Introduction

The United Nation’s Food and Agriculture Organization (FAO) has projected the earth’s population will exceed 9 billion by 2050 (Alexandratos and Bruinsma, 2012).  This increase will require an additional billion tons of cereal crops alone; nearly a 33% increase over current levels. The FAO expects most of the gains to come from increased yield and increased land in production. In developed countries production demands will increase but those gains will not come from increased land in production. Even with this 33% increase in production, the FAO projects that water demand will increase by only 11%. The reduced rate of increase is expected to come from improvements in water use efficiency and a reduction in rice production. Most of this increase in efficiency will come from improvements in stress tolerance and reduced water needs (Baulcombe, 2010).  In developed nations, efficiency gains must come from improved management practices. In these countries, the FAO projects an 8% decrease of land in production. An increase in production demand, increase in water use, and a reduction of land all indicate that pressure on farms will continue into the future.

It is not necessary to look beyond the United States to find evidence of pressure on irrigated farms. Irrigated agriculture in the US accounts for 80-90% of the consumptive water use and approximately 40% of the value of value of agricultural production (Schaible and Aillery, 2012). This value, totaling nearly $118 billion US dollars, is produced on 57 million acres. According to the most recent Farm and Ranch Irrigation Survey (USDA, 2012) , of the 296,303 irrigated farms 25,853 reported diminished yield caused by a shortage of ground or surface water.  Also of concern is than 6011 farms reported a discontinuance of irrigation, up more than 30% from the last survey.

The need for a standard

Agriculture has become a data driven endeavor. New sources of information about soil, weather, crop status, machine operation, marketing, and economics all facilitate the evidence-based decision-making that defines precision agriculture. Using these new data streams requires tools and the evidence of this is found in the proliferation of new apps for utilizing agricultural data. A search of the Google Play store for the works “Agriculture” or “irrigation” yields 92 and 82 results respectively. Even though these tools improve accessibility of the data, the grower is responsible for relating the data to other aspects of the farm enterprise. Relating these data can involve combining multiple source into a single output, performing calculations that transform the data in to recommended actions, or feeding models that forecast potential outcomes. Each of these tasks requires moving and transforming data. The process as a whole is integration. Each source of information has value by itself, but only when combined do they provide the evidence needed for evidence based management. In other words, integration produces decision-making power that is greater than the sum of the individual streams.

An example of integration’s value is found in the study by (Wang and Cai, 2009) .  The authors examined the management benefits of incorporating weather forecasts in to irrigation scheduling. In this case, the ‘new’ stream of data was the forecast but providing this new product required integrating the information into an irrigation recommendation. The results of the study were what were expected: including weather forecasts improved management. However, what the growers received was more than another source of information. The researchers were integrating the forecasts into the recommendation and this integration played an important role in improving management.

Although weather data is increasingly available online, not all weather networks are providing data in forms readily useable by web based applications. For example, the Agricultural Water Conservation Clearinghouse has a list of weather stations and ET networks (Colorado State University, n.d.). Hillyer and Robinson (2010) conducted a survey of the Of the 14 networks listed and only one (CIMIS) provided weather data in an XML format. Nearly all of the networks provide their data in a ‘csv’ format that is readily useable by spreadsheet applications. As of this writing, that list has grown to 25 websites or networks and 7 provide data in XML or other machine oriented format; a 4 fold increase.

Growers have long recognized the value of integration, even if not explicitly. One example of this is the Eldorado Irrigation District (EID) in northern California (Taylor, 2009).  The EID has developed an irrigation management service for its members. This service includes installation and maintenance of soil and weather sensor, data entry, and a software system build specifically for the EID (TruePoint Solutions, 2008).  Probably the most valuable aspect of this system is the trust that the EID has developed with its members. The growers receive irrigation recommendations via the TrueISM software and members trust the EID to do the work associated with providing the service. While a majority of the effort goes into maintaining the hardware, a significant part of the labor effort goes into collecting the data and moving it into the software system. Thus, a significant part of the value of this service comes from integration of sensor data, water applications, and weather information.

Many tools exist for managing irrigation (Smith et al., 2010).  A majority of growers does not use physically based tools. Figure 1 shows data from Table 22 of the most recent Farm and Ranch Irrigation Survey (“Methods used in Deciding When to Irrigate”) (USDA, 2012).  The methods that involve evidence based management (AKA Scientific Irrigation Scheduling) are selected by 64,037 times and non-evidence based methods are selected 369,917 times. This imbalance has persisted over the last seven surveys dating back to 1988.

