Thank you for visiting nature.com. The browser version you are using has limited CSS support. For the best experience, we recommend using the latest browser version (or disabling compatibility mode in Internet Explorer). Additionally, to ensure continued support, this site will not include styles or JavaScript.
We have launched a new geospatial dataset, NPKGRIDS, which for the first time provides data on application rates of three key plant nutrients – nitrogen (N), phosphorus (P, expressed as phosphorus pentoxide P2O5), and potassium (K, expressed as potassium oxide K2O) – for 173 crops as of 2020 at a geospatial resolution of 0.05° (approximately 5.6 km at the equator). NPKGRIDS was developed using a data fusion approach that combines crop mask information with eight published datasets of fertilizer application rates derived from georeferenced data or national and local statistics. In addition, total N, P2O5, and K2O application rates are compared with country-level information provided by FAO and the International Fertilizer Association (IFA), and validated against data provided by national statistical offices (NSOs). NPKGRIDS can be used for global modeling, decision making and policy development to maximize crop yields and minimize environmental impacts.
The use of chemical and mineral fertilizers has increased almost tenfold over the past six decades1, making a decisive contribution to the growth of crop and livestock production over the same period in the context of global economic expansion2. At the same time, excessive and inappropriate use of fertilizers from agricultural lands has created serious environmental problems that can affect ecosystems and human health at all levels: from local soil and water pollution to regional eutrophication hotspots and marine dead zones at the confluence of major rivers flowing through important agricultural regions3,4,5,6. The dual objective of ensuring food supplies to meet global demand while eliminating and reducing environmental damage is a major challenge for people and the planet and is fundamental to the 2030 Agenda for Sustainable Development6,7 and the Global Biodiversity Framework8, and is of particular importance for the efficient use of fertilizers to achieve agricultural production and sustainable development.
Currently, two global datasets, FAO1,9,10 and IFA11, provide extensive information on nitrogen (N), phosphorus (P) and potassium (K) inputs to agriculture, with country statistics for the period 1961–2022 updated annually. In addition, limited data on crop-specific fertilizer inputs are available12. This information is a recognized global source of information for analysing fertilizer use in agriculture and its trends at the national, regional and global levels, as evidenced by dozens of published articles13, international reports14, sustainability indices15 and planetary boundaries data16.
At the same time, studies focusing on local or regional issues often require more detailed subnational-scale information to assess interactions between fertilizer use and key covariates (e.g., climate conditions, soil properties and water flows, ecosystem and crop distribution, farm management type, infrastructure, and population data). To address these needs, global spatial maps of fertilizer distribution have begun to appear in the literature17,18,19,20,21,22,23,24, primarily to inform global biogeochemical studies and models in the Earth system sciences. These products are a useful step toward refining information from national to subnational and gridded data, although they have a number of important limitations. First, these new maps are often created by spatially transforming existing national-level information and do not incorporate more detailed published data and subnational information from national statistical offices. Second, the creation of such maps requires significant amounts of data and computational resources to develop and validate, so existing products are largely ad hoc and lack the coordination needed to ensure continuous improvement and updating. In fact, the most widely used geospatial dataset to date is limited to 2003 data on nitrogen, phosphorus, and potassium application rates by crop type25 (hereafter MFM, for Mueller et al.’s Fertilizer Maps). Significant changes in agricultural land use and fertilizer application over the past 20 years9, coupled with dramatic increases in computing power and data storage, suggest that the time is ripe for a major update of existing products.
Here we present the results of a major new data fusion initiative to create NPKGRIDS, an updated dataset of global inorganic fertilizer application rates by crop type for 2020, with a focus on the major plant nutrients nitrogen (N), phosphorus (P2O5), and potassium (K2O). NPKGRIDS includes fertilizer application rates for 173 crops at a global spatial resolution of 0.05° (~5.6 km at the equator). NPKGRIDS was developed using a data fusion approach combining crop mask information recently available in CROPGRIDS26,27 with other relevant published data sources, as shown below. We first searched and collected existing peer-reviewed and crop-specific fertilizer use data, selecting eight datasets that contain information specifying individual crops or aggregated crop groups in either geographic or tabular format. We then selected the most appropriate dataset for each crop and local unit using the same data merging optimization process and quality assessment system as for CROPGRIDS. We then conducted a comparative analysis of NPKGRIDS using aggregated national application statistics published by FAO, international agricultural associations, and national statistical offices.
