PRECISION AGRICULTURE INITIATIVE FOR TEXAS HIGH PLAINS

2001 ANNUAL COMPREHENSIVE REPORT

Texas Agricultural Experiment Station and Texas Agricultural Extension Service

Texas A&M University System

Principal Investigator: Robert J. Lascano (r-lascano@tamu.edu)

Texas A&M University – USDA-ARS

3810 4^th Street

Lubbock, TX 79415

Name Title/Agency Location

Cooperators: Jill Booker Res. Assoc., TAES Lubbock

Kevin Bronson Asst. Prof., TAES Lubbock

Ted Wilson Prof. and Res. Direc., Beaumont

Jim Bordovsky Res. Eng. Halfway

Project Title: The Balance and Variability of Water/Nitrogen within Large Agricultural Fields.

Primary Research Location: Lamesa and Helms Farm, TX

Project Objectives:

Objective 1: Quantify the spatial and temporal variability of factors that can be addressed by precision agriculture practices.

Objective 2: Develop and evaluate instrumentation and software to measure and analyze variability in crop production and plant response to that variability.

Objective 3: Evaluate the application of variable rate irrigation to cotton.

Reporting Period: 1 January 2001 — 31 December 2001

A. Summary of Progress (Address all applicable objectives; data must be included)

Objective 1: Quantify the spatial and temporal variability of factors that can be addressed by precision agriculture practices.

Lint Yield as a Function of Variability of Soil Factors in a Large Field. Field heterogeneity may affect yield potential, irrigation responses, and N transport. A water and N balance study was conducted in a large cotton field beginning in the spring 1998 to determine lint yield and N uptake potential related to local water and N input. Two irrigation levels, 50% ET and 75% ET, were applied from north to south across the field. In 1998, lint yields increased significantly with increasing irrigation (P > 0.0012). There was neither an effect of N inputs nor interaction between water and N inputs (P > 0.6780). However, the covariance of the model was significant on lint yield and N uptake (P > 0.0001). Therefore, a state-space approach was used to identify the spatial soil variability along transects. Lint yield, soil water, P₂O₅ content, and elevation were correlated in space. The state-space equations were determined through multivariate autoregressive processes to quantify the spatially correlated parameters that create spatial difference in a heterogeneous soil. Results showed that the natural landscape variability could affect lint yield and N uptake potential related to water and N use. Furthermore, the variability of our field measurements can be described with state-space models. These results are summarized in the publications listed below.

Objective 2: Develop and evaluate instrumentation and software to measure and analyze variability in crop production and plant response to that variability.

Multispectral Remote Sensing Related to Water and Nitrogen Use in Irrigated Cotton. Assessment of real–time crop and soil conditions using remotely sensed data promises to realize site-specific water and N application in large fields under semiarid conditions. Plant/soil reflectance and spectral vegetation index have been used in characterizing soil, water, nutrient, and plant development conditions, in forecasting crop yield, and in making day-to-day farm management decisions in irrigation and fertilization. The objectives of this study were (i) measurement of cotton/soil reflectance related to specific irrigation and N fertilization, (ii) determination of cotton/soil spectral and agronomic characteristics, (iii) assessment of variability in N status and lint yield across large cotton field, and (iv) to forecast irrigation and N fertilization using spectral vegetation index. These results are summarized in the publications listed below.

Objective 3:

Evaluate the application of variable rate irrigation to cotton.

Summary. The performance of a variable rate irrigation system on a center pivot was evaluated during the 2001-growing season. This performance included two objectives: 1) evaluation of delivery rates and positioning, and 2) phenological responses of cotton irrigated with variable amounts of water. Both objectives were preliminary and the evaluation was done at the Texas Agricultural Experiment Station at the Helm’s Farm, near Halfway, TX. Our results only concentrate on objective 2, as the first objective is part of an ongoing project by Jim Bordovsky.

