Amy L. Kaleita

Relationship Between Soil Moisture Content and Soil Surface Reflectance

Depending on the topography and soil characteristics of an area, soil moisture, an important factor in crop productivity, can be quite variable over the land surface. Thus, a method for determination of soil moisture without the necessity for exhaustive manual measurements would be beneficial for characterizing soil moisture within a given region or field. In this study, soil surface reflectance data in the visible and near-infrared regions were analyzed in conjunction with surface moisture data in a field environment to determine the nature of the relationship between the two, and to identify potential methods for estimation of soil moisture from remotely sensed data in these wavelengths. Results indicate that it is feasible to estimate surface (0 to 7.6 cm) soil moisture from visible and near-infrared reflectance, although estimating moisture regimes rather than precise water content is perhaps more likely. Furthermore, an exponential model was appropriate to describe soil moisture from spectral reflectance data. In particular, the visible region of the electromagnetic spectrum works well with such a model. A partial least squares analysis with improved R2 values over the single-band models indicated that mulitspectral data may add more useful information about soil moisture as compared to single-band data. The results also suggested that the performance of reflectance models for moisture estimation is a function of soil types; the estimation results were better for the lighter of the two soils in this study.

FIELD CALIBRATION OF THE THETA PROBE FOR DES MOINES LOBE SOILS

Knowledge of soil moisture is needed to understand crop water use, hydrology, and microclimate. A reliable, rapid technique is needed, and recently an impedance soil moisture probe (Theta Probe) has been accepted by the scientific community. The purposes of this study were to calibrate the probe for soils of Central Iowa through field sampling, to determine the number of samples needed for calibration, and to determine the effect of temperature on calibration. Laboratory calibration was conducted on Des Moines lobe soils across a range of water contents and temperatures. Including a temperature term increased the R 2 from 0.85 to 0.87. Field calibration was based on Theta Probe measurements on similar soils combined with gravimetric sampling and soil temperature determination. Although some scatter existed, the field calibration was adequate for Iowa soils (R 2 = 0.77). Inclusion of temperature did not significantly improve the calibration for the field data. To determine the appropriate number of samples needed for the field calibration, regression equations were determined from sample numbers ranging from 2 to 89, and the standard error was determined for each. Based on the standard error analysis, 20 samples was an adequate number, with no further improvement for additional data points.

USING CROSS-VALIDATION TO EVALUATE CERES-MAIZE YIELD SIMULATIONS WITHIN A DECISION SUPPORT SYSTEM FOR PRECISION AGRICULTURE

Crop growth models have recently been implemented to study precision agriculture questions within the framework of a decision support system (DSS) that automates simulations across management zones. Model calibration in each zone has occurred by automatically optimizing select model parameters to minimize error between measured and simulated yield over multiple growing seasons. However, to date, there have been no efforts to evaluate model simulations within the DSS. In this work, a model evaluation procedure based on leave-one-out cross-validation was developed to explore several issues associated with the implementation of CERES-Maize within the DSS. Five growing seasons of measured yield data from a central Iowa cornfield were available for cross-validation. Two strategies were used to divide the study area into management zones, one based on soil type and the other based on topography. The decision support system was then used to carry out the model calibration and validation simulations as required to complete the cross-validation procedure. Results demonstrated that the models ability to simulate corn yield improved as more growing seasons were used in the cross-validation. For management zones based on topography, the average root mean squared error of prediction (RMSEP) from cross-validations was 1460 kg ha-1 when two growing seasons were used and 998 kg ha-1 when five years were used. Model performance was shown to vary spatially based on soil type and topography. Average RMSEP was 1651 kg ha-1 on zones of Nicollet loam, while it was 496 kg ha-1 on zones of Canisteo silty clay loam. Spatial patterns also existed between areas of higher RMSEP and areas where measured spatial yield variability was related to topography. Changes in the mean and variance of optimum parameter sets as more growing seasons were used in cross-validation demonstrated that the optimizer was able to arrive at more stable solutions in some zones as compared to others. Results suggested that cross-validation was an appropriate method for addressing several issues associated with the use of crop growth models within a DSS for precision agriculture.

