967f64a | heisterm | 18 February 2014, 12:46:19 UTC | MIN: In the global section of module bufr, we now check whether the BUFR software is operational. If it turns out that the BUFR software (mainly decbufr) is not operational on that system, module bufr will raise an ImportError. In that case, util.import_optional will raise an exception. This change was made to allow for optional import of the BUFR module. | 18 February 2014, 12:46:19 UTC |
d4165fe | heisterm | 18 February 2014, 12:05:17 UTC | MIN: The import of the bufr module is now via lazy loading in read_BUFR in module io. The global import of the bufr module in the io module was removed. | 18 February 2014, 12:05:17 UTC |
27b1ba1 | heisterm | 18 February 2014, 12:04:01 UTC | FIX: Fixed an error in the call of import_optional | 18 February 2014, 12:04:01 UTC |
bd9360f | heisterm | 18 February 2014, 11:17:34 UTC | MIN: removed import of module bufr from module __init__. | 18 February 2014, 11:17:34 UTC |
b10da93 | heisterm | 18 February 2014, 11:14:39 UTC | MIN: Using lazy loading and optional import for the speedup module in module dp. This means the dp does not try anymore to globally import speedup. Insstead, speedup is imported by util.import_optional within the function dp.unfold_phi. No change in behaviour, execpt no warning will show if you run dp but not use dp.unfold_phi. | 18 February 2014, 11:14:39 UTC |
3cd1f1c | heisterm | 18 February 2014, 11:11:38 UTC | ENH: First prototype of optional_import function in module util to facilitate lazy loading of optional modules. This porottype will be refined, soon. | 18 February 2014, 11:11:38 UTC |
dca27eb | heisterm | 18 February 2014, 11:10:37 UTC | REF: Removed all "from module import *" statements that import funcitons directly into the wradlib namespace. This requries refactoring of all code that calls wradlib funcitons directly from the wradlib namespace instead of from the corresponding modules. So code such as >>> wradlib.readDX(f) will not work anymore. Instead, you need to specifiy the module by >>> wradlib.io.readDX(f) This change is a necessary requirement to have a better namespace hygiene. | 18 February 2014, 11:10:37 UTC |
92ceb39 | heisterm | 18 February 2014, 08:18:27 UTC | MIN: Modification of aggregate_equidistant_tseries in module util. The function was changed in a way that we now check for each target time interval whether source and target time specs are consistent. If they are not, the target value for that target interval will be NaN. Bfore that change, the entire target series was set to NaN if the specs for only one target interval was inconsistent. | 18 February 2014, 08:18:27 UTC |
cad11e3 | heisterm | 18 February 2014, 07:46:04 UTC | DOC: Minor documentation edits in dp module overview. | 18 February 2014, 07:46:04 UTC |
a147b29 | heisterm | 28 January 2014, 08:07:15 UTC | DOC: Fix in dp doc. | 28 January 2014, 08:07:15 UTC |
46faf60 | heisterm | 28 January 2014, 08:05:41 UTC | DOC: Minor changes in dp documentation. | 28 January 2014, 08:05:41 UTC |
e045128 | Thomas Pfaff | 23 January 2014, 10:35:01 UTC | FIX: changed order of dimension slices. Fixes issue #16 The recipe had not been updated to reflect the change in coordinate dimension order convention for 3-D data. | 23 January 2014, 10:35:01 UTC |