For the purpose of variable rate irrigation a land area under the pivot was divided into 3 crop management zones (CMZ), each receiving different quantities, i.e., variable, of water during the growing season. Irrigation applications were determined according to the potential moisture holding capacity of the soil, which varied spatially due to variability in soil texture and topography. These two parameters were determined by taking soil samples to measure texture and elevation measurements to determine slope. Texture and slope were selected based on past research showing the importance of these two factors in determining cotton lint yield across a field. Crop water requirements were based on the evapotranspiration (ET), calculated from a Penman-Monteith equation. Our criterion for irrigation was to apply less water to areas with soils with a higher water holding capacity, which received a lower application rate, i.e., 60% ET, than to areas with soils with a lower water holding capacity, which received a higher application rate, i.e., 100% ET. In our research, each CMZ was divided in 2 parts, one receiving a base rate of 80% ET, or a variable rate according to the pre-determined soil water holding capacity.

To evaluate the cotton phenological response to variable rate irrigation, soil and plant measurements were done during the growing season (July – September). Soil measurement consisted of measuring soil moisture profiles every two weeks using neutron attenuation at 64 locations. Plant samples were collected twice a month at the same 64 locations, measuring plant height, number of leaves, leaf area index (LAI), number of bolls/squares, plant fresh biomass and plant dry weight. Additionally, plant reflectance was measured using a portable multi-radiometer. The crop was hand-harvested (24 October 2001) at the 64 sampling locations.

We evaluated our preliminary results by comparing for each CMZ the base and variable rate. Soil water content showed an increase for areas with higher application rates compared to the same area with lower rates. Results from LAI, plant height and crop reflectance measurements showed that plants under similar conditions showed better development, i.e., higher measurements of LAI, height and plant biomass, for higher application rates than for lower. Results showed that VR leveled out differences in the field. Measured differences in plant height and crop reflectance for cotton in different sections under the base application rate were similar to those under variable rates. Fresh plant biomass and number of bolls per plant showed that cotton under variable rate application with 60% ET on clay soils was water stressed. Lint yield showed a gain for variable rate on the slope with 100% ET and a loss on clay with 60% ET when compared to lint with the 80% base application rate. Lint yields irrigated with the 80% ET base application rate were fairly constant throughout the field. Results are preliminary due to equipment installation delays the cotton crop was irrigated late into the growing season. These experiments will be repeated during the 2002-growing season.

Introduction. Lack of water during the growing season is the limiting factor in the production of agricultural crops on the Texas High Plains (THP). The semi-arid climate in combination with persistent high winds, results in a high water demand. As a result of the climatic conditions the THP heavily rely on the use of groundwater for the production of crops. Gardner et al. (1996) expressed it, “…irrigation is the basis for an economy and a way of life”. Because of good soils and irrigation, the THP region, alone, would rank very high in agricultural production as a separate state in the U.S. or as another country in the world (Howell and Musick, 1997). Water for irrigation is obtained from the non-replenishable underground Ogallala aquifer. Water has been withdrawn from the aquifer for many years and as a result of overdraft, the future of Texas agriculture is uncertain. The effects of overdraft are likely to be the most severe in the southern Great Plains, where significant irrigated area may be removed from production within the next 20 years (Vaux et al., 1996). Since approximately 95% of the water pumped from the Ogallala is for irrigation (McReynolds, 2001), the severe decline of the water table prompted for changes in water management strategies. One of the changes is the use of deficit irrigation. Crop water demand is determined from the evapotranspiration (ET), consisting of the evaporation of soil water and the transpiration of a crop, as calculated with a Penman-Monteith equation. While regular irrigation applies water according to the ET, deficit irrigation exposes plants to a certain level of water stress by applying less than the water demand. Optimum crop yield can be obtained through allowing a certain level of yield reduction of a given crop while higher returns can be obtained with saved water, which can be diverted to irrigation of other areas (Kirda and Kander, 1998). Deficit irrigation is now widely practiced on the THP for cotton and literature regarding crop performance with various water treatments and crops is available (Lyle and Bordovsky, 1995; Bordovsky et al., 1992). Additionally, reducing irrigation water losses and increasing water use efficiencies requires a new approach towards irrigation technology: precision irrigation. Precision irrigation is a term given to irrigation methods that recognize and manage spatial and temporal variations in the soil-plant-atmosphere system. The objective of precision irrigation is to improve the control of water input, together with fertilizer and chemicals, to increase application efficiency and to reduce water use (Cook and Bramley, 1998).

