International Journal of Applied Earth Observation and Geoinformation
Remote sensing net primary productivity (NPP) estimation with the aid of GIS modelled shortwave radiation (SWR) in a Southern African Savanna
Introduction
In studies of ecosystem processes and environmental management, reliable net primary productivity (NPP) measurements are crucial as NPP represents an important energy flux in ecosystems. Traditionally, NPP has been measured using biometric measurements, i.e., sample surveys and field measurements (Tao et al., 2003). Although these traditional field-based measurements have been used successfully with accurate NPP output for small scale observations (Ogawa, 1977, Li et al., 2005), they are often time consuming and laborious (Lu, 2006). The methods are also hard to extend to the estimation of NPP on large scales because of the sparse measurements network. Also the need to fell sample trees at the target research site may adversely affect the site, for example, loss of habitat, biodiversity and carbon sequestration potential (Wang et al., 2007). Therefore, it is necessary to apply alternative methods of NPP estimation to replace or supplement traditional approaches in collecting ecological data.
The development of remote sensing has by far enhanced the ability to study and understand ecosystems with improved accuracy (Lu, 2006). Remote sensing provides an invaluable opportunity to improving the estimation of NPP at landscape and regional scales in a cost effective, efficient and accurate approach (Running et al., 1999, Running, 2000) at high temporal and spatial scales (Myneni et al., 1997, Myneni et al., 1998, Tucker et al., 2001, Zhou et al., 2001, Lu, 2006). As a result, various NPP estimation models that use remote sensing data have been developed (Goetz et al., 2000). Recently, the moderate resolution imaging spectroradiometer (MODIS) NPP, based on a micrometeorological approach was developed by Rahman et al. (2004) to provide a consistent, continuous estimate of photosynthetic production (Heinsh et al., 2006) hereinafter referred to as the MODNPP model. This model is derived from the Monteith model (Monteith, 1977) which is given by the equation:where, LUE is light use efficiency and APAR is absorbed photosynthetically active radiation by vegetation. However, the model’s limitation has been on its demands for densely measured shortwave radiation (SWR) flux for APAR estimation, as well as LUE values. The LUE and SWR flux measurements have mainly been obtained from coarse resolution satellite-based weather data, making NPP studies difficult in cases where intensively measured LUE and SWR information is required (Turner et al., 2003, Rahman et al., 2004, Running et al., 2004). Scientists have therefore largely depended on calibrated LUE (Rahman et al., 2004) and SWR (Kumar et al., 1997) values whose accuracy is still largely unknown. Thus, the development of efficient and accurate methods for estimating LUE and SWR is critical.
Modifications to the MODNPP model established by Rahman et al. (2004) can however be made. For example, Kumar et al. (1997) used a digital elevation model (DEM) and geographic information system (GIS) radiation model to compute “continuous-field” SWR flux measurements which offer an alternative source of SWR. The availability of DEM data for virtually any location on earth since the launch of space-borne earth observation sensors like ASTER (Stevens et al., 2004) and the Shuttle Radar Topography Mission (SRTM) (Rabus et al., 2003) has enhanced opportunities of using fine to medium scale DEMs in ecological modelling at landscape scale (Ndaimani et al., 2012). In the present study, a considerable modification to the MODNPP model after Rahman et al. (2004) was made. SWR, a key input of NPP estimation in this model, has been estimated using GIS and DEM based SWR model contrary to previous studies which has used SWR measured from weather stations. To the best of our knowledge, we are not aware of any study that has used DEM and GIS modelled SWR to estimate NPP. Hence, testing whether SWR modelled from a DEM in a GIS can successfully be used to estimate NPP is important.