Development of an android app to estimate chlorophyll content of corn leaves based on contact imaging

An app was implemented on android phone to estimate chlorophyll content of corn leaf.New method of imaging, contact imaging, was used to reduce effects of real conditions.Both linear and neural network models were developed to estimate SPAD values.Stepwise and sensitivity analysis were used to extract superior features.The app is a practical and low-cost alternative to measure SPAD. In this study a new android app for smartphones to estimate chlorophyll content of a corn leaf is presented. Contact imaging was used for image acquisition from the corn leaves which captures the light passing through the leaf directly by a smartphones camera. This approach would eliminate the needs for background segmentation and other pre-processing tasks. To estimate SPAD (Soil Plant Analysis Development) values, various features were extracted from each image. Then, superior features were extracted by stepwise regression and sensitivity analysis. The selected features were finally used use as inputs to the linear (regression) and neural network models. Performance of the models was evaluated using the images taken from a corn field located in West of Ames, IA, USA, with Minolta SPAD 502 Chlorophyll Meter. The R2 and RMSE values for the linear model were 0.74 and 6.2. The corresponding values for the neural network model were 0.82 and 5.10, respectively. Finally, these models were successfully implemented on an app named SmartSPAD on the smartphone. After installing the developed app on the smartphone, the performance of the models were evaluated again using a new independent set of data collected by SmartSPAD directly from maize plants inside a greenhouse. The SmartSPAD estimation compared well with the corresponding SPAD meter values (R2=0.88 and 0.72, and RMSE=4.03 and 5.96 for neural network and linear model, respectively). The developed app can be considered as a low cost alternative for estimating the chlorophyll content especially when there is a demand for high availability.

Technical Note: Field-Scale Surface Soil Moisture Patterns and Their Relationship to Topographic Indices

Understanding variability patterns in soil moisture is critical for determining an optimal sampling scheme both in space and in time, as well as for determining optimal management zones for agricultural applications that involve moisture status. In this study, distributed near-surface gravimetric soil moisture samples were collected across a 3.3 ha field in central Illinois for ten dates in the summer of 2002, along with dense elevation data. Temporal stability and consistency of the moisture patterns were analyzed in order to determine a suitable grid size for mapping and management, as well as to investigate relationships between moisture patterns and topographic and soil property influences. Variogram analysis of surface moisture data revealed that the geospatial characteristics of the soil moisture patterns are similar from one date to another, which may allow for a single, rather than temporally variable, variogram to describe the spatial structure. For this field, a maximum cell size of 10 m was found to be appropriate for soil moisture studies on most of the sampling occasions. This could indicate an appropriate scale for precision farming operations or for intensive ground sampling. While some areas had consistent behavior with respect to field mean moisture content, no conclusive relationships between the overall patterns in the moisture data and the topographic and soil indices were identified. There were, however, some small but significant correlations between these two sets of data, particularly plan and tangential curvature, and also slopes. In areas of convergent flow, moisture content exhibited a slight tendency to be wetter than average. There also seemed to be a small influence of scale on the relationship between moisture patterns and topographic curvatures.

Using Aerial Hyperspectral Remote Sensing Imagery to Estimate Corn Plant Stand Density

Since corn plant stand density is important for optimizing crop yield, several researchers have recently developed ground-based systems for automatic measurement of this crop growth parameter. Our objective was to use data from such a system to assess the potential for estimation of corn plant stand density using remote sensing images. Aerial hyperspectral remote sensing imagery was collected on three dates over three plots of corn in central Iowa during the 2004 growing season. The imagery had a spatial resolution of 1 m and a spectral resolution of 3 nm between 498 nm and 855 nm. A machine vision system for early-season measurement of corn plant stand density was also used to map every row of corn within the three plots, and a complete inventory of corn plants was generated as a rich ground reference dataset. A principal component regression analysis was used to assess relationships between plant stand density measurements and principal components of hyperspectral reflectance for each plot, on each image collection date, and at three different spatial resolutions (2, 6, and 10 m). The maximum R 2 for regressions was 0.79. Estimates of corn plant stand density were best when using imagery collected at the later vegetative and early reproductive corn growth stages. Quantization effects due to row width complicated corn plant stand density estimates at 2 m spatial resolution, and better estimations were typically seen at resolutions of 6 m and 10 m. Among the different cases of plot, image date, and spatial resolution, the principal components of reflectance most highly correlated with plant stand density were able to be classified into four distinct types, denoted as types A, B, C, and D. Type A principal components contrasted all available visible red wavelengths with all available near-infrared wavelengths. Type B principal components contrasted green wavelengths (531 to 552 nm) plus shorter wave near-infrared (759 nm) with red wavelengths (675 to 693 nm) plus longer wave near-infrared (852 nm). Type C principal components summed green wavelengths (528 to 546 nm) and near-infrared wavelengths (717 to 855 nm). Type D principal components contrasted blue/green wavelengths (498 to 507 nm) with the red edge (717 nm). Remote sensing can be best used to estimate corn plant stand density at mid-season as long as plant stand variability exists and variability due to other factors is minimal.