We examined and collected georeferenced and tabular datasets reflecting N, P, and K fertilizer application rates at the national and/or subnational levels. To ensure data reliability, we collected data only from peer-reviewed and national sources. We then analyzed these datasets following the workflow presented in Figure 1, which consists of three main steps: step 1) standardization of input datasets into a tabular format at the subnational level; step 2) identification of endogenous data quality indicators; and step 3) global georeferencing of fertilizer application rates.
NPKGRIDS development workflow. Step 1: Transform input datasets into tabular format; Step 2: Define endogenous data quality metrics; Step 3: Global spatial distribution of fertilizer application; Step 4: Validation. MFM: Mueller et al. Fertilizer Map. MRF: Monfreda et al. Dataset. GAUL: Global Administrative Units Layers Dataset.
The starting point for NPKGRIDS was CROPGRIDS26,27, a recently developed georeferenced crop map dataset containing detailed data on crop locations and harvested areas. We then reviewed existing peer-reviewed literature and official national statistics to obtain georeferenced and tabulated fertilizer use datasets that specify individual crops and/or aggregated crop groups. We included only datasets with data years after 2003 (i.e. the latest time coverage of MFM25) and crop names that followed the FAO Indicative Crop Classification (ICC)28. We excluded datasets that were not crop-specific or included aggregated crops without further specifying the constituent crops. The collected datasets contained data on the quality and/or application rates of total nitrogen (N), total phosphorus (P or P2O5), and total potassium (K or K2O) from individual fertilizers and/or combined fertilizers. From these, we selected eight N datasets and seven P2O5 and K2O datasets (Table 1 and Supplementary Table 1). Among the selected datasets, fertilizer use in intermittent crops (FUBC)29 covers crop-specific and aggregated crop group data for 63 countries from 2016 to 2018. We divided it into two datasets: one for individual (IDV) crops only (FUBC18-IDV) and the other for aggregated (AGG) crop groups only (FUBC18-AGG). The Historical Fertilizer Use in Crop Dataset (HFUBC)12 combines fertilizer use data for all crops and crop groups from 1978 to 2018 for 111 countries from the International Federation of Agriculture (IFA) and the Food and Agriculture Organization of the United Nations (FAO). This study used only 12 individual crop data from the HFUBC for 65 countries from 2006 to 2018. It is important to note that FUBC18 and HFUBC are nationally resolved datasets. In addition, datasets from four national statistical offices (NSOs) were included: the United States30 (US), Belarus31 (BY), the United Kingdom32 (UK), and Australia33 (AU), which provided crop-specific fertilizer data at provincial resolution. These eight datasets were used as input to construct NPKGRIDS (Table 1).
We also used additional auxiliary datasets to compute and spatially represent the NPKGRIDS data. Specifically, we used two datasets of global georeferenced crop maps to identify crop locations and harvested areas: the CROPGRIDS26,27 dataset containing maps of 173 crops at 0.05° latitude in 2020, and the Monfreda et al.34 dataset (hereafter referred to as MRF after the authors’ initials) containing maps of 175 crops at 0.0833° latitude around 2000. Both datasets use the Food and Agriculture Organization of the United Nations (FAO) crop nomenclature. When the selected dataset did not contain data on fertilizer application but only total fertilizer application, we estimated fertilizer application per unit area harvested using publicly available harvested area data at the national level from the National Statistical Office (NSO) (e.g. CROP-AU35 and CROP-BY36) or the FAOSTAT database37. To define country boundaries and subnational units, we used the FAO Global Administrative Units Level (GAUL) dataset38 (Table 2).
Eight input datasets (Table 1) are combined into a single tabular format where nitrogen (N), phosphorus (P₂O₂) and potassium (K₂O) application rates are expressed as application rates per unit area of harvested crop. The table resolution is the maximum level for each dataset, e.g. provincial (level 1) for MFM, USA, Belarus, Australia, UK and national (level 0) for FUBC18-IDV, FUBC18-AGG and HFUBC.
