From the ground up: Soil best management practices for vegetable production on rooftop farms

Project Overview

Project Type: Partnership
Funds awarded in 2016: $14,999.00
Projected End Date: 12/31/2017
Grant Recipient: Cornell University
Region: Northeast
State: New York
Project Leader:
Dr. Thomas Whitlow
Cornell University

Annual Reports


  • Vegetables: beans, beets, broccoli, cabbages, carrots, cauliflower, celery, cucurbits, eggplant, greens (leafy), leeks, onions, peas (culinary), peppers, radishes (culinary), tomatoes, brussel sprouts
  • Additional Plants: herbs


  • Education and Training: decision support system, extension, on-farm/ranch research, participatory research
  • Soil Management: nutrient mineralization, soil chemistry, organic matter, soil physics, soil quality/health
  • Sustainable Communities: urban agriculture

    Proposal abstract:

    We propose to develop and test new soil-less media suited to outdoor farming on urban rooftops. Cities present a growing demand for fresh, local vegetables, which has stimulated the growth of urban farms across America. Available space would appear to be a limiting factor, yet considering that New York City alone has over 8600 ha of flat roof space (25X the area of Central Park) that could support production agriculture, it is easy to imagine that commercial urban farming is both practical and a growth sector in urban economies. A rooftop farm differs from both in-ground farming and green house production, so production models used in these traditional systems require approaches that have yet to be formalized and we are only now starting to understand. Among the many practical challenges facing a rooftop farm is the need for almost constant irrigation and fertilizer, with the parallel need to avoid water pollution from runoff drainage. At the center of both these issues is the need for a sustainable, soil-less growing medium for vegetables that retains water and nutrients instead of discharging them into storm sewers. Since May, 2014, we have been monitoring water use and nutrient fluxes at the Brooklyn Grange, a 0.6 ha farm on the roof of the Brooklyn Navy Yard and found that in one year, over 8 million liters of water/ha and 2500 kg of nitrogen/ha are leaking from the farm, so the opportunity to increase resource use efficiency is quite substantial.

    Project objectives from proposal:

    Several coinciding problems motivate this project. First is environmental pollution. “Dead zones” in the Chesapeake Bay and the Gulf of Mexico resulting from agricultural runoff are well known. Could they have been avoided if we had developed Best Management Practices (BMPs) to prevent the problem in the first place? Regarding roof top farms, we are on the rising limb of the curve and have the extraordinary opportunity to develop BMPs that avoid excessive water use and nutrient runoff before they become problems. Drainage from the Grange discharges into the East River/Open Waters catchment, designated as Impaired by NYDEC due to combined sewer overflows. Second, rooftop vegetable production is sufficiently different from conventional in-ground farming, greenhouse production, and extensive green roofs that conventional practices adopted from these other systems are not a good fit. At the few rooftop farms in existence, fertilization and water requirements have been borrowed from in-ground vegetable farming in native mineral soil. Natural precipitation is insufficient to support vegetable production on a roof, so irrigation with potable water is needed even in the humid NE. From green roofs, rooftop farms have borrowed the lightweight, fast draining rooting medium to avoid overloading the roof. Green roof media resemble artificial greenhouse media, with the important difference that they are not replaced at the end of each cropping cycle. Importantly, on a conventional green roof, rapid growth and vegetable quality are not considerations. Third, the absence of mineral soil in the roof mix causes gradual leaching of potassium and corresponding deficiency symptoms in leafy vegetables that affect quality. Further, high concentrations of ammonium in detected in plant sap suggest that drought in the growing medium inhibits nitrification despite daily irrigation, suggesting that rapid drainage depletes water in the zone where roots occur. Clearly, we need a novel cultural system specifically suited to rooftop farming. In this light, the challenge is to develop an optimal growing medium that retains more water and nitrogen than the current mix without exceeding the bearing capacity of an industrial roof. This will both avoid excess runoff and nutrient leaching.

    PROPOSED SOLUTION: Soil is the central component for delivering water and nutrients required for plant growth. We believe that a growing medium designed specifically for rooftop farming is the most strategic point of intervention in order to reduce both leaching and drainage to storm sewers and ultimately into surface water bodies. We will develop an improved soil-less medium that optimizes water retention and reduces drainage and leaching of nitrogen, potassium and other nutrients.

    PAST WORK: There are no studies of a rooftop farm’s N budget, but studies of similar media used in green infrastructure for stormwater runoff abatement provide a useful starting point. Existing literature reports N retention and application rates between 32% (121.1 N Kg ha-1 y-1) and 54% (336.7 N Kg ha-1 y-1). Both studies suggest that N in compost based media exceeded the optimum N levels for ornamentals used in bioswales and rain gardens. We have partnered with the Brooklyn Grange since 2013, and they are now in the process of installing a third farm. Together, we have a manuscript in press describing the potential risks to human health resulting from deposition of atmospheric particulates from vegetables produced on a roof. The Brooklyn Grange is a pilot project of the Community-Based Green Infrastructure Program funded by the New York City Department of Environmental Protection. Construction of the farm cost $592,730 in 2012.

