Limits...
Assessing the Potentialities of FORMOSAT-2 Data for Water and Crop Monitoring at Small Regional Scale in South-Eastern France

View Article: PubMed Central

ABSTRACT

Water monitoring at the scale of a small agricultural region is a key point to insure a good crop development particularly in South-Eastern France, where extreme climatic conditions result in long dry periods in spring and summer with very sparse precipitation events, corresponding to a crucial period of crop development. Remote sensing with the increasing imagery resolution is a useful tool to provide information on plant water status over various temporal and spatial scales. The current study focussed on assessing the potentialities of FORMOSAT-2 data, characterized by high spatial (8m pixel) and temporal resolutions (1-3 day/time revisit), to improve crop modeling and spatial estimation of the main land properties. Thirty cloud free images were acquired from March to October 2006 over a small region called Crau-Camargue in SE France, while numerous ground measurements were performed simultaneously over various crop types. We have compared two models simulating energy transfers between soil, vegetation and atmosphere: SEBAL and PBLs. Maps of evapotranspiration were analyzed according to the agricultural practices at field scale. These practices were well identified from FORMOSAT-2 images, which provided accurate input surface parameters to the SVAT models.

No MeSH data available.


Related in: MedlinePlus

a) Rainfall and irrigation events over the study period. b) Measured vegetation height with vertical bars corresponding to the 95% confidence interval of measurements. c) Comparison between estimated and measured LAI of irrigated meadow over the study period.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3675555&req=5

f7-sensors-08-03460: a) Rainfall and irrigation events over the study period. b) Measured vegetation height with vertical bars corresponding to the 95% confidence interval of measurements. c) Comparison between estimated and measured LAI of irrigated meadow over the study period.

Mentions: For LAI estimation, the best fitted values found for NDVI∞, NDVIs, and KLAI were respectively 0.9, 0.1, and 0.70, with a RMSER of 26.8%. The results compared to ground measurements were globally satisfactory as displayed in figure 6a. Those values were very comparable to those obtained in other studies [27, 35]. However, we can notice more dispersion for irrigated meadows at high LAI values. Discrepancies between simulations and measurements for high values of LAI (>4) hare already been observed by several authors [17, 6]. In order to better understand this dispersion, we have analyzed the temporal variation of LAI. Figure 7 displays a comparison between the dynamics of estimated and measured meadow LAI, the evolution of vegetation height, hveg, and the main cultural practices performed over the study period (3 cuts and irrigation every 11 days approximately). We can first notice that the 3 cuts, represented by vertical arrows on the figure, are well identified by LAI measurements and FORMOSAT data. We saw that simulations gave better results if they were only compared to the measurements with their confidence intervals than to interpolated data. The choice of the linear interpolation of LAI appears indeed questionable. Other empirical models used classically to interpolate LAI data such as those proposed by [53] can be applied to short period for the meadow between each cut. However such empirical models do not take into account small variations due to various factors such as water stress which can affect LAI for small periods between two irrigation events.


Assessing the Potentialities of FORMOSAT-2 Data for Water and Crop Monitoring at Small Regional Scale in South-Eastern France
a) Rainfall and irrigation events over the study period. b) Measured vegetation height with vertical bars corresponding to the 95% confidence interval of measurements. c) Comparison between estimated and measured LAI of irrigated meadow over the study period.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3675555&req=5

f7-sensors-08-03460: a) Rainfall and irrigation events over the study period. b) Measured vegetation height with vertical bars corresponding to the 95% confidence interval of measurements. c) Comparison between estimated and measured LAI of irrigated meadow over the study period.
Mentions: For LAI estimation, the best fitted values found for NDVI∞, NDVIs, and KLAI were respectively 0.9, 0.1, and 0.70, with a RMSER of 26.8%. The results compared to ground measurements were globally satisfactory as displayed in figure 6a. Those values were very comparable to those obtained in other studies [27, 35]. However, we can notice more dispersion for irrigated meadows at high LAI values. Discrepancies between simulations and measurements for high values of LAI (>4) hare already been observed by several authors [17, 6]. In order to better understand this dispersion, we have analyzed the temporal variation of LAI. Figure 7 displays a comparison between the dynamics of estimated and measured meadow LAI, the evolution of vegetation height, hveg, and the main cultural practices performed over the study period (3 cuts and irrigation every 11 days approximately). We can first notice that the 3 cuts, represented by vertical arrows on the figure, are well identified by LAI measurements and FORMOSAT data. We saw that simulations gave better results if they were only compared to the measurements with their confidence intervals than to interpolated data. The choice of the linear interpolation of LAI appears indeed questionable. Other empirical models used classically to interpolate LAI data such as those proposed by [53] can be applied to short period for the meadow between each cut. However such empirical models do not take into account small variations due to various factors such as water stress which can affect LAI for small periods between two irrigation events.

View Article: PubMed Central

ABSTRACT

Water monitoring at the scale of a small agricultural region is a key point to insure a good crop development particularly in South-Eastern France, where extreme climatic conditions result in long dry periods in spring and summer with very sparse precipitation events, corresponding to a crucial period of crop development. Remote sensing with the increasing imagery resolution is a useful tool to provide information on plant water status over various temporal and spatial scales. The current study focussed on assessing the potentialities of FORMOSAT-2 data, characterized by high spatial (8m pixel) and temporal resolutions (1-3 day/time revisit), to improve crop modeling and spatial estimation of the main land properties. Thirty cloud free images were acquired from March to October 2006 over a small region called Crau-Camargue in SE France, while numerous ground measurements were performed simultaneously over various crop types. We have compared two models simulating energy transfers between soil, vegetation and atmosphere: SEBAL and PBLs. Maps of evapotranspiration were analyzed according to the agricultural practices at field scale. These practices were well identified from FORMOSAT-2 images, which provided accurate input surface parameters to the SVAT models.

No MeSH data available.


Related in: MedlinePlus