For the georeferenced MFM dataset, we first constructed a table of fertilizer application rates at the original resolution of the dataset (i.e. 0.0833°, approximately 10 km at the equator) using a GAUL mask of level 1. For local units with missing fertilizer application rate data, we imputed the missing information using the national weighted average fertilizer application rate FMFM [kg ha−1] (fertilizer n, crop i and country j), which was calculated as follows:
where fMFM(n,i,j,r) in [kg ha−1] is the fertilizer application rate n available for crop i in provincial unit r of country j in the MFM dataset, and AMRF is the corresponding harvested area obtained from the MRF dataset.
For all other tabular datasets, the matching process involved transforming the variables into application rates expressed as the mass of N, P₂O₂, and K₂O applied per unit crop area harvested. If a dataset contained information on application mass only, we calculated application rates using the crop-specific harvested areas for the relevant year from the corresponding output dataset (Table 2). Specifically, the HFUBC, FUBC18-IDV, and FUBC18-AGG datasets used crop area harvested data from FAOSTAT, while the AU and BY datasets used CROP-AU and CROP-BY data, respectively. For the U.S. datasets that contained information on application rates per unit area and percentage of harvested area fertilized, we calculated application rates for the entire harvested area of each crop by multiplication. For the BY dataset, P and K masses were multiplied by 2.29 and 1.20, respectively, to convert them to P₂O₂ and K₂O. In the UK dataset, fertiliser application rates for some crops varied seasonally. Since within-year information was missing from our products, in these cases we used the average seasonal fertiliser application rates to determine the average annual fertiliser rates for crops. For the Australian dataset and some country datasets in FUBC18-IDV and FUBC18-AGG where the data span two calendar years, we used the first calendar year as the reference year. For datasets containing a list of crops within a group (e.g. FUBC18-AGG and the UK), we assigned fertiliser application rates for all crops within a group to that group based on the summary information in Supplementary Table S3.
Other georeferenced and tabulated output and control datasets (Table 2) were also converted to the same data format and administrative unit level as the input datasets.
We developed a multi-criteria ranking scheme to identify the most appropriate fertilizer application rates for specific crops in subnational units representing multiple sources in the eight selected input data sets. The ranking was based on three endogenous data quality indicators: Qc (crop specification); Qr (data resolution); and Qy (synchronicity). Each indicator was assigned a value from 0 (lowest quality) to 1 (highest quality). For each data set, indicator values may vary across crops and subnational units.
The Qc indicator indicates whether the fertilizer data relate to a single crop or a group of crops:
Datasets that include both crop-specific data and aggregate crop data (e.g. UK) will have different Qc values between crops, with individual crops having higher ranks.
The Qr metric evaluates the administrative resolution of datasets, with higher resolution datasets ranking higher, as shown below:
The Qy indicator is used to assess the degree of synchronization between the base year Yr of the input dataset and the base year of NPKGRIDS, which is set to the period from 2015 to 2020, hereinafter referred to as “around 2020″, and is defined as
Qy increases as \({Y}_{r}\) approaches the 2015–2020 period and may differ for different base years \({Y}_{r}\) within the same data set (e.g. HFUBC and US).
The above endogenous data quality measures for all datasets used to compile NPKGRIDS are summarized in Table 3. From a practical perspective, we compare the average endogenous quality for each dataset k, crop i, and provincial unit r as follows:
The georeferenced application rates of nitrogen, phosphorus (P₂O₂), and potassium (K₂O) for each crop at the global scale were summarized according to the algorithm presented in Figure 2. First, we decomposed the national application rates of the three fertilizers (HFUCB, FUBC18-IDV, and FUBC18-AGG) into provincial application rates and allocated them using the national and local proportions calculated using MFM (if MFM data are available). In this step, the application rate fk(n,i,j,r) for a specific fertilizer data set nk for crop i in local unit r in country j is calculated using the following formula:
where fMFM is the application rate for the respective crop and local unit in MFM, and α is a scaling factor defined as
where Fk(n,i,j) is the national fertilizer rate n for crop i and country j in dataset k, and ACR is the corresponding cropped area in CROPGRIDS. In equation (7), we assume that these coefficients α remain constant between 2000 and 2020. This is only true if we assume that such geographic differences are due primarily to agrometeorological differences rather than management practices, or if geographic differences in management practices remain similar in both time periods.
