| Title |
Investigators | Department | Objectives | Approach Keywords | Progress Reports | Impact Statements | Publications | |
Project * COL00736 | |
| Title | *Field Scale Monitoring and Modeling of Salinity |
| Investigator(s) | Garcia, LA |
| Department | Civil Engineering |
| Objectives | The objective of this project is to continue detailed data collection activities at the field scale for soil salinity, depth to groundwater, groundwater quality, rainfall amounts, evapotranspiration, irrigation water quality (including selenium), and crop yield in ten fields located in the Lower Arkansas River Basin. The data collected will be used to determine the severity of crop losses due to salinity and waterlogging using a Geographic Information Systems (GIS) model that is being completed as part of an AES project terminating in June. This information will also be used to provide insights into the fields scale process for a sub-regional project that is ongoing, as well as to evaluate the impact of changes in agricultural practices on crop yields and water quality. Workshops will be conducted to disseminate the results. |
| Approach | This research project will continue field investigations, simulations of existing conditions using GIS and a numerical model in order to evaluate the impacts of any proposed management changes. The data needed to compute crop yield losses are soil salinity, depth to water table, soil type, irrigation schedule, and crop ET. All these data are measured as point data and therefore must be converted to spatial data by appropriate interpolation or weighting techniques. Once all data have been captured or converted into a GIS grid database, this database is linked with a spatial model to compute the potential yield loss for each cell in the database. At each of the fields between 8 and 12 observation wells have been installed where boundary conditions were expected (next to canals or drains) or in areas of the field where high soil salinity was determined. The observation wells were dug using a giddings rig and are between 3.5 and 6 meters in depth with a perforated well screen at the bottom. At each of the selected fields a Davis Rain Collector was installed along with a Hobo Event Rainfall Logger along with an ET Gauge sensor (ET Gauge Company). The depth to water table has been measured using a Solinst model 101 water lever indicator. The groundwater conductivity and temperature has been measured using a YSI Model 30 SCT meter. Each of the wells is pumped once a month using a battery operated submersible pump to make sure the water in the wells is representative of the groundwater in the area. Groundwater quality samples are taken with well bailers weekly during field visits and a subset of these samples will be analyzed for selenium. Currently ten fields are monitored in the area around Holly and La Junta Colorado. These fields were selected to represent different conditions to encompass the variability of site conditions typical in the Arkansas Valley. There is also a need to maintain current observations into the future and to monitor the implementation of Best Management Practices (BMPs) recommended by previous work. Parameters impacting crop yields change spatially as well as temporally. Therefore, a series of animations were created for each field to see how parameters change spatially and temporally. These animations have proven to be very valuable, an example can be viewed at the following URL: http://www.ids.colostate.edu/projects/arkansas/. Selenium will be sampled in the irrigation water applied to the fields in the study and samples will be taken from the Arkansas River to determine contributions and background conditions for selenium in the area. The samples from the groundwater wells will be used in conjunction with other data to evaluate the impact of agricultural practices on selenium at the field scale. Dr. Jim Loftis (Civil Engineering) and Dr. Dean Heil (Soil and Crop Sciences) will conduct the work related to selenium. |
| Keywords | Salinity, Irrigation, Drainage, Computer Modeling, Monitoring, GIS |
| Progress Reports | |
| 1998 | This goal of this project is to obtain a good estimate of the salt mass balance in the S. Platte River Basin and the Arkansas River Basin as a first step toward understanding the sources and causes of the increase in salinity. Effective strategies for future management of salt problems in either basin must be based on such an understanding. For the S. Platte Basin, a basin-level salt balance developed earlier is being refined and improved. The present study makes use of existing conductivity and flow data, supplemented by a few samples collected in the field, to identify the major sources of salt in the basin--agricultural, urban, natural. While many assume that most salt is contributed by agricultural return flows, the present study suggests that much of the salt load