The THP landscape is characterized by the presence of center pivot (CP) systems, which are considered an efficient means of irrigation. A center pivot is a moving irrigation system that rotates around a fixed point, a pivot (Broner, 1998). The application rate varies laterally because the CP lateral covers more area per unit length towards the outer end. Center pivots are classified according to pressure or nozzle type. High-pressure systems have a pressure of >3.5 kPa at the pivot; medium pressure systems work with pressures between 2.5 to 3.5 kPa, and low-pressure systems have < 2.5 kPa.

Originally, CP’s sprayed water under high pressures through the air. However, losses due to water evaporation and low uniformity resulted in low irrigation and application efficiencies. With the introduction of LEPA (Low Energy Precision Application), an irrigation system designed for maximizing irrigation efficiency and increasing rain utilization, a forward step towards significant savings in both water and energy requirements was made. LEPA is a self-moving circular or linear irrigation system that applies water for the production of crops and forage in an energy efficient manner. Water is distributed directly to the furrow at a very low pressure through drop tubes and orifice controlled emitters (Lyle and Bordovsky, 1980). Since water is not sprayed through the air by high pressures, water losses due to evaporation are significantly reduced. Additionally, the water is applied directly into the furrow, which leads to very high application efficiencies. Since the introduction of LEPA much research has been conducted on their efficiency. High application efficiencies from 95-98% for LEPA when compared to spray sprinkler methods were found, with negligible runoff or deep percolation (Schneider, 1999).

Present technology applies water uniformly throughout the field, even though the field may not be uniform. Although efficiency has improved due to LEPA, the spatial variability of the soil is not considered in the application of water (Kaspersma and Sonnemans, 2000).

A technology that is still in development and could potentially save more water is Variable Rate irrigation (VR). According to Bordovsky et al., 2000, areas within a field may require different quantities of irrigation due to variation in soil texture, depth of soil, and the effect of topography on rainfall runoff (lower elevations benefiting, or suffering, from runoff from higher elevations). Instead of treating entire fields uniformly, precision VR allows on-the-go adjustments in the rate of water delivery to specific portions of the field (Leidner, 2000). Non-uniform distribution of irrigation based on topography and water holding capacity of the soil profile could better utilize both rain and irrigation water and improve water use efficiencies. Although this does not necessarily mean water savings, re-allocation of the water may result in higher water use efficiencies and thus crop yield. Another advantage is that crops could also benefit by changing irrigation amounts at points (along the pivot lateral) as the relationships among soil chemical properties, pest infestations, production inputs, and crop yields are determined (Bordovsky, 2000).

Research reports on variable rate application of irrigation water are available. Kincaid (1997) developed a sprinkler head with a pin that moves smoothly in and out of the nozzle that reduced the flow to about 35 % of a nozzle at full capacity. Another type of variable flow sprinkler head consists of a digitally controlled metering device (Camp et al., 2000). Other types of variable rate systems use multiple manifolds to vary the flow-rate, in which several individually controlled manifolds each delivering discrete but different flow rates, and each designed at a certain water volume that could be combined into a range of flow rates at the drop (Bordovsky et al., 1992; Duke et al., 1992; Fraisse et al., 1992; Camp et al., 1998). Currently, most of this technology is not yet commercially available and their implementation is mainly restricted to research applications.

The two objectives of this research were to test and evaluate the functioning of a CP VR system designed by J.P. Bordovsky (a similar VR system on a linear move irrigation system was described in Bordovsky et al., 1992), and to find differences in crop phenological response between VR application, taking spatial variability in the field into account, and uniform water application, thus neglecting spatial variability in the field. However, in this report we will only discuss results from our second objective.

Material and Methods

Description of the Research Area

The research was conducted at the Texas Agricultural Experiment Station in Halfway at the Helms farm (long. 101° 57’, lat. 34° 11’), on the Texas High Plains (THP). Fig. 1 is a map of Texas showing annual precipitation and location of the experiment site in Halfway. The altitude of the research site is 1045 m above sea level. The semi-arid climate in this area is characterized by an average annual rain of 450 mm and an annual water evaporation of 2300 mm. For the production of crops, ground water is pumped from the Ogallala aquifer, and used for irrigation in almost 50% of the area under cultivation.

