYRIGHT 2001 American Society of Agricultural Engineers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2001 Gale, Cengage Learning. All rights reserved.

 Reader Opinion

Title:

Comment:



 

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:forecasting effects of extreme weather on crops
Author:Srinivasan, Raghavan
Publication:Resource: Engineering & Technology for a Sustainable World
Geographic Code:1MEX
Date:May 1, 2001
Words:1289
Previous Article:Zeroing in on Forest Enemies.
Next Article:Stalking a New Fuel Source.
Topics:

Terms of use | Copyright © 2014 Farlex, Inc. | Feedback | For webmasters