1ad733f | Thomas Pfaff | 22 January 2014, 10:27:10 UTC | FIX: removed assertion to avoid errors with new DX-files. Apparently the DX-format writers at the DWD have a strange way to calculate product length. It looks like the 0x03 character used to separate the header from the data is not counted, or counted once too many or, I don't know. Fact is, you can have every combination of even and odd byte length header and even or odd byte length data section. However the data must be even length and apparently it doesn't hurt to read one byte less than implied by product-length - header-length, so that's going to be the way we move forward. | 22 January 2014, 10:27:10 UTC |
79ed503 | heisterm | 21 January 2014, 16:01:05 UTC | FIX: pia_from_kdp did not consider gate length when computing PIA from KDP. | 21 January 2014, 16:01:05 UTC |
d6fe311 | heisterm | 21 January 2014, 15:55:31 UTC | FIX: pia_from_kdp did not consider gate length when computing PIA from KDP. | 21 January 2014, 15:55:31 UTC |
28fdd38 | Stephan Jacobi | 20 January 2014, 12:12:36 UTC | FIX: in correctAttenuationConstrained2-Function, where reference-pia interpolation and pia-calculation in small sectors of incorrect beams only takes place now, if they exist indeed; - changed internal variabel name invalidbeams -> small_sectors for better understanding - minor documentation changes | 20 January 2014, 12:12:36 UTC |
b428033 | Thomas Pfaff | 20 January 2014, 11:08:50 UTC | ENH: made plot_ppi always return axes with equal aspect (avoiding radar 'eggs'). | 20 January 2014, 11:08:50 UTC |
cb6716e | Thomas Pfaff | 20 January 2014, 11:07:21 UTC | MIN: minor code beautifications. | 20 January 2014, 11:07:21 UTC |
40f6f89 | Thomas Pfaff | 16 January 2014, 08:58:53 UTC | ENH: added examples for the new plot_ppi routine | 16 January 2014, 08:58:53 UTC |
e7fbf81 | Thomas Pfaff | 16 January 2014, 08:58:53 UTC | NEW: Added function vis.ppi_plot_crosshair | 16 January 2014, 08:58:53 UTC |
57c04ae | Thomas Pfaff | 17 January 2014, 07:55:53 UTC | DOC: added some in-code documentation to vis.plot_ppi and vis.plot_rhi | 17 January 2014, 07:55:53 UTC |
8beb473 | Thomas Pfaff | 16 January 2014, 08:58:53 UTC | ENH: Added ability to georeference and project coordinates in plot_ppi. | 16 January 2014, 08:58:53 UTC |
99cc86b | Thomas Pfaff | 16 January 2014, 08:58:53 UTC | NEW: Introduced functions plot_ppi and plot_rhi. These functions plot data assumed to be in the radar polar coordinate system directly on a cartesian matplotlib axes, allowing zooming and easier additions of other data. | 16 January 2014, 08:58:53 UTC |
39c698e | Thomas Pfaff | 14 January 2014, 16:15:59 UTC | Merged in kai_muehlbauer/wradlib_miub (pull request #13) Minor changes in cg_plot and GAMIC hdf reader | 14 January 2014, 16:15:59 UTC |
6b77d61 | heisterm | 13 January 2014, 13:52:44 UTC | Merged in thpfaff/wradlib (pull request #12) ENH: Introduced polar2latlonalt_n and helper functions. | 13 January 2014, 13:52:44 UTC |
956ba1f | Thomas Pfaff | 10 January 2014, 10:14:52 UTC | FIX: changes to readDX, fixes issue #4 - fully interpret and return the header information - read only the amount of data indicated by the header regardless of file size - remove static truncation to 128 range bins | 10 January 2014, 10:14:52 UTC |
0d5c87d | Thomas Pfaff | 07 January 2014, 11:23:13 UTC | FIX: changed import datatype in read_radolan_composite to uint16 and added type cast in order to avoid problems similar to those reading RX-data | 07 January 2014, 11:23:13 UTC |
a8f45c5 | heisterm | 20 December 2013, 14:59:35 UTC | MIN: removed rho as argument from process_raw_phidp_vulpiani because it is not used anymore. | 20 December 2013, 14:59:35 UTC |