The soil of the research area consists of an Olton Loam (fine, mixed, thermic Aridic Paleustolls), characterized by the presence of a caliche layer that on average is 1 m below the surface. Caliche is a layer of indurated material near the surface, more or less cemented by secondary carbonates of calcium and magnesium, precipitated from the soil solution. The presence of such a layer affects penetration of roots into the soil and the movement of soil water.

For this research, a cotton crop variety Paymaster 2200, planted on 24 May 2001, was irrigated with a standard ¼ mile Zimmatic™ center pivot equipped with a VR system. The CP consisted of 8 spans each 48.8 m in length, covering a field with a radius of almost 400 m and 47.8 ha. On the outer 3 spans, a VR system was mounted. Except for the research area, the field was irrigated with a deficit, receiving 80% ET of the crop, which was uniformly applied with regular LEPA. The center pivot received water from 3 wells, which at top capacity pumped 2.3 m³/min, sufficient to irrigate the crop at 80% ET.

The research area, 4.8 ha, covered a 60º angle over the outer 3 spans at the south side of the field (Fig. 2) and received water with the VR system. The south side of the field had a lower elevation than the north side and there existed some variation in soil texture. According to spatial variability in soil texture and topography, the field –as described in Determination of the Crop Management Zones- for our purpose was divided into 3 different crop management zones (CMZ), each representing areas of similar characteristics. Application rates in this area varied between 60-100% of the ET.

The main advantage of the VR system compared to a conventional one is that it allows irrigation according to local requirements in the field by changing flow rates over part of a CP span (Fig. 2), instead of uniform application of irrigation water over the field. Different application rates for the CMZ’s were programmed into the control system for the control program of the center pivot allowed flow-rate changes in the field over 3º increments. The minimum area allowing variation in the irrigation rate, consisted of the length of 1 manifold over a 3º angle of the circle, equaling about 250 m², which is 0.025 ha.

The design of the VR system required water to be supplied from the pivot lateral through pressure regulators and solenoid valves to each of three manifolds comprising the manifold unit. Three manifold units were present per span. Hoses were used to direct water from the manifolds to the modified LEPA applicator. Initial nozzle sizes for each applicator provided rates of 1x, 2x, and 3x, which, in various combinations, which provided 6 discrete irrigation amounts ranging -with 20% increments- from 20% to 120% of a base irrigation rate. In total, the three spans were equipped with ninety-six specially constructed LEPA variable rate applicators (Bordovsky et al., 2000). To avoid large pressures variations in the main system, a pump at one of 3 wells is equipped with a pressure regulator to adjust and maintain the pressure at certain level at any delivered flow rate. In case of high flow rates, a booster pump is connected at the pivot to maintain system pressure. Before the water enters the manifolds and drops with pressure-designed nozzles, it passes pressure regulators to ensure the right pressure for the nozzles.

Data Collection

Data available for this research were collected during the summer of 2001, from May to October. In the first week of June, 64 aluminum neutron access tubes were installed pre-determined locations over 9 east west transects in the research area over the CP spans 6,7 and 8. Holes with a depth of 1.5 m were drilled using a Giddings Coring Drill. Additionally, soil samples were collected at 33 locations. The location of each access tube was determined by using a GPS Sattloc SLX.

Data on field characteristics such as elevation and slope were collected for 64 locations by using a field level (Model 8114, David White Instruments, Menomonee Falls, WI, USA). The elevation data were plotted in a GIS using Arcview 3.2, and an overview of the elevation and slope of the research area was created. Soil samples were collected over 4 depths, 0.2 m each, at 33 locations throughout the field for determination of the soil texture in the research area. Samples were air-dried and sieved to 2 mm. Soil textures were determined in the laboratory using the hydrometer method (Gee and Bauder, 1986). Soil texture together with the elevation was used to determine the crop management zones.

