SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Yield models and crop estimate accuracy are discussed within the Large Area Crop Inventory Experiment. The wheat yield estimates in the United States, Canada, and U.S.S.R. are emphasized. Experimental results design, system implementation, data processing systems, and applications were considered.
A technology assessment of the LACIE data processing and information systems was discussed during the Large Area Crop Inventory Experiment Symposium. Crop inventories of wheat yield in the United States as well as several other nations (such as the U.S.S.R., Canada, etc.) were discussed, along with the methodology involved in acquiring this data.
The author has identified the following significant results. LACIE acreage estimates were in close agreement with SRS estimates, and an operational system with a 14 day LANDSAT data turnaround could have produced an accurate acreage estimate (one which satisfied the 90/90 criterion) 1 1/2 to 2 months before harvest. Low yield estimates, resulting from agromet conditions not taken into account in the yield models, caused production estimates to be correspondingly low. However, both yield and production estimates satisfied the LACIE 90/90 criterion for winter wheat in the yardstick region.
The author has identified the following significant results. The Large Area Crop Inventory Experiment (LACIE), completed June 30, 1978, has met the USDA at-harvest goals (90% accuracy with a 90% confidence level) in the US Great Plains and U.S.S.R. for two consecutive years. In addition, in the U.S.S.R., LACIE indicated a shortfall in the '76-'77 wheat crop about two months prior to harvest, thus demonstrating the capability of LACIE to make accurate preharvest estimates.
The author has identified the following significant results. Labeling of wheat rather than total grains, particularly with only one acquisition, led to significant overestimates in some segments. The Image-100 software and procedures were written to facilitate classification of the LACIE segments but were not designed to record data for later accuracy assessment. A much better evaluation would have been possible if accuracy assessment data had been collected following each satisfactory classification.
The author has identified the following significant results. Estimates of soil moisture were obtained from visible, near-IR gamma ray and microwave data. Attempts using GOES thermal-IR were unsuccessful due to resolutions (8 km). Microwaves were the most effective at soil moisture estimates, with and without vegetative cover. Gamma rays provided only one value for the test site, produced by many data points obtained from overlapping 150 meter diameter circles. Even though the resulting averaged value was near the averaged field moisture value, this method suffers from atmospheric contaminants, the need to fly at low altitudes, and the necessity of prior calibration of a given site. Visible and near-IR relationships are present for bare fields but appear to be limited to soil moisture levels between 5 and 20%. The densely vegetated alfalfa fields correlated with near-IR reflectance only; soil moisture values from wheat fields showed no relation to either or near-IR MSS data.
An econometric model was run to compare the current forecasting system with a hypothetical (Landsat Follow-on) space-based system. The baseline current system was a hybrid of USDA SRS domestic forecasts and the best known foreign data. The space-based system improved upon the present Landsat by the higher spatial resolution capability of the thematic mapper. This satellite system is a major improvement for foreign forecasts but no better than SRS for domestic forecasts. The benefit analysis was concentrated on the use of Landsat Follow-on to forecast world wheat production. Results showed that it was possible to quantify the value of satellite information and that there are significant benefits in more timely and accurate crop condition information.
The papers outline the remote sensing activities being carried out in a number of countries throughout the world, and present details on a variety of individual projects. Topics studied include a worldwide approach to remote sensing and mineral exploration, a land use information system based on statistical inference, acoustic radar and remote sensing in the boundary layer, procedure for land-use analysis in developing countries, an airborne geochemical system, remote sensing of snowpack with microwave radiometers for hydrologic applications, wheat production forecasts based on Landsat data, airborne lidar aerosol measurements over the U.S. and Europe, Landsat inventory of agricultural and forest resources in Bangladesh, application of satellite imagery to flood plain mapping in Thailand, and vegetation mapping of Nigeria from radar.
The Suits bidirectional reflectance model (1972) for vegetation canopies was used to calculate reflectance values for the four Landsat spectral bands under a variety of canopy and observation conditions. Relationships between characteristics of developing wheat fields and associated reflectances were studied. Coincident measurements of crop reflectance and Landsat signals were compared using regression analysis techniques. The model was also applied to forest situations both to help establish the extent to which Landsat may provide direct information about conditions occurring beneath forest overstories and to help develop methods for reducing the effects of terrain slope and aspect in the processing of Landsat data. Finally, a relationship based on model calculations was used to adjust Landsat signal values and was found to reduce topography-related variability and improve subsequent classification performance.