Algorithm for synthesizing a global fertilizer application map. The names of the data sets are given in Tables 1 and 2.
For each fertilizer n applied to crop i in local unit r, we checked whether the application rate was available from multiple data sets. If only one data set k was available, the selected application rate was f(n,i,r) = fk(n,i,r) (Fig. 2). If multiple data sets were available, we selected the best-fitting data set kbest, which had the highest endogenous quality Qk,i,r as defined in equation (5), such that f(n,i,r) = \({f}_{{k}_{{best}}}\)(n,i,r). If two data sets had the same Qk,i,r, the data set with the closest reference year was selected as kbest. Alternatively, if the base year of these data sets was the same, \({f}_{{k}_{{best}}}\left(n,i,r\right)\) was calculated as the average of all data sets with the same Qk,i,r and base year. If a data set was missing, we filled the gap by first checking for the availability of data for neighboring provincial units. Specifically, if fertilizer quantity n for crop i is available in w neighboring provincial units, the area-weighted average fertilizer quantity favg(n,i,w) for w neighboring provincial units is calculated as follows:
where nw is the number of grid cells of the common boundaries. If adjacent provincial units do not have a fertilizer application policy n for crop i, we estimate f(n,i,r) based on fertilizer application rates for similar crops according to three criteria defined by FAO39: (a) classification (i.e. cereals, legumes, nuts, fruits and berries, spices, perennial oilseeds, annual oilseeds, forages, fibre crops, vegetables and other perennial crops); (b) durability (i.e. temporary or permanent); and (c) stem type (i.e. herbaceous, shrubby or woody; see Supplementary Table 4). Crops i and c are considered similar if they share at least two of the above three criteria. If the fertilizer application rates n are known for a similar crop c within a national unit r (Fig. 2), we calculate f(n,i,r) = favg(n,c,r), where favg(n,c,r) is the area-weighted average fertilizer application rate for c similar crops within the national unit r, i.e.,
If there are no similar crops in subnational unit r (Figure 2), we calculate f(n,i,r) = favg(n,c,g), where favg(n,c,g) is the area-weighted average application rate c of similar crops in all subnational units g, that is,
Finally, to generate georeferenced maps of global fertilizer application rates by nutrient and crop, we uniformly distributed fertilizer application rates across provinces across the grid cells that crop i belongs to within the provincial unit using crop masks in CROPGRIDS. Figure 3 shows example application rate maps for nitrogen, phosphorus (P₂O₂), and potassium (K₂O) for cotton, along with the corresponding overall data quality and data sources used to generate the maps.
Example of cotton distribution in NPKGRIDS data. From left to right: N, P2O5, and K2O; from top to bottom: fertilizer application, data quality, and data source.
The NPKGRIDS dataset contains georeferenced maps of N, P₂O₂, and K₂O fertilizer application rates in 2020 for 173 crops (see Supplementary Table 4 for the list of crops) at 0.05° resolution (~5.6 km at the equator) with a bounding box from −180° to 180° in longitude and from −90° to 90° in latitude in the WGS-84 coordinate system. Georeferenced maps are distributed as NetCDF files with grid cells containing ocean/water marked as “-1”. The list of files in the dataset is provided in Table 4. The NPKGRIDS dataset is available for download from the figshare40 repository at https://doi.org/10.6084/m9.figshare.24616050. The application rates for phosphorus and potassium fertilizers are presented in terms of oxides. They can be converted to application rates based on elemental composition using the following formulas: 1 kg P2O5 is equivalent to 0.436 kg P, and 1 kg K2O is equivalent to 0.83 kg K.