originates from natural sources along the Front Range. This finding could have significant implications for management. As another part of this effort, two computer tools are being developed to allow users to calculate the amount of Consumptive Use (CU) from a particular area. This task is being conducted in cooperation with several water organizations in the S. Platte River Basin. The first tool is called the S. Platte Mapping and Analysis Program (SPMAP) and allows users to select the area that they are modeling using a map. SPMAP then generates all the input data for the second tool, a CU model called SPMAP CU. Both of these tools are being tested and upgraded. The third general task of this project is the use of a simulation model (Colorado State University Irrigation and Drainage Model (CSUID)) for evaluating best management practice (BMP) alternatives for irrigation and drainage systems in arid and semi-arid irrigated areas. The goal of the model and suggested BMPs is to improve the sustainability of agriculture. Two fields in the Arkansas Valley have been identified as test plots. Two water level recorders were installed in one of the fields, and soil samples were taken in one-foot increments to a depth of four feet from both fields. Field soil samples were taken at 27 sites on one field and at 9 sites on the other field. These samples will be analyzed, and the results will be used to determine whether more intensive sampling is required. The information obtained from the soil samples will be used to generate input data sets for CSUID. |
| 1999 | The goal of this project is twofold: to obtain a good estimate of the salt mass balance in the South Platte River Basin as a first step toward understanding the sources and causes of increased salinity, and in the Arkansas River Basin to monitor the salinity of several fields to be able to quantify the salinity problem and evaluate its temporal and spatial variability. In the Arkansas River Basin four fields were selected for detailed monitoring last year. In each of the fields 8 to 11 monitoring wells were installed. At each of these wells, water table salinity and depth to water table were recorded weekly; hourly water table depths were recorded on 11 of the wells using automatic level recorders. Soil salinity readings (between 60 and 120 readings per field) were also collected in each of the fields using a hand-held EM38 and a differential GPS receiver. Maps of soil salinity, water table salinity, and depth to water table have been generated. Additional data about soil properties will be collected this year in order to develop a data set for an irrigation and drainage model for each of the fields. The goal is to evaluate the impact that Best Management Practices would have in reducing salinity at the field scale and the impacts on drainage effluent from the field. Farmers in the lower South Platte River are experiencing declining productivity of their fields due to increasing salinity. To develop effective management options, the origin of the salt must be determined. If it is contributed by subsurface return flows from the irrigated fields themselves, reduced leaching will help solve the problem. If the salt originates upstream of the irrigated areas, improved irrigation efficiencies will do little to reduce the salt load in the river. In order to provide this information, one portion of the project is an effort to identify the primary sources of salt in the lower South Platte River. Existing water quality data for salinity are being used, combined with a conceptual mass-balance model developed in an earlier project. The chemical composition of the salt is being used as an indicator of the source. Preliminary results from the earlier study suggested that most of the salt in the river originated above the major irrigated areas, i.e., above Greeley, CO. The present study confirms those results and is attempting to further identify the geologic factors that contribute to the salt load and to identify the contribution of urban areas. |
| 2000 | The goal of this project is two fold: (1) to obtain a good estimate of the salt mass balance in a portion of the South Platte River Basin as a first step toward understanding the sources and causes of increased salinity, and (2) to monitor the salinity of several fields to be able to quantify the salinity problem and evaluate its temporal and spatial variability in the Arkansas River Basin. Soil salinity levels are limiting the crop yields in both the Lower South Platte and the Arkansas River Basin. The following conclusions have been reached thus far for the section of the Arkansas River Basin between Manzanola and La Junta. First, there is significant spatial variability in each of the fields studied for soil and groundwater salinity, as well as depth to groundwater. All of the fields have experienced moderate to severe crop yield reduction (in some areas