dc12d08 | Kai Mühlbauer | 20 December 2013, 11:10:27 UTC | wradlib/wradlib mit default zusammengeführt | 20 December 2013, 11:10:27 UTC |
889dfef | heisterm | 20 December 2013, 10:08:48 UTC | FIX: Fixed an error related to the window length in the finite difference scheme of kdp_from_phidp_finitediff. | 20 December 2013, 10:08:48 UTC |
2dbe300 | Kai Mühlbauer | 20 December 2013, 07:01:11 UTC | wradlib/wradlib mit default zusammengeführt | 20 December 2013, 07:01:11 UTC |
15c9c39 | heisterm | 19 December 2013, 13:37:17 UTC | FIX: Removed several bugs from new function process_raw_phidp_vulpiani in module dp. | 19 December 2013, 13:37:17 UTC |
cb62dcd | heisterm | 19 December 2013, 12:04:50 UTC | REF: A lot of formal changes in the dp module including renaming of functions and an enhancement of process_raw_phidp2. Providing more informative function names kdp_from_phidp --> kdp_from_phidp_finitediff kdp_from_phidp2 --> kdp_from_phidp_linregress kdp_from_phidp3 --> kdp_from_phidp_sobel kdp_from_phidp4 --> kdp_from_phidp_convolution process_raw_phidp2 --> process_raw_phidp_vulpiani process_raw_phidp_vulpiani now also does the iterative phdp kdp estimation. The number of iterations is given by argument niter. The function now also returns the kdp array that corresponds to phidp. | 19 December 2013, 12:04:50 UTC |
41b9ba2 | heisterm | 18 December 2013, 16:41:34 UTC | ENH: Added function aggregate_equidistant_tseries to util module. This function aggregates time series data under the assumption that the source and the target time series are equidistant, and that the source time steps regularly fit into the target time steps (no shifts at the boundaries). This is the most trivial aggregation scenario. However, the function allows for gaps in the source data. This means, we assume the data to be equidistant (each item represents a time step with the same length), but it does not need to be contiguous. NaN values in the source data are allowed, too. The output, though, will be a contiguous time series. This series will have NaN values for those target time steps which were not sufficiently supported by source data. The decision whether the support was "sufficient" is based on the argument *minpercvalid*. This argument specifies the minimum percentage of valid source time steps inside one target time step (valid meaning not NaN and not missing) which is required to compute an aggregate. The default value of minpercvalid is 100 percent. This means no gaps are allowed, and a target time step will be NaN if only one source time step is missing. Aggregation methods at the moment are "sum" and "mean" of the source data. | 18 December 2013, 16:41:34 UTC |
f0b75f3 | Thomas Pfaff | 18 December 2013, 15:22:42 UTC | ENH: Added function timestamp2index, which allows to calculate indices of an equidistant time series from time information (human readable strings or python datetime/timedelta objects). ... and helper functions. | 18 December 2013, 15:22:42 UTC |
2274741 | Kai Mühlbauer | 18 December 2013, 08:17:57 UTC | wradlib/wradlib mit default zusammengeführt | 18 December 2013, 08:17:57 UTC |
52e5e57 | heisterm | 18 December 2013, 08:11:21 UTC | MIN: Passing the window size for KDP retrieval consistently as a kwarg in process_raw_phidp2. | 18 December 2013, 08:11:21 UTC |
ffb3d93 | heisterm | 18 December 2013, 08:07:43 UTC | MIN: Some minor changes in ErrorMetrics. Including that (i) pbias is now computed in percent as it should, (ii) pbias is reported by default, (iii) avoiding reports and plot attempts of no valid pairs of obs and est are available. | 18 December 2013, 08:07:43 UTC |