Due to the lack of crop fertilization data other than those used in this paper, we estimated NPKGRIDS data using total national fertilization inputs of N, P₂O₂ and K₂O provided by FAOSTAT41 (160 countries) and IFA11 (110 countries) (Table 1). To do this, we first calculated the total national fertilization input M(n,j) in country j estimated by NPKGRIDS. The results are presented below:
Where ACR(p,i,j) is the harvested area of crop i in grid p for country j in CROPGRIDS26; f is the corresponding amount of fertilizer applied n in NPKGRIDS. Country boundaries were defined based on the GAUL38 dataset (level 0). We then compared M(n,j) with the corresponding fertilizer use reported by FAOSTAT and IFASTAT, MFAO and MIFA respectively (average over the period 2015–2020). These comparisons were characterized using the coefficient of determination R2 (similar to the Nash-Sutcliffe efficiency coefficient), the concordance correlation coefficient (CCC) and the normalized root mean square error (NRMSE), which are expressed as
where \({O}_{x}\) represents the logarithm of MFAO or MIFA, \(E\) represents the logarithm of the country-level application mass (M) calculated according to NPKGRIDS. \(\bar{{O}_{x}}\) and \(\bar{E}\) are the corresponding means for all countries, \({{\sigma }_{{O}_{x}}}^{2}\) and \({{\sigma }_{E}}^{2}\) are the corresponding variances, and \(\rho \) is the Pearson correlation coefficient between Ox and \(E\). Mx represents MFAO or MIFA, \({M}_{x,\max }\) and \({M}_{x,\min }\) are the corresponding maximum and minimum fertilizer masses for all countries respectively, and \({n}_{p}\) is the number of data points.
NPKGRIDS shows that global nitrogen fertilizer application was 100 million tonnes, about 10% lower than the 2020 global estimates provided by FAOSTAT and IFASTAT (110 million tonnes and 112 million tonnes, respectively). At the country level (left column of Figure 4), nitrogen fertilizer application estimated by NPKGRIDS agrees well with FAOSTAT (R2 = 0.76, CCC = 0.89 and NRMSE = 0.01) and relatively well with IFASTAT (R2 = 0.66, CCC = 0.87 and NRMSE = 0.01). When compared with FAOSTAT data, underestimation of nitrogen fertilizer application was found mainly in Africa, such as the Democratic Republic of Congo, Namibia and Madagascar. NPKGRIDS consistently overestimates nitrogen fertiliser use in Iraq, Syria and Jordan compared to FAOSTAT and IFASTAT data.
Comparison of NPKGRIDS data with FAOSTAT (top row) and IFASTAT (bottom row) data for N (left column), P₂O₂ (middle column) and K₂O (right column). Each marker in the scatterplot represents a country, and the black line indicates a 1:1 ratio.
For phosphorus, the NPKGRIDS estimate is 46 million tonnes, which is very close to the FAOSTAT and IFASTAT estimates for 2020 (48 million tonnes and 49 million tonnes, respectively). A comparison of total P2O5 use at the country level shows the strongest correlation with FAOSTAT (R2 = 0.82, CCC = 0.91, NRMSE = 0.02) and IFASTAT (R2 = 0.70, CCC = 0.88, NRMSE = 0.01, Figure 4, middle column). In general, as with nitrogen data, the differences in phosphorus input data between NPKGRIDS and FAOSTAT and IFASTAT data are more pronounced in countries in Africa and the Middle East.
For potash, global application according to NPKGRIDS is 40 million tonnes, which is in good agreement with FAOSTAT and IFASTAT estimates (39 million and 41 million tonnes, respectively). However, a comparison of total K2O applications by country shows lower agreement with FAO/IFA estimates (Figure 4, right column), with low correlation for both FAOSTAT (R2 = 0.68, CCC = 0.84, NRMSE = 0.01) and IFASTAT (R2 = 0.50, CCC = 0.77, NRMSE = 0.01). NPKGRIDS tends to overestimate potash application in North Africa and West Asia.
We obtained non-crop-specific total nitrogen, phosphorus, phosphorus and potassium (N) input data at the national and provincial levels for 37 countries and 166 provincial units from 2006 to 2020, including 32 European countries42, India43, Pakistan44, China45, Iran46 and Sri Lanka47 (Table 2, Supplementary Table 2). Of the 37 countries, 11 provided data at the provincial level and 26 only at the national level. Only five countries (99 provincial units) provided data for potash and K. We calculated the average for 2015–2020 for all NSO data except Iran, for which the latest available data are from 2006. We aggregated pixel-level data from NPKGRIDS into national and provincial annexes for total N, P₂O₂, and K₂O using Equation 11, where j represents the national (level 0) or subnational (level 1) unit defined within the GAUL38 administrative unit boundaries. The quality of the comparison of NPKGRIDS and NSO data was quantified using R2 (Equation 12), CCC (Equation 13), and NRMSE (Equation 14).