up to 75% crop yield reduction was observed). Second, a set of computer tools has been developed to generate maps for the data collected (soil salinity, depth to water table, groundwater salinity). Data was collected from hourly to monthly depending on the data type and was interpolated into surface maps to show values for different time periods (hours to days). These maps were then concatenated into animations for display of the spatial as well as the temporal changes in parameters throughout the growing season. Third, a spatially based model has be developed and calibrated using one of the fields. This model uses the GIS data to study the impact of soil salinity, water table depth, and groundwater salinity on crop yield. Fourth, in most of the fields there is a need to lower the water table, this could be done in some cases by reducing seepage from canals and/or by installing drainage systems. The following conclusions have been reached thus far in the South Platte (middle section, between Denver and Julesberg, Colorado). First, the three major tributaries (St. Vrain Creek, the Big Thompson and the Poudre River) are major contributors to both the salinity levels and the salt loads of the main stem. There is a very large jump in salinity levels for each of the three tributaries as they leave the mountains and enter the plains. This is almost certainly due to the presence of shallow saline formations in the region, since salt deposits are readily visible in low-lying areas. Second municipal wastewater treatment plants appear to be a potentially significant source of dissolved solids, roughly estimated as 21% of the salt load in the South Platte River at Kersey (near Greeley). Third, the estimated mean annual salt load at Kersey is actually greater than the mean annual salt load at Julesberg (near the Nebraska border), even though the salt concentrations in the river continues to increase all the way to Julesberg. This is possible because the mean annual flow at Kersey is greater than the mean annual flow at Julesberg. Thus it would appear that the major agricultural areas downstream of Greeley are not adding salt to the river on a net basis. Rather it appears that salt from the river is being deposited in the agricultural areas or in the underlying materials and ground water. |
| 2001 | The primary concern in the Arkansas Valley with respect to the waterlogging and salinity problems is the economic losses the region experiences. Therefore, quantifying crop yield reductions due to salinity and waterlogging is essential to understanding the extent of the problem. A spatial model for quantifying the waterlogging and salinity effects within individual fields has been developed as part of this project and validated based on field data collected this year. This method utilizes soil and water data commonly collected in field-scale studies. The result is a GIS-integrated model that does not require extraordinary data collection but will provide practical insight into the spatial effects of salinity and waterlogging on crop yields. |
| 2002 | Last year the primary concern in the Arkansas Valley was the drought and the availability of water. Due to the lack of water there was significantly less irrigation, and as a result, we saw a decrease in the water table and reduction in the waterlogging problems. About 25% of the fields that we have monitored in the past were left fallow due to the drought. Also an increased emphasis has been placed on improved irrigation techniques. This year one of the fields that we monitored was a sub-surface drip irrigation system where the farmer was growing peppers. We were interested in evaluating the impact of sub-surface drip on the production of peppers through the use of chemical injection and irrigation scheduling. Materials and Methods: The drip irrigation system (30 acres) is located near Holly, CO. It is fully automated and consists of five zones and has the capability of maintaining a present pH in the irrigation water. The system accomplishes this by sensors built into the controller. The system was setup to maintain the pH of the irrigation water at values of 6.3, 6.5, 6.7 and 6.9 for zones 2, 3, 4, and 5 respectively. Maps of salinity for the farm were generated three times during the season. Water samples from the well were collected several times during the year and sent to a lab for analysis. Results: Data was collected during the year on the pH of the water and the amount of different chemicals used in each of the zones. Irrigation: The four zones were irrigated a total of 113 days for an average of 2.68 acre feet of water per acre. Production: Control zone 1 (6.5 pH with TMR 23 chilie), produced 18,020 plans per acre with an average of 4.2 lbs of pepers per plant. For each of the Zone the plants per Acre (ppa) and the lbs per Plant (lbspp) are the following: Zone 2, ppa 22340 lbspp 4.8, Zone 3, ppa 17860, lbspp 