From July to September, crop samples were collected every other week at the 64 neutron access tube locations in the research area. From the collected plants, plant height, number of leaves, leaf area index (LAI) were measured. Number of squares and bolls were counted, and fresh and dry weight of the plants and bolls were measured using a scale. During July and September, a crop scan (MSR5 Multispectral Radiometer, Cropscan Inc.) was used to measure crop reflectance, which is an indicator for green mass of the plants. In October, lint yield samples were collected by hand as an indicator of lint yield.

Soil moisture content was measured at the 64 neutron access tube locations every 10-14 days during July to September. Using a neutron probe (CPN Corporation, Model 503 Hydroprobe, Martinez, CA), soil moisture measurements were taken in 0.3 m increments to 1.5 m depth.

During the irrigation season, VR flow rates, pressures and positioning system were collected. Flow rates were measured using a hang scale, stopwatch and bucket. Pressures were measured with a pressure gauge. For this, small Schreader valves (comparable to valve-stem in car tubes that can be fixed into almost any pipe conduit) were located throughout the system to which the pressure gauge was connected. For the positioning system, stakes were placed at known positions in the field, which were determined using a tape measure (every 3º in the research area in a transect from west to east). Weather data of the season 2001 were obtained from the Texas Agricultural Experiment Station in Halfway, located 1 km south of our research plot.

A. Determination of the Crop Management Zones (CMZ)

The CMZ were assigned using soil texture and slope.

Soil Texture

Soil samples were collected at the neutron access tube locations, each 30 m apart, of the middle (east west) transect of the CP spans 6, 7 and 8. In total, 33 samples of each 4 layers with 0.2 m increments (0-0.2 m, 0.2-0.4 m etc.) were used for the determination of the soil texture. Because of time constraints before the irrigation with VR started, only the two top depths, i.e., 0-0.2 m and 0.2-0.4 m, were analyzed for soil texture. At a later time during the research, also of the 2 other depths, i.e., 0.4-0.6 and 0.6-0.8 m, soil textures were determined. Results of this soil analysis for clay content were plotted in a GIS map.

Slope

The slope of a field may affect the application efficiency of irrigation and rainfall due to potential runoff by which the water is lost for a specific location. With the development of LEPA, a technology to reduce runoff in the furrow by creating small dikes was introduced (Lyle and Bordovsky, 1981), which was also implemented in the research area. Despite diked furrows, runoff can occur due to erosion or overtopping of the dikes. Additionally, alternate furrows application was practiced ((Bordovsky et al., 1984) where runoff may take place in the intermediate ‘dry not-diked furrows’ in case of rain.

Elevation measurements were done using a level and rod and then used to determine the field’s slope. Elevation was measured at all 64 locations where neutron access tubes were situated at the top of the ridge and data were plotted in a GIS. Using the inverse distance weighing method (IDW), a map was created with elevation of the research area in reference to the center pivot (Fig. 3). Because runoff is not caused by the slope alone but also by the slope of the furrow, the slope between every two neighboring points in a row was calculated individually. Dividing the measured difference in elevation between points by the distance of the points in the row, furrow slope data were obtained. Finally, points with the same slope were connected with a line, creating a map with furrow slope contour lines in 5 categories, i.e., 1%, 0.8%, 0.6%, 0.4% and 0%. Division into these categories was done for the practical reason that these were the numbers retrieved from the calculations that could be connected easily.

Layout Crop Management Zones

The actual CMZ’s were determined according to their water holding capacity. Soils containing more clay have a higher potential water holding capacity than those with sand. The same applies for soils without a slope, which have a higher potential to use the applied irrigation water and rain, than sites with a slope that are susceptible to runoff.

Roughly, the research area could -according to soil texture- be divided into two zones, a clayey zone with a high water holding capacity, containing > 40% of clay and a sandy zone with a lower water holding capacity, containing < 40% clay. In the case of slope of the furrows, it was assumed that runoff occurs mainly when the slope >0.5%. Considering the map with furrow slope contour lines, the area within the 0.6% or more contour line was defined as sufficient slope for runoff. The area with <0.6% slope was considered less susceptible to runoff, having a higher potential to utilize irrigation water and rain. Reason for using the 0.6% contour line was instead of a 0.5% was that it was easily derived from slope calculations, while the 0.5% contour line was not.