97e8eac | heisterm | 18 December 2013, 08:05:22 UTC | MIN: Removed a minor TODO warning that is always and annoyingly printed to the screen. | 18 December 2013, 08:05:22 UTC |
9ff76f4 | Kai Muehlbauer | 18 December 2013, 07:53:05 UTC | MIN: cg_plot.colorbar routine added check for integer/real numbers for proper display of ticklabels | 18 December 2013, 07:53:05 UTC |
29e67f0 | Kai Mühlbauer | 18 December 2013, 07:25:10 UTC | wradlib/wradlib mit default zusammengeführt | 18 December 2013, 07:25:10 UTC |
0b8893e | Thomas Pfaff | 18 December 2013, 07:23:11 UTC | Merged in kai_muehlbauer/wradlib_miub (pull request #11) #11: Improvement io.py - read generic hdf5 | 18 December 2013, 07:23:11 UTC |
9814d26 | Thomas Pfaff | 17 December 2013, 15:54:22 UTC | MIN: one last cleanup. | 17 December 2013, 15:54:22 UTC |
16ea87c | Thomas Pfaff | 17 December 2013, 15:53:59 UTC | FIX: made read_RADOLAN_composite return its data properly multiplied by the precision factor found in the header. This is a major change in behavior! Even though it may be considered as a bug fix it will break code for most users who worked with this function (and around the bug) so far. | 17 December 2013, 15:53:59 UTC |
6445ce3 | Thomas Pfaff | 17 December 2013, 15:50:30 UTC | FIX: made read_RADOLAN_composite close its opened file. | 17 December 2013, 15:50:30 UTC |
77a5c85 | Thomas Pfaff | 17 December 2013, 15:49:05 UTC | ENH: introduced missing keyword to read_RADOLAN_composite. To avoid hardcoded values as much as possible | 17 December 2013, 15:49:05 UTC |
5ca2e4f | Thomas Pfaff | 17 December 2013, 15:47:16 UTC | ENH: made parsing of header info more independent from individual locations (as recommended in the latest RADOLAN composite format documentation [v2.2.2]) | 17 December 2013, 15:47:16 UTC |
391f91f | Thomas Pfaff | 17 December 2013, 15:45:04 UTC | MIN: Some code beautifications | 17 December 2013, 15:45:04 UTC |
bbc9618 | Thomas Pfaff | 17 December 2013, 15:44:08 UTC | ENH: changed print statements inside read_RADOLAN_composite to warnings | 17 December 2013, 15:44:08 UTC |
0f7c8b5 | Thomas Pfaff | 16 December 2013, 15:24:10 UTC | FIX: added radar height to final altitude | 16 December 2013, 15:24:10 UTC |
6061b34 | Thomas Pfaff | 16 December 2013, 15:09:43 UTC | DOC: Updated doctests for polar2latlonalt and polar2latlonalt_n | 16 December 2013, 15:09:43 UTC |
ad29cd6 | Thomas Pfaff | 16 December 2013, 15:08:46 UTC | FIX: hammered out one big bug | 16 December 2013, 15:08:46 UTC |
40420b6 | Thomas Pfaff | 16 December 2013, 15:07:54 UTC | ENH: made call signature of polar2latlonalt_n similar to that of polar2latlonalt | 16 December 2013, 15:07:54 UTC |
c752228 | Thomas Pfaff | 16 December 2013, 13:47:02 UTC | FIX: moved conversion of angles from degrees to radians, so that behavior is more intuitive (working with degrees on the user side, and thus actually following the documentation). | 16 December 2013, 13:47:02 UTC |
da9a659 | Thomas Pfaff | 13 December 2013, 15:02:41 UTC | ENH: Introduced polar2latlonalt_n and helper functions. calculates geographical (sphere based) coordinates from radar coordinates, taking both earth's curvature and the refraction of radar beams in the air into account. | 13 December 2013, 15:02:41 UTC |
dcf1082 | Kai Mühlbauer | 11 December 2013, 10:07:25 UTC | wradlib/wradlib mit default zusammengeführt | 11 December 2013, 10:07:25 UTC |