4.3, Zone 4, ppa 16430, lbspp 3.8 and Zone 5, ppa 13670, lbspp 3.7. Groundwater: A couple of groundwater samples were taken and sent to Ward Laboratories in Kearney Nebraska. Sample were taken on June 29 and August 19, 2002. It is interesting to note that the pH of the well water is around 7.5-7.6 and therefore had to be significantly reduced to meet the targets of this experiment. Also the EC of the irrigation water is very high with values of around 2.5 mmho/cm. Soil Salinity: Soil salinity data was collected three times during the year using an EM-38. Along with the EM readings GPS points were collected allowing us to generate maps of the soil salinity. The salinity data is fairly low at the beginning of the season (May 15) but it increased significantly by mid season (July 22) and was starting to decrease by the end of the season (September 7). We will collect soil salinity data at the beginning of next season to determine the amount of salts leached during the winter and early spring. |
| 2003 | Salinity levels in the Arkansas River increase from 300 mg/L to 4000 mg/L over the 150-mile stretch from Pueblo to the Kansas Border. Salinity in the lower Arkansas River Valley in Colorado is reducing crop yields and threatening the future of agriculture in the area. Agricultural lands in the area are experiencing substantial crop yield reduction due to waterlogging, capillary movement of salts to the soil surface exacerbated by waterlogging, and from the application of saline irrigation water. Researchers are collecting field data (water table levels, soil and water salinity, and crop yields) in the lower Arkansas River Valley for use in calibrating the Colorado State University Irrigation and Drainage Model (CSUID). The CSUID Model estimates the effects of salinity in the area and can be used as a tool to judge the efficacy of the use of various agricultural practices intended to ameliorate the salinity situation. The CSUID Model simulates principal field processes in three-dimensional detail at daily and smaller time steps. Recent improvements in the model include updates to the salt solver to make it fully three-dimensional and to give the user more control for soil and water salinity calibration; the revision of the code to effectively handle three-dimensional input data; the reduction of the relative crop yield and plant transpiration due to matric, osmotic, and waterlogging stresses; the incorporation of bare-soil evaporation during the off-season; and the introduction of dynamic timesteps to make the solution process more robust. Researchers are using the CSUID Model results along with field data to test various combinations of agricultural practices including reducing infiltration and deepening artificial drainage. After calibrating the results of the model with field data, researchers are seeing that any gains in relative crop yield due to the lowering of the water table appear to be slight in the first year, but in simulated future years, as infiltration is reduced, it appears that various agricultural management changes are likely to improve crop yield. |
| 2004 | Problem: Salinity in Colorado's lower Arkansas River Valley is reducing crop yields and threatening the future of agriculture in the area. To assess the magnitude of the problem, soil salinity maps of fields in the area are being created by combining field data and remote sensing technology. For the past several years, soil salinity readings have been taken three times a year at over 60 locations in each field using an EM-38 probe. In addition, over 100 soil samples have been collected and analyzed in the lab against the EM-38 readings. Soil salinity maps were then generated using GPS coordinates for each monitoring point. However, data collection using the EM-38 is time consuming, and it is expensive to collect such data for large areas. Consequently, we have been working on using 4m Ikonos multi-spectral satellite imagery to determine the severity of soil salinity and its impact on crop yield. We are: 1) training the multi-spectral data to distinguish between different levels of salinity based on the reflectance of the crop and 2) evaluated ordinary least squares (OLS), spatial autoregressive (SAR), and spatial lag (SLAG) models for determining if soil salinity can be accurately predicted from remote sensing data. So far, we have focused our research on corn fields. Elevated levels of soil salinity affect the appearance and growth of corn crops, which can be detected using satellite imagery. By enhancing the image, we can separate the crop condition into several classes. Using spatially referenced ground data collected at the study area, we can relate each class in the satellite image to a level of soil salinity. We can use these classes to create a signature file and classify other areas planted with corn.We have tested the accuracy of our classifications using remote