Comparison of NPKGRIDS and NSO data for N application at the national and provincial levels showed relatively good agreement: R2 = 0.80, CCC = 0.90 and NRMSE = 0.03 (Figure 5), while P2O5 and K2O estimates were weaker: R2 values were 0.74 and 0.75 from NSO data, respectively.
The NPKGRIDS system compares all national and provincial crop fertiliser application data with data from national statistical offices (NSOs). Total fertiliser applied includes (a) N, (b) P₂O₂ and (c) K₂O. Colour codes correspond to NSOs: EU (European Union), LK (Sri Lanka), PK (Pakistan), IR (Iran), IN (India) and CN (China). Black lines correspond to a 1:1 ratio.
A comparison of the 32 country data with EUROSTAT42 data shows good consistency at both national and local levels, with some exceptions. In particular, total N and P fertiliser applications were significantly underestimated in Iceland and Ireland (N and P2O5) and Malta (N). This underestimation is likely due to the high uncertainty in the area data provided by CROPGRIDS in these countries, particularly in Ireland due to uncertainty in fertiliser application to grassland. In Iceland, NPKGRIDS only mapped potatoes. In contrast, nutrient applications were slightly overestimated in China.
NPKGRIDS takes into account uncertainties and errors in input datasets, such as the original fertilizer dataset and the CROPGRIDS dataset used to construct the spatial distribution tables of fertilizer applications. Uncertainties can arise from erroneous or missing data on fertilizer application and crop area reported in national and international reports. For example, MFM faces data limitations in many low- and middle-income countries, and phosphorus and potash fertilizer data show more anomalies than nitrogen fertilizer data. CROPGRIDS, on the other hand, is constructed by harmonizing multiple data sources, including surveys, remote sensing, and models, each of which has uncertainties that are passed on when constructing NPKGRIDS.
NPKGRIDS spatially distributes national and provincial data across grid cells, which introduces additional uncertainty. For example, the spatial distribution of national data (e.g., HFUBC, FUBC18-IDV, FUBC18-AGG) assumes that the relative shares of fertilizer use within a country follow the same pattern as in the MFM (Equation 7), ignoring relative changes in planting practices that may have occurred between different provinces within a country. In addition, information obtained directly from the MFM does not account for changes in fertilizer use that may have occurred in these regions over the past 20 years.
Finally, due to spatial resolution limitations, NPKGRIDS excludes a number of small countries and territories, including the Falkland Islands, Faroe Islands, French Southern and Antarctic Territories (SAT), Hart Island, Isle of Man, Kingman Reef, Kiribati, Matansala, Mayotte, Netherlands Antilles, Palau, Réunion, Saint Pierre, South Georgia, Svalbard, and the Virgin Islands.
To quantify potential uncertainty, we calculated subnational data quality scores based on endogenous quality scores and compared them with FAOSTAT and IFASTAT data. The overall quality of NPKGRIDS data for nutrient n (i.e. N, P₂O₂ and K₂O) for crop i in subnational unit r in country j, \(Q(n,i,j,r)\), is calculated as follows:
where Qk is the endogenous quality of the selected dataset calculated using equation (5), and the base quality of QFAO and QIFA with the FAOSTAT41 and IFASTAT11 datasets is defined as
x is FAO or IFA, and Qx values range from 0 (low quality) to 1 (high quality). For subnational units where the application rate has filled a gap, we assign the corresponding Qk value to zero. Data quality maps are published with the NPKGRIDS dataset. An example of a data quality map for cotton is shown in Figure 3 (second row).
All georeferenced maps published in the NPKGRIDS40 dataset are in the standard NetCDF4 format. These files can be read and analyzed using various programming languages (e.g. MATLAB, Python, Julia, R) and software (ArcGIS, QGIS, Panoply). The NPKGRIDS dataset contains the same crops as the CROPGRIDS26 dataset and follows the naming system used by FAO28.
Post time: Jul-28-2025