Evaluation of model performance and accuracy assessment of the estimated results are important aspects in the NPP estimation procedure (Lu, 2006). Scientists have often adopted two main methods to evaluate model performances. One is based on the coefficient of determination (R2) that is if the models were developed using multiple regression analysis and the other assesses the root-mean-squared error (RMSE) such as in Muinonen et al. (2001) and Rasib et al. (2007). In general, a high R2 or a low RMSE value often indicates a good fit between the model developed and the sample plot data. However, due to unavailability of field-measured NPP, NPP model results can be compared with outputs of other NPP-models. Lu (2006) noted that previous research on NPP estimation have failed to provide accuracy assessments due to the difficulty in collecting ground reference data or the discrepancy between field measurements and NPP estimation results. Although there are many challenges associated with obtaining ground reference data, it is important to test modelled results with results from other measurement approaches as a way of improving accuracy before the model can be adopted for further research.
In this study we tested whether and to what extent the DEM and GIS modelled SWR can be used in the MODNPP model to estimate NPP in an African Savanna landscape. To achieve this, we tested whether the NPP estimated using the modified model significantly differ with those obtained using an established NPP-Rainfall regression model developed by Lieth and Whittaker (1975). Next, we compared our model results with those found in refereed scientific literature in an African Savanna. Finally we tested whether our modelled NPP was significantly related to dry matter productivity (DMP) results from VITO (http://www.geoland2.eu/). Specifically, we tested the hypothesis that our MODNPP results do not significantly differ from NPP results of other NPP models within the same ecological landscape.
Section snippets
Study site
This study was conducted in the Savanna landscape of the Great Limpopo Transfrontier Conservation Area (GLTFCA) which lies across Zimbabwe, South Africa and Mozambique, between 30.70 °E and 35.00 °E, and 25.50 °S and 20.30 °S. The GLTFCA is a union of the Limpopo National Park (LNP) in Mozambique, Kruger National Park (KNP) in South Africa, Gonarezhou National Park (GNP), Manjinji Pan Sanctuary (MPS) and Malipati Safari Area (MSA) in Zimbabwe, as well as two areas between Kruger and Gonarezhou,
Results
The modelled SWR is significantly (p < 0.05) positively related to SWR measured from weather stations with an R2 of 0.68 (Fig. 3). In addition, a Student t-test analysis showed that measured SWR is not significantly (p > 0.05, N = 24) different from the modelled SWR.
The range of NPP values for the MODNPP and Lieth NPP are almost similar with NPP ranging from 0 to 6 ton C ha−1 year−1 whilst NPP values range up to 30 ton C ha−1 year−1 for the DMP (Fig. 4(a)–(c)).
Table 1 shows the distribution of NPP (ton C ha−1
Discussion
Our results suggest that our modified MODNPP model can be used to successfully estimate NPP in savanna ecosystems. This fills an important knowledge gap as previous studies on the estimation of savanna NPP have mainly been using biometric measurements (Miller et al., 2004). Most importantly, our results fall within the range established by a range of NPP studies in African Savannas (Table 2).
Specifically, results of the GLTFCA long-term annual MODNPP mean falls within 95% CI of 5.8 ± 1.9 ton C ha−1
Conclusion
The objective of this study was to test whether and to what extent the modified MODNPP model can successfully be used to model NPP in an African Savanna landscape. Based on the results, we can conclude that the modelled SWR can effectively be used in the MODNPP model in place of calibrated SWR weather station data. We also conclude that the modified MODNPP model can be used successfully to model NPP in African Savanna landscapes. Results of this study imply that GIS modelled SWR is a useful
Acknowledgements
This research was conducted within the RP-PCP Research Platform. We thank the French Embassy in Zimbabwe for their financial support through CIRAD (RP-PCP grant/Project ECO#1). Several persons at the University of Zimbabwe and CIRAD Zimbabwe helped in many ways during this research. Of particular mention are Dr Michel Degarine, Dr Mhosisi Masocha and Fadzai Zengeya for their technical support.
References (34)
- et al.
Forest stand characteristics estimation using a most similar neighbor approach and image spatial structure information
Remote Sensing of Environment
(2001) - et al.
The shuttle radar topography mission a new class of digital elevation models acquired by spaceborne radar
ISPRS Journal of Photogrammetry and Remote Sensing
(2003) - et al.
NASA EOS Terra ASTER: volcanic topographic mapping and capability
Remote Sensing of Environment
(2004) - et al.