sensing on several fields with very encouraging results. For the fields that we have tested, the difference between soil salinity measurements taken with EM-38 and the satellite image is less than 15%, and part of this discrepancy could be a result of the errors that occur in EM-38 salinity maps. As for assessing the model that is best for use in conjunction with remote sensing data to accurately predict soil salinity, the OLS model was determined to best fit the criteria. The combination of spectral bands used consisted of the blue band, the infrared band, the normalized difference vegetation index (NDVI), and the infrared band divided by the red band (IR/R). The p-values of the selected model for the intercepts of blue, infrared, NDVI, and IR/R were 0, 0.0372, 0.0001, 0.0174, and 0.0113 respectively. Each of these p-values is less than 0.05 which indicates a strong correlation with soil salinity data. The standard errors of these variables were all relatively small at 1.7585, 0.0047, 0.0022, 2.1378, and 0.3496 respectively. The p-value of Moran's I (residuals) was 0.3814, a number greater than 0.05, meaning that the residuals are spatially independent. The p-value of Lagrange was 0 indicating that there is a strong correlation between the selected variables and soil salinity. |
| 2005 | Problem: Salinity in Colorado's lower Arkansas River Valley is reducing crop yields and threatening the future of agriculture in the area. To assess the magnitude of the problem, soil salinity maps of the area are being created by combining field data and remote sensing technology. Extensive field data has been collected, but now an effort is being made to cut down on the expensive and time-consuming field work by producing maps using remote sensing techniques. Different remote sensing techniques are being used to create salinity maps which are being compared for accuracy against the field data. This year, we made a major push towards exploring the relationship between soil salinity and evapotranspiration (ET). Narrative: Crop ET can be estimated using satellite imagery by applying an energy balance approach. As crop ET is often reduced when the osmotic potential in soil pore water increases due to salinity, remote sensing of crop ET should offer insights into salinity levels of agricultural lands. We have developed a methodology called RESET (Remote Sensing of ET) for calculating ET based on remote sensing. RESET is based on the Surface Energy Balance Algorithm for Land (SEBAL) and is enhanced to account for spatial and temporal variability. This approach uses the thermal information from the infrared band of the satellite image as well as the crop reflectance or Normalized Difference Vegetation Index (NDVI). RESET improves on previously developed models because it uses multiple weather stations for obtaining ground data needed for the model and it allows for the calculation of cumulative ET over time between available satellite images. Previously developed models mainly use one weather station for ground data (wind speed, ET, Rain) making their results only valid for areas that where the weather is fairly uniform over an area. Our model is able to use data from several weather stations by developing a surface interpolated from all stations that fall within the targeted area. RESET also accounts for changes occurring over time. To test the relationship between soil salinity and evapotranspiration, GPS referenced ground measurements of soil salinity were taken in several corn fields. ET was calculated using RESET, and a regression was generated showing the relation between ET and soil salinity. Although there are many factors that influence crop ET at a given location in a field, our results suggest that there is a clear correspondence between ET and soil water salinity for corn. The root mean square error (R) was over 0.86. Rates of ET start decreasing in corn at soil salinity ECe values that exceed 2 to 3 dS/m. These results suggest that this approach could be used to detect and evaluate soil salinity with acceptable accuracy for large areas at less cost than conventional approaches. |
| Impact | |
| 1999 | The demonstrations in the Arkansas River Basin show farmers the extent of the salinity problem at the field scale, accomplished by developing maps and video presentations showing the spatial and temporal variability of salinity in these fields. In the South Platte River Basin our efforts are helping people understand the actual sources of salts in the basin: agricultural, municipal and/or natural. |
| 2000 | Evaluation of the salinity problems at field scale requires the ability to accurately determine locations were crop yields are diminished due to high soil salinity and/or high water tables at specific times during the growing season. This data is highly variable spatially and temporally and therefore the tools and data collected during this project will provide valuable information and tools in assisting farmers to plan long-term strategies for reducing impacts to crop production. In the South Platte this effort is helping identify the sources of salts which is an essential step before solutions can be formulated. |