The different CMZ’s were assigned to receive an application rate according to their potential water holding capacity, based on their potential ET as determined with the Penman-Monteith equation. One of the objectives of VR was to vary flow rates without an increase of the total water demand of the crop by reallocation of application rates. The base application rate in this field consists of 80% ET. Areas with high potential water holding capacity received less than this 80% ET base rate and areas with low potential water holding capacity received more than the base rate (BR).

Finally, three CMZ’s were distinguished:

1. A clayey zone without slope, which was assigned to receive the lowest flow-rate of 60% ET,

2. A sandy zone without slope, which was assigned to receive an intermediate flow-rate of 80% ET,

3. A sandy zone with slope, which was assigned to receive the highest flow-rate of 100% ET.

To assure good monitoring of differences in crop phenological response and soil water content (SWC) between different water treatments, 32 out of 64 sample locations (neutron access tube locations) received the BR application and 32 received VR application. Of the 9 manifold units on 3 spans, 5 applied VR at 60%, 80% or 100% of the ET, depending on the location of the CP in the field. The 4 manifold units left, which was every other manifold, applied BR.

Crop Phenological Response

Methodology

B. Plant sampling

During July-September 2001, two weekly crop samples were taken. Six plants were cut at random at every of the 64 sample locations, shown in Fig. 3. Plant samples were processed in a laboratory. We measured plant height, number of leaves per plant, number of squares and bolls, plant fresh weight and dry weight, leaf area of the plants from which leaf area index (LAI) was calculated.

Plant Density

To determine the number of plants per m row for every sampling location, a plant count was done at the beginning of the growing season. At every sample location, 5 random plant counts were done by placing a 1 m rod along the row and counting of the plants within that m. Plant counts were average and this value was used for the calculations of LAI, fresh/dry weight of the crop and hand harvested cotton yield.

Leaf Area Index

The leaf area (LA) was measured using a LI-3100 Area Meter. The LAI was calculated as follows:

LAI = (plants per m²/ plants in sample) * (LA / area)

Where, LAI, is the leaf area index in m²/ m²; plants per m is the number of plants per m obtained from the plant count; plants in sample is the number of plants in sample (6 plants); LA is the leaf area measured with LI-3100; and Area is the area of 1 m row in m². The rows were 0.76 m apart.

Fresh/Dry Weight

Fresh and dry weights of the 64 samples were measured every sampling date. Before the dry weight was measured, all samples were dried in an oven at a temperature of 75º C. Depending on the size of the bolls in the sample, the samples were dried for 2-3 days. Fresh and dry weights are expressed in kg/ha, using the following formula:

Weight = (plants per meter/ plants in sample) * weight sample * 43.75

Where Weight is the fresh/dry crop weight in kg/ha; Plants per m, is the number of plants per m obtained from stem count; Plants in sample is the number of plants in sample (6 plants); Weight sample, is the fresh/dry weight of sample in g; and 43.75 is a conversion factor from weight in kg/m row to weight in kg/ha in m/m².

C. Crop Reflectance

Additionally to plant samples, crop reflectance measurements were used as an indicator of crop development. Due to memory lost we were only able to collect data for 2 sampling dates. A crop scan measures the crop reflectance at a height of 2 m above the crop canopy, viewing a ground area of 1 m diameter. Spectral radiometer readings in millivolt were converted into spectral reflectance as percentage of reflectance. Center wavelengths measured were 460, 485, 500, 560, 600, 660, 700, 750, 800, 830, 880, 940, 1100, 1260, 1480 and 1650 nm.

Different wavebands correlate with specific plant components. Li et al., 2000, found for example that near infrared reflectance (NIR) increased significantly with increasing irrigation, soil water content (SWC), plant fresh biomass (PFB) and plant N-content. One of the most significant relations between spectral and agronomic characteristics found by Li et al., 2000, was that NIR reflectance varied as a function of PFB and cotton lint yield was found to be related to the reflectance based normalized difference vegetation index (NDVI). The NDVI was determined as follows:

NDVI = (NIR – RED) / (NIR + RED)

Where NIR is the near infrared reflectance band between 797 and 829 nm, and RED is the red band between 648 and 674 nm

The NDIV and reflection of several wavelengths were plotted into a GIS to make differences in various CMZ’s the field visible. Additionally, reflections and NDVI were presented over a transect through the research area.