36cf3a1 | Thomas Pfaff | 09 December 2013, 08:04:15 UTC | FIX: added conversion to 32-bit int for RX-data, because otherwise the data type of the array would not be sufficient to store -9999 nodata values. | 09 December 2013, 08:04:15 UTC |
a5a050e | Thomas Pfaff | 09 December 2013, 08:02:09 UTC | DOC: minor fixes | 09 December 2013, 08:02:09 UTC |
b504b44 | heisterm | 06 December 2013, 16:59:49 UTC | FIX: Fixed wrong datetime interpretation in RADOLAN composite and added new 150 km radius. | 06 December 2013, 16:59:49 UTC |
4801304 | heisterm | 18 November 2013, 15:59:53 UTC | MIN: Trying to make MFB adjustment more robust. Before an adjusment facator is computed, we fit a linear regression and evaluate slope, p-value and correlation. We compute an adjustment factor only if a strong relationship is indicated (slope > 0.1, r > 0.5, p < 0.01). These thresholds might later be made available as function arguments. | 18 November 2013, 15:59:53 UTC |
a9781f7 | heisterm | 18 November 2013, 15:57:26 UTC | MIN: removed the range_lim argument from read_EDGE_netcdf. Limiting the range is something tpo be done outsode the reading function. Now, the entire range is returned. | 18 November 2013, 15:57:26 UTC |
e094a2e | heisterm | 12 November 2013, 14:43:29 UTC | MIN: Introducing exception handling for reading input files in read_EDGE_netcdf. The aim is to at least close files if an error occus during reading, using try except finally. This might be a model for all input functions. The exception will be raised (or can be caught outside the function), but the file will be closed if it was successfully opened. | 12 November 2013, 14:43:29 UTC |
9844826 | heisterm | 12 November 2013, 10:28:57 UTC | FIX: Passing range length to sub-functions within process_raw_phidp2. | 12 November 2013, 10:28:57 UTC |
a949455 | Kai Mühlbauer | 08 November 2013, 07:01:24 UTC | wradlib/wradlib mit default zusammengeführt | 08 November 2013, 07:01:24 UTC |
d1cc60d | heisterm | 07 November 2013, 15:42:25 UTC | ENH: Added new functions for PhiDP reconstruction based on Vulpiani (2012). The functions involve unfold_phidp_2 and a new convenience function process_raw_phidp2. Once we know more about the behaviour of these function, it will be time to clean up the dp module a little. | 07 November 2013, 15:42:25 UTC |
74d7979 | heisterm | 07 November 2013, 14:30:38 UTC | FIX: adding dr to argument list of kdp_from_phidp. | 07 November 2013, 14:30:38 UTC |
1df5c17 | Kai Mühlbauer | 07 November 2013, 12:58:11 UTC | wradlib/wradlib mit default zusammengeführt | 07 November 2013, 12:58:11 UTC |
0ff9297 | heisterm | 07 November 2013, 12:48:15 UTC | ENH: Added fourth experimental Kdp retrieval based on convolution filter suggested by Kai Mühlbauer. The function is named kdp_from_phidp4. It also takes care of remaining NaN values. If phidp does not contain NaNs, only the convolution filter is applied. | 07 November 2013, 12:48:15 UTC |
b1e8d66 | Kai Mühlbauer | 07 November 2013, 10:16:16 UTC | wradlib/wradlib mit default zusammengeführt | 07 November 2013, 10:16:16 UTC |
989640f | heisterm | 07 November 2013, 10:12:41 UTC | FIX: Fixing bug in handling the end of the beam in kdp_from_phidp2. Also inserted the explicit handling of start and end of the beam to kdp_from_phidp3. | 07 November 2013, 10:12:41 UTC |
d123bd8 | heisterm | 07 November 2013, 09:33:27 UTC | FIX: Accounting for forward/backward propagation effects in kdp_from_phidp2 and kdp_from_phidp3. | 07 November 2013, 09:33:27 UTC |