    PROJECT METHODS: Hypotheses H1: Optimizing media characteristics and irrigation rate will achieve N retention rates above 54%, an ideal for leafy vegetables grown in soil. H2: Focused delivery of water and nutrients around the root zone will enhance plant’s performance and reduce N leaching. Shallower media with increased water and nutrient holding capacity will reduce the need for supplements. Overview: In greenhouse trials, we will grow arugula as a sensitive indicator crop in growth media composed of composted food waste mixed with varying proportions of expanded shale, biochar and coconut coir and installed at 3 depths. The 2 best performing combinations will be tested in plots at the Grange farm at the Brooklyn Navy Yard to validate findings and allow us to develop guidelines for irrigation and fertilization schedules. Greenhouse Experiments Experiments to identify optimal mixtures and media depths will be conducted in the greenhouse using specially made PVC containers. The 2 best combinations will be used in subsequent on-farm trials. We will add 3 amendments (expanded shale, hardwood biochar, coconut coir) to food-waste compost at 3 different volumetric ratios (25:75, 50:50, 75:25).

    Each mix will be tested at 4 different media depths (10, 20, 30, 40 cm). 56 treatments each replicated 5 times (280 units total), will be tested at Cornell. The response variables include 1) water holding capacity, 2) nutrient leaching, 3) mineralizable N, and 4) plant yield, nutrient content and quality. We will use arugula, which is especially sensitive to nutrient imbalances, as a bioindicator of treatment performance.

    Field Experiment: Farm-scale Biogeochemistry Water Budget. The water and nutrient budget of the Brooklyn Grange has been monitored since May 2014 using a micro meteorology station, online irrigation controllers, wireless soil moisture sensors, passive atmospheric deposition collectors, and mesh bags containing ion exchange resins buried at strategic locations in planting beds on the rooftop. During the 2014 growing season, the average water applied for irrigation was equivalent to 8 × 106 Liter · y-1 ha-1. A V-notch weir equipped with a sonic depth sensor monitors drainage from one of 8 zones across the farm. In conjunction with ET demand calculated from the micro met data (Penman-Monteith Method) we can calculate the water balance for the farm. Results to date indicate that irrigation and drainage rate are virtually identical on sunny days; average volumetric soil moisture content is only 20%; and rainfall is ineffective in reducing the irrigation demand. This ongoing monitoring of the water budget will continue for the duration of this study.

    N Budget: The atmospheric input and the leaching loss of N are estimated by analyzing ion-exchange resin contained in atmospheric bulk collectors, and resin-filled mesh bags buried under soil beds and around roof drains. Resin has been collected every 6 weeks since May 2014 and will continue through 2016. During the 2014 growing season, findings show that at least 500 N Kg ha-1 y-1 of mineral N was leached from the growing media to the municipal storm drain. Combustion analysis of soil sampled before and after the growing season in 2014 indicated that over 2000 N Kg ha-1 y-1 of organic N was lost from the farm. Based on the farm records, N application rate in 2014 growing season was 117.2 N Kg ha-1 y-1, and the N export by harvesting the vegetable was 85.2 N Kg ha-1 y-1. Thus, the N flux resulting from fertilization and harvesting was much smaller than the total N loss from the farm< (2500 N Kg ha-1 y-1) For comparison, conventional in-ground farms lose nitrate-N at rates between 4-155 N Kg ha-1 y-1, suggesting substantial room for improvement on the rooftop farm. Observations of farm-scale water and N fluxes will continue for the 2017 growing season following installation of the best performing growth medium identified in 2016. (Note that we expect the project to continue after the period of the this grant.)

    Field Experiments-Improved Growing Media:In the proposed study, two media will be selected from the greenhouse experiment: one that maximizes crop yield and the second optimizes yield against reduced nutrient leaching. 0.6 acre of the Brooklyn Grange is dedicated for the salad greens, where 0.4 acre will be amended based on the each of two selected media while 0.2 acre will be left untreated as a control. The input, retention, loss from harvested vegetables, and leaching of nutrients and water are being monitored in the ongoing biogeochemical monitoring since May 2014. Year 1 (2016) of the proposed study will determine which growing medium achieves best crop performance while reducing nutrient leaching. The optimum design will be applied to the entire farm in 2017 after this grant terminates. The biogeochemical performance of the farm will continue to be monitored in the framework of ongoing experimentation.

    Outreach plan

    The immediate outcome of this study will be a new soil-less medium for rooftop farming, accompanied by guidelines for optimum irrigation, and nutrient application, referenced to produce yield and quality, and rates of water and nutrient loss. This package is the first of this kind and will immediately inform cities with major rooftop farming initiatives, including NYC, Boston, Jersey City, Chicago, and Seattle. Because the Grange is a pilot project funded by the Community-Based Green Infrastructure Program of the New York City Department of Environmental Protection ($592,730 in 2012), our project will aalso directly inform the NYCDEP Best Management Practice Inventory and future projects funded by the program. Additionally, because the Grange has an internship and green job-training programs for college students, underrepresented local residents, and international visitors, this project has a ready-made, face-to-face extension audience. Information will be made available to a diverse stakeholder audience via a web page maintained by Cornell University’s Section of Horticulture, The Cornell Small Farms Program, and the Grange website as well as presentations at both national and international meetings.

    Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.