D. Soil Moisture Content

For the measurement of the soil moisture content of the soil, a neutron probe was used. Since a neutron probe is not very sensitive to small differences in soil moisture, measurements were done every 10-14 d. The neutron probe measures moisture in the soil by use of a radioactive source and neutron detector. The probe is lowered into an access tube and neutron counts were obtained for five depths of each 30 cm (layer 1 = 0-30 cm, layer 2 = 30-60 cm etc.). In 1997, the neutron meter was calibrated, resulting in the following linear relation (Bordovsky, personal communication):

SW = ((CR₅₀₃ * 7.10) + 0.43)/0.3

Where SW is the soil moisture content in m³/ m³; CR₅₀₃ is the count in specific soil layer/standard count. The standard count consists of 10 readings on a neutral (not lowered into the soil) position.

E. Harvest Data

On 24 October 2001, cotton lint was hand harvested at all 64 sampling locations. At each location, two areas, each 0.76 x 5.33 m, equaling 1/2500 ha (30 inch rows x 17.5 ft = 1/1000 acre) were harvested. While harvesting, the number of plants in a sample and the number of bolls of the first 6 plants in the sample were counted and recorded to be able to combine these data with data previously collected in the research. The samples were ginned at the Texas Agricultural Experiment Station, where the total, seed and lint weight were measured. These data were then plotted into a GIS.

Additionally, lint yield was harvested with a cotton stripper. The yield monitor records the weight of harvested cotton (including burr and seed) every second together with the GPS position in the field. From the hand harvesting, a lint yield factor was derived indicating the ratio between total weight and lint weight. Data recorded with the yield monitor were multiplied with this factor and plotted into a GIS.

F. Mapping in Arcview

Soil properties and plant development were mapped with Arcview, a geographical information system (GIS). In addition, information on elevation, soil (chemical) properties, soil water content, fresh and dry weight of plant matter, lint yield and NDVI’s, were also mapped using Arcview (v. 3.2), showing spatial variability of properties and characteristics in the research plots. Using the inverse distance weighted method (IDW), data collected at the 64 sample locations were interpolated, creating overview maps for the whole research area. The IDW option in Arcview interpolates data from the nearest neighbors and adds value to data from surrounding points by multiplying these with the inverse of the distance. The further a point, the smaller its weight in the interpolation. The number of neighbors was set at 6. The inverse power on the distance was set at 2. Although Arcview automatically determines map classifications, the number of classes was decreased to in most cases 5, and classifications were pragmatically reclassified into more equal and logic classes to assure a clear overview the maps created.

G. Data Analysis over a Transect

A transect from the SE to NW corner of the research area was delineated and used for spatial analysis of the research data. The transect was defined from low to high elevation, crossing all 3 CMZ’s, by connecting the mentioned corners with a line between them, crossing the middle point of the research area. Every point (sample location) in the research area was connected perpendicular to the transect. Cross-sections were given a number in chronological order from southeast to northwest, starting with 1 and ending with 64. Our defined transect combined points and data from different locations, and offered the possibility to take a different look at the dataset. Data analysis along this transect proved reliable. Besides the transect, regression curves between various variables were analyzed. To find relations between different soil, water and lint yield properties, several variables were analyzed:

Results and Discussion

Crop Phenological Response

G. Plant Sampling

Fresh Weight

Graphs of VR and BR plant fresh biomass (PFB) are presented in Fig. 4. Our results show large variations in PFB for VR and rather constant and higher values for BR. Fresh weight in the clay zone is for VR almost overall the lowest and for BR overall the highest, indicating that crop growth in the clay zone is affected due to the reduced 60% flow rate. Variation in PFB for cotton between VR and BR on sand and slope are small.

Dry Weight

Graphs of the plant dry weight (PDW), show the same trend as PFB, except for the fact that where curves of PFB decrease, the PDW increases. This is because bolls, which are about half of the total PDW, have a bulk density than leaves and stems. Overall, the dry biomass is 10-20% lower for VR than for BR cotton during the last sampling dates for which no direct explanation can be given.