Rice Yield Prediction via Localized DSSAT Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining rice target yield and nitrogen fertilizer amount are limited by their reliance on historical data, lack of precise soil data, and inability to accurately account for climate variability, leading to inaccurate predictions and inefficient fertilizer use.
Innovation Solution
A soil-climate intelligent type determining method that integrates a climate year-type simulation system and 100 m×100 m plot-scale soil property simulation model, using historical and real-time weather data, and nitrogen fertilizer gradient test data to predict target yields and recommend nitrogen application amounts through a localized crop growth model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If crop growth models are used to determine target yield, then prediction accuracy is improved, but data collection and model parameter calibration become time-consuming and labor-intensive
Solution Approach 1:
The patent performs preliminary action by collecting and processing soil data, weather data, and variety data before the growing season, and pre-calibrating model parameters using historical data. This preparation work is done in advance so that during the actual growing season, the model can quickly generate target yield predictions without requiring extensive real-time data collection and calibration, thus resolving the contradiction between accuracy and time consumption
Solution Approach 2:
The patent uses historical weather data and soil data as copies or proxies for current conditions when calibrating models. By using representative historical data that captures typical growing conditions, the model can be calibrated without requiring extensive real-time measurements, reducing the time and labor needed while maintaining prediction accuracy
2Ease of operation
If average weather data from past years is used to drive crop models, then model operation is simplified, but accuracy for specific climate years and small-scale plots is reduced
Solution Approach 1:
The patent applies local quality by transitioning from regional average weather data to plot-specific microclimate data. It uses weather station data from locations close to each plot, and incorporates local soil property variations at the plot level. This localized approach maintains ease of operation by using standardized model procedures while significantly improving accuracy for specific climate years and small-scale plots by capturing local variations in weather and soil conditions
Solution Approach 2:
The patent segments the weather data and soil data from regional to plot-specific levels. Instead of using uniform average data for entire regions, it divides the area into smaller plot units and assigns specific weather and soil characteristics to each plot. This segmentation allows the model to operate simply while accounting for local variations, thereby improving prediction accuracy for specific conditions
3Quantity of substance
If large-scale soil property data is used, then data availability is improved, but plot-scale yield simulation accuracy is insufficient
Solution Approach 1:
The patent segments soil property data from large-scale regional data to plot-specific data. It divides the study area into individual plots and assigns soil properties to each plot based on available data and spatial interpolation. This segmentation maintains data availability by using existing large-scale soil surveys while improving plot-scale accuracy by capturing local soil variations that affect yield simulation
Solution Approach 2:
The patent uses spatial interpolation methods as intermediaries to bridge between large-scale soil survey data and plot-specific requirements. The interpolation process generates continuous soil property surfaces that can be sampled at plot locations, providing plot-scale soil data without requiring direct measurements at every plot. This intermediary approach maintains data availability while improving the precision needed for plot-scale yield simulation
Data Source
AI summary
A soil-climate intelligent type determination method for rice target yield and nitrogen fertilizer amount includes steps: construct a basic database; obtain historical weather data from the region where the rice planting area is to be determined over the past few years, and obtain real-time GFS weather data of the region in a predicted year; obtain 100 m×100 m grid sampling point soil data and a rice field vector layer of rice planting area to be determined; obtain at least three years of nitrogen fertilizer gradient test data of the rice planting area to be determined, rice variety information and management data of each farmer in each year; obtain accurate weather data of the prediction year; prepare soil data of 100 m×100 m plot-scale; localize genetic parameters of the rice growth model DSSAT; and run the localized DSSAT model to predict the plot-scale target yield and recommend the nitrogen application amount.