Scaling Gross Primary Production (GPP) over boreal and deciduous forest landscapes in support of MODIS GPP product validation
Remote Sensing of Environment
(2003) - et al.
The State of the Environment of the Southern Africa
(1998) ArcView GIS 3.2
(1998)- et al.
Interannual variability of global terrestrial primary production: results of a model driven with satellite observations
Journal of Geophysical Research
(2000) - et al.
Productivity and carbon fluxes of tropical savannas
Journal of Biogeography
(2006) Photosynthesis and carbon balance of a Sahelian fallow savanna
Global Change Biology
(1998)- et al.
Evaluation of remote sensing based terrestrial productivity from MODIS using regional tower Eddy flux network observations
IEEE Transactions on Geoscience and Remote Sensing
(2006)
Above-ground standing-crop, protein content and dry matter digestibility of a tropical grassland range in the Nairobi National Park, Kenya
African Journal of Ecology
Modelling topographic variation in solar radiation in a GIS environment
International Journal of Geographical Information Science
Improvement, comparison, and application of field measurement methods for grassland vegetation fractional coverage
Journal of Integrative Plant Biology
Primary productivity of natural grass ecosystems of the tropics, a reappraisal
Plant and Soil
The potential and challenge of remote sensing-based biomass estimation
International Journal of Remote Sensing
Biometric and micro-meteorological measurements of tropical forest carbon balance
Ecological Applications
Cited by (28)
Assessing canopy nitrogen and carbon content in maize by canopy spectral reflectance and uninformative variable elimination
2022, Crop JournalCitation Excerpt :The highest prediction accuracy of soybean LCC was obtained with the expression (SDr − SDy)/(SDr + SDy) [9], indicating that the VIs of the red edge area with the derivative spectra are helpful for estimation of leaf carbon status. Net primary productivity (NPP) is an indicator used to evaluate vegetation ecosystems and carbon cycling [54]. Satellite remote sensing, such as by MODIS and Landsat, is often used to estimate NPP on a large scale or even globally [55,56].
Estimating aboveground net primary productivity of reforested trees in an urban landscape using biophysical variables and remotely sensed data
2022, Science of the Total EnvironmentCitation Excerpt :This is because the spatio-temporal variability of NPP across landscapes occurs at multiple scales (ranging from small-to-broad). MOD17 model permits integration of photochemical reflectance index (PRI) derived from narrow-wavebands (e.g. 530–570 nm) as a surrogate of LUE (Lin et al., 2019; Pachavo and Murwira, 2014) and aforementioned APAR. In urban landscapes, numerous ecosystem processes and production often occur at fine resolutions, thus coarse resolution remotely sensed imagery such as MODIS is not best suited for evaluating fine-scale ecological processes and impacts.
Estimating Aboveground Net Primary Production (ANPP) Using Landsat 8-Based Indices: A Case Study From Hir-Neur Rangelands, Iran
2020, Rangeland Ecology and ManagementCitation Excerpt :Carlos et al. (2014) observed a significant relationship between ANPP and their remote sensing data, which they used to create a map of ANPP of the north-central forests and rangelands of Mexico. Pachavo and Murwira (2014) also estimated ANPP from remote sensing data with the aid of GIS-modeled shortwave radiation in South African Savanna. Ariapour et al., (2015) estimated the suitability of forage production modeling using remote sensing data.
Evaluating the grassland net primary productivity of southern China from 2000 to 2011 using a new climate productivity model
2016, Journal of Integrative AgricultureEffects of atmospheric correction and pansharpening on LULC classification accuracy using WorldView-2 imagery
2015, Information Processing in AgricultureCitation Excerpt :Changes of NPP can greatly affect global carbon balance and climate change [6], which has been a key issue of ecological studies since last decades [8]. Recently, many studies have concerned prediction of regional NPP [9–16], and many studies indicated that the potential of carbon sequestration could be achieved by land management practices, such as sustainable timber production and farm afforestation [5,17–19]. Because of constant change of NPP among terrestrial ecosystems or LULC types [8], an accurate LULC map is very important to support a precise estimation of NPP or carbon sequestration.