Identifying sampling locations for field‐scale soil moisture estimation using K‐means clustering

Identifying and understanding the impact of field-scale soil moisture patterns is currently limited by the time and resources required to do sufficient monitoring. This study uses K-means clustering to find critical sampling points to estimate field-scale near-surface soil moisture. Points within the field are clustered based upon topographic and soils data and the points representing the center of those clusters are identified as the critical sampling points. Soil moisture observations at 42 sites across the growing seasons of 4 years were collected several times per week. Using soil moisture observations at the critical sampling points and the number of points within each cluster, a weighted average is found and used as the estimated mean field-scale soil moisture. Field-scale soil moisture estimations from this method are compared to the rank stability approach (RSA) to find optimal sampling locations based upon temporal soil moisture data. The clustering approach on soil and topography data resulted in field-scale average moisture estimates that were as good or better than RSA, but without the need for exhaustive presampling of soil moisture. Using an electromagnetic inductance map as a proxy for soils data significantly improved the estimates over those obtained based on topography alone.

Identification of Optimal Sampling Locations and Grid Size for Soil Moisture Mapping

The most robust site-specific management system is based in part on high quality maps of soil data. However, creating useful, accurate maps of soil information can be complicated by the natural variability of soil characteristics. This study addresses optimal sampling locations and grid sizes for soil moisture mapping, which is a valuable input for site-specific applications that depend on moisture status, such as precision irrigation and application of chemicals which require moisture transport into the root zone. Variogram analysis of surface moisture data for a central Illinois field revealed that the geospatial characteristics of the soil moisture patterns are similar from one date to another, which may allow for a single, rather than temporally variable, variogram to describe the spatial structure. For this field, a maximum cell size of 13 meters was found to be appropriate for soil moisture. This could indicate an appropriate scale for precision farming operations or for intensive ground sampling. The temporal stability of moisture patterns was studied in order to identify optimal sampling points for field-average soil moisture. Such points were identified by calculating their deviation over time from field average. Topographic data were analyzed to determine if these sampling points could be identified from time-invariant data. While no topographic indices were identified as being strong indicators of these locations, the points tended to be located in areas that were neutral in plan curvature compared to the field average.

Feasibility of using smart phones to estimate chlorophyll content in corn plants

New spectral absorption photometry methods are introduced to estimate chlorophyll (Chl) content of corn leaves by smart phones. The first method acquires light passing through a leaf by smartphone camera, compensating for differences in illumination conditions. In order to improve performance of the method, spectral absorption photometry (SAP) with background illumination has been considered as well. Data were acquired by smartphone camera in Iowa State University maize fields. Various indices were extracted and their correlation with Chl content were examined by Minolta SPAD-502. Hue index in SAP reached R2 value of 0.59. However, with light-aided SAP (LASAP), R2 of 0.97 was obtained. Among traits, the vegetation index gave the most accurate indication. We can conclude that the high performance of LASAP method for estimating Chl content, leads to new opportunities offered by smart phones at much lower cost. This is a highly accurate alternative to SPAD meters for estimating Chl content nondestructively.

Genetic algorithms for Hyperspectral Range and Operator Selection

A novel genetic algorithm was developed using mathematical operations on spectral ranges to explore spectral operator space and to discover useful mathematical range operations for relating spectral data to reference parameters. For each range, the starting wavelength and length of the range, and a mathematical range operation were selected with a genetic algorithm. Partial least squares (PLS) regression was used to develop models predicting reference variables from the range operations. Reflectance spectra from corn plant canopies were investigated, with proportion of plants (1) with visible tassels and (2) starting to shed pollen as reference data. PLS models developed using the spectral range operator framework had similar fitness than PLS models developed using the full spectrum. This range/operator framework enabled identification of those spectral ranges with most predictive capability and which mathematical operators were most effective in using that predictive capability. Detection of operator locality may have utility in sensor and algorithm design and in developing breeding stock for other algorithms.