A technique utilizing transformed Landsat digital data for detection of agricultural drought was empirically defined during the 1976 South Dakota drought. During 1977, the procedure was expanded to the Great Plains for evaluation as a technique for detecting and monitoring vegetative water stress over large areas. The technique, Green Index Number (GIN), uses Landsat digital data from 5 by 6 nautical mile sampling frames (segments) to indicate when the vegetation within the segment is undergoing drought. At known growth stages for wheat, segments were classified as drought or non-drought areas. The remote-sensing-based information was compared to a weekly ground-based index (Crop Moisture Index) provided by the United States Department of Commerce. This comparison demonstrated a high degree of agreement between the 18-day remote sensing technique and the ground-based weekly data. Maps based on GIN of parts of the USSR and Australia were produced with a two-week lag and later compared with other crop assessments of crop conditions in these areas. These maps were judged to be in general agreement with the other data sources.
Aims, history and schedule of the Large Area Crop Inventory Experiment (LACIE) conducted by NASA, USDA and NOAA from 1974-1977 are described. The LACIE experiment designed to research, develop, apply and evaluate a technology to monitor wheat production in important regions throughout the world (U.S., Canada, USSR, Brasil) utilized quantitative multispectral data collected by Landsat in concert with current weather data and historical information. The experiment successfully exploited computer data and mathematical models to extract timely corp information. A follow-on activities for the early 1980's is planned focusing especially on the early warning of changes affecting production and quality of renewable resources and commodity production forecast.
Results for both the United States and Canada show that econometric models can provide estimates of confusion crop ratios that are more accurate than historical ratios. Whether these models can support the LACIE 90/90 accuracy criterion is uncertain. In the United States, experimenting with additional model formulations could provide improved methods models in some CRD's, particularly in winter wheat. Improved models may also be possible for the Canadian CD's. The more aggressive province/state models outperformed individual CD/CRD models. This result was expected partly because acreage statistics are based on sampling procedures, and the sampling precision declines from the province/state to the CD/CRD level. Declining sampling precision and the need to substitute province/state data for the CD/CRD data introduced measurement error into the CD/CRD models.
Aggregation formulas are given for production estimation of a crop type for a zone, a region, and a country, and methods for estimating yield prediction errors for the three areas are described. A procedure is included for obtaining a combined yield prediction and its mean-squared error estimate for a mixed wheat pseudozone.
The design, implementation, and operational functions of the three phases of LACIE supported the data needs of all other elements of the project and required several types of data in addition to LANDSAT multispectral digital data. The nonelectronic data base consisted of statistical data, printed reports, periodicals, ground observed data received from intensive test sites and operational segments, and full-frame multispectral scanner CIR photographs. The following data were collected for the test sites in the United States and Canada: land use inventories, periodic crop observations, solar radiometer measurements, rainfall, and wheat yield for selected fields.
The organization and products of the yield estimation subsystem (YES) are described with particular emphasis on meteorological data acquisition, yield estimation, crop calendars, weekly weather summaries, and project reports. During the three phases of LACIE, YES demonstrated that it is possible to use the flow of global meteorological data and provide valuable information regarding global wheat production. It was able to establish a capability to collect, in a timely manner, detailed weather data from all regions of the world, and to evaluate and convert that data into information appropriate to the project's needs.
The primary responsibility of the crop assessment subsystem (CAS) during the three phases of LACIE was to produce crop reports that included estimates of wheat area, yield, and production, as well as a specified set of associated statistical descriptors. The operations of CAS are described with emphasis on sampling strategy, input/output data, evolution of aggregation/reporting system capabilities, and CAS aggregation procedures.
The accuracy and reliability of LACIE estimates of wheat production, area, and yield is determined at regular intervals throughout the year by the accuracy assessment subsystem which also investigates the various LACIE error sources, quantifies the errors, and relates then to their causes. Timely feedback of these error evaluations to the LACIE project was the only mechanism by which improvements in the crop estimation system could be made during the short 3 year experiment.
Meteorological data and LANDSAT spectral data from growing regions in the U.S. Great Plains, the U.S.S.R., and Canada were used to assess growing conditions and to document where anomalies such as drought, floods, and freezes were impacting the crop yield and appearance of spring and winter wheat. In the United States, the weekly rainfall and temperature data were used to estimate soil moisture, which was then related to crop needs by a crop moisture index. The transformation of LANDSAT digital data into a green index number provided a procedure whereby data from a LACIE segment could be classified as drought affected or not. The growing seasons encountered in each LACIE country during the three phases are described.