Figure 1. Table 22, Farm And Ranch Irrigation Survey, 2013 Census of Agriculture

One potential explanation for the poor adoption is the effort required to use physically based methods. A soil moisture sensor can provide useful guidance about when to irrigate. In recent years, remote telemetry for soil moisture sensors will place the data directly into the grower’s hands via smart phone apps. When maximizing the value of water, however, a soil moisture sensor only tells part of the story. Growers also need to know how much water was applied, how much the crop has used, what the weather has done, and condition of the crop. Sensors for each of these sources are available but they exist as separate tools. Maximizing the value of water requires integrating all these sources and therein lies the problem: the tools do not talk to each other. Integration is the grower’s responsibility and the effort required can be discouraging.

The Farm Management Information System (FMIS) is an obvious point of integration for all these disparate information sources. By facilitating the integration, the FMIS alleviates some of the grower’s burden. Implementing this integration requires the FMIS to have special code to interoperate with each different data source. As new streams emerge, the FMIS must continue to expand. If each of the different tools could produce data in the same format, integration would be a simpler and thus cheaper proposition. A common data format would lower the cost of implementing integration and likely lead to a proliferation of new and more comprehensive FMIS. This proliferation would in turn lead to increased adoption of more efficient methods for deciding when to irrigate.

In the EID example above, the district’s software system acts as a FMIS. In broader context, the FMIS is a central integrator because it acts as an information nexus for the farm enterprise. FMIS have been around for many years but recent changes in the ISO 11783 standard have made FMIS a central component of machine oriented agricultural operations and the data associated with them. The revised ISO 11783 standard does not include irrigation operations in any significant form.

The presence of new data sources, the multitude of existing irrigation management technology, availability of cheaper telemetry, and the expanding role of FMIS all indicate all portend an important opportunity to improve irrigation management. However, there is no established framework for integrating these disparate data sources or tools. These factors create an immediate need for an irrigation related data exchange standard. Without a standard irrigation will miss the Big Data revolution and will instead remain a “by hand” management activity as evidence by the decision preferences shown in Figure 1.

The PAIL Project

The PAIL project seeks to develop a common language that enables integration of all the disparate sources of water management technology.

“The purpose of the PAIL project is to provide an industry-wide format that will enable the exchange and use of data from irrigation management systems, which are currently stored in a variety of proprietary formats” (“PAIL Project Narrative - PAIL Project - WIKI,” n.d.)

PAIL’s Objectives

The PAIL project will cover a broad range of data but will focus especially on two types: Operations and Observations.

Observations are the field, atmospheric, plant, or other in situ measurements that apply to irrigation management. This includes weather stations, soil moisture sensors, or crop related sensing.

Operations are all of the activities associated with the application of water with an irrigation system. This includes but the management level communications and record keeping. The operations data set is based around a Work Order, which describes what irrigation is desired, and a Work Record, which describes what actually occurred.

There are several deliverables that will come from the PAIL projects. Of those, four are important for this paper.

• The XML Schema (Fallside and Walmsley, 2004) is the primary technical deliverable.  This document contains a structured and unambiguous definition of the data and how it is formatted.
• Use Cases (Jacobson, 1992) that describe, in a semi-structured way, the typical management scenarios where data exchange occurs.  These documents effectively define the scope of the standard.
• BPMN Diagrams (von Rosing et al., 2015) that will provide exposition of the key process that make up the irrigation management process.  Explanation of BPMN is beyond the scope of the paper but there are two aspects relevant to PAIL. First is that BPMNs are based on a business process, that is, the management process currently used, as seen from a farmer’s perspective, who’s goal it to operate as a profitable enterprise. The second element is that the process of building the BPMN results in identification of a set of messages (and data thereof) that define the communications that occur during irrigation management.
• A field trial that will serve to expose any potential conflicts or shortcomings of the standard. The trial will also serve as a demonstration the standard’s utility to potential adopters. The PAIL team has already conducted one trial in 2015 and is conducting a second during the 2016 season.
• A formal standard, submitted to ASABE. A standards project, X632, is already in progress in the ASABE irrigation management committee, NRES-244. Drafting of this standard is underway and submission for balloting is expected in Q3/Q4 of 2016. Ultimately, the ASABE standard will become an ISO standard also.

In this paper, we present the core elements of PAIL, the business process from whence those elements were derived, and an introduction to the data structures defined in the standard. This paper will prepare the reader to begin adoption of PAIL. The intended audience is both engineering research professionals who will review the standard and practitioners who will ultimately adopt the standard. This paper will enable interested persons to decide if the PAIL standard can help their organization serve the irrigation industry and, ultimately, the irrigators themselves.