| 2001 | The yield data collected and the output of the model that was developed have shown that yield has a significant spatial variability. The average yields collected in four fields in which corn was grown last year range from 90 bushels/acre to 260 bushels/acre. In the field with an average yield of 260 bushels/acre the model predicted that 70% of the area of the field is not affected by salinity or waterlogging, that 20% of the area of the field had a yield reduction of more than 10%, and the remaining 10% of the area of the field has a yield reduction of 20-30%. In contrast, in the field that had an average yield of 90 bushels/acre the model predicted that 30% of the area of the field had a yield reduction in excess of 70%, another 40% of the area of field had yield reductions between 40-70%, and only 5% of the area of the field had no yield reduction due to salinity and waterlogging. When the area of three fields are aggregated and the results of the model are area weighted the results are the following: 1) 40% of the area has no yield reduction due to salinity or waterlogging; 2) 30% of the area has yield reductions of 10-30%; 3) 30% of the area has yield reductions in excess of 30%. |
| 2002 | The results showed that the lowest pH zone which had a target pH of 6.3 and an actual average pH of 6.32 had the highest per plan production of 4.8 lbs. There was significant reduction (20-25%) in yield at the higher pH level (6.9) with an average per plant production of 3.7 lbs. We have monitored this field for two years and there was no significant accumulation of salts this year as compared to last year. The results of this research are encouraging since the amount of water used was 2.68 acre-ft per acre which is easily a 50% reduction in the amount of water that is normally applied in this area. Also, the fact that we have not detected a salinity buildup in the soil is very positive sign. From this results it appears that maintaining the pH of the soil at the lower values (6.3-6.5) is a definite advantage. |
| 2003 | This project brings to light areas in data collection and modeling of the salinity problem that need to be addressed. Useful outcomes in the form of improved management strategies for individual fields in the lower Arkansas Valley or for the whole region are being developed and tested. |
| 2004 | Impact: Mapping soil salinity identifies the magnitude of the soil salinity problem and is the first step in formulating solutions. Using remote sensing to determine soil salinity allows us to evaluate soil salinity over large areas (river basins) in an economical way. Changes taking place over years can be identified by using remote sensing data from multiple years. This allows us to determine the rate of change and identify problem areas before salinity gets to the point of significantly impacting yields. |
| 2005 | Mapping soil salinity identifies the magnitude of the soil salinity problem and is the first step in formulating solutions. Using remote sensing to determine soil salinity allows us to evaluate soil salinity over large areas, such as Colorado's lower Arkansas River Valley, in an economical way. Since we are using Landsat images we can use historical data to detect the changes taking place over the last few years. This allows us to determine the rate of change and identify problem areas before salinity gets to the point of significantly impacting yields. |
| Publications | |
| 2000 |
Garcia, L.A., Elhaddad, A., and Patterson, D. (2000), "Modeling Irrigation and Drainage System for Drainage Water Quality Management", presented at the International Challenges Facing Irrigation and Drainage in the New Millennium, June 20-24, 2000, Ft. Collins, CO. Haby, P. and J. Loftis. 2000. Salinity Characterization and Source Assessment in the South Platte River Basin, Northeastern Colorado. ASCE Watershed 2000 Conference, Fort Collins, Colorado, June. |
| 2002 |
Foged, Nathan. 2002. A GIS Based Tool to Estimate Relative Reductions in Crop Yield Due to Salinity and Waterlogging in a Corn Field in the Lower Arkansas River Basin in Colorado. M.S. Thesis. Colorado State University. Fort Collins, CO. |
| 2004 |
Eldeiry, A. and Garcia, L.A. (2004). "Spatial Modeling using Remote Sensing, GIS, and Field Data to Assess Crop Yield and Soil Salinity." Published in the proceeding of the 24th Annual Hydrology Days, Fort Collins, CO, March 10-12. Elgaali, E. and Garcia, L.A. (2004). "Neural Network Modeling of Climate Change Impacts on Irrigation Water Supplies in Arkansas River Basin." Published in the proceeding of the 24th Annual Hydrology Days, Fort Collins, CO, March 10-12. |