e71c619 | Kai Mühlbauer | 07 November 2013, 09:09:49 UTC | wradlib/wradlib mit default zusammengeführt | 07 November 2013, 09:09:49 UTC |
b408ddb | heisterm | 07 November 2013, 08:58:52 UTC | ENH: New experimental Kdp retrieval function kdp_from_phidp3. This function should be robust AND fast. For contiguous valid regions, it applies a sobel filter, for remaining NaN regions with valid neighbours it performs a local linear regression. Beyond, all the kdp_from_phidp functions now have an argument dr which specifies the gate length (in km). | 07 November 2013, 08:58:52 UTC |
980a7c9 | heisterm | 06 November 2013, 08:43:04 UTC | MIN: kdp_from_phidp2 now explicitely handles the start and the end of each beam. For the start and the end of the beam, we do not use a "central window" around a gate, but we use a forward window for the first L/2 gates and a backward window for the last L/2 gates. | 06 November 2013, 08:43:04 UTC |
8b0db86 | Kai Mühlbauer | 05 November 2013, 09:11:06 UTC | wradlib/wradlib mit default zusammengeführt | 05 November 2013, 09:11:06 UTC |
e3a9fca | heisterm | 05 November 2013, 08:50:37 UTC | ENH: Adding new experimental function for Kdp retrieval (kdp_from_phidp2). This function uses a moving window linear regression and should also work for data with gaps. STATUS is experimental. | 05 November 2013, 08:50:37 UTC |
f2ff0ab | heisterm | 05 November 2013, 08:48:16 UTC | FIX: Adapting VPR functions to new z-y-x order of 3-D arrays | 05 November 2013, 08:48:16 UTC |
dcdbbc8 | Stephan Jacobi | 29 October 2013, 16:33:38 UTC | Added correctAttenuationConstrained2 as an attenuation correction very close to Kraemer algorithm. | 29 October 2013, 16:33:38 UTC |
625072d | Stephan Jacobi | 29 October 2013, 16:30:46 UTC | Added correctAttenuationConstrained2 as an attenuation correction very close to Kraemer algorithm. | 29 October 2013, 16:30:46 UTC |
0be1ba4 | heisterm | 25 October 2013, 14:17:25 UTC | ENH: Added percent bias (pbias) as verificaiton measure to ErrorMetrics. | 25 October 2013, 14:17:25 UTC |
1520e02 | heisterm | 25 October 2013, 13:39:34 UTC | MIN: Provides the option to pass parameters accepted by the functions to process_raw_phidp. This change was necessary so the user is not forced to use the default parameters of all the functions that are used in the convenience funciton process_raw_phidp. | 25 October 2013, 13:39:34 UTC |
6e27419 | heisterm | 25 October 2013, 13:29:40 UTC | MIN: Changing the default value of the margin argument of fill_phidp to 3. | 25 October 2013, 13:29:40 UTC |
6c04945 | heisterm | 25 October 2013, 13:06:05 UTC | MIN: Added the requirement to MFB that there should be at least "mingage" number of valid raw and obs pairs. Also added a try/except statement to the linear regression in case line fitting fails. | 25 October 2013, 13:06:05 UTC |
4ad7c01 | heisterm | 25 October 2013, 11:57:29 UTC | DOC: Exposing the class adjust.GageOnly in the documentation. | 25 October 2013, 11:57:29 UTC |
c7b1d09 | heisterm | 25 October 2013, 11:32:23 UTC | ENH: Making Mean Field Bias Adjustment (MFB) more robust by using linear regression. If biasby argument to MFB.__call__ was "mean", the bias was computed by simply computing the mean of the ratios between obs and est. Now a new argument value is introduced which is "linregr" (now the default value). In that case, the bias is computed from the inverse slope of a linear regression forced through (0,0). | 25 October 2013, 11:32:23 UTC |
f9a7594 | Kai Muehlbauer | 22 October 2013, 13:20:05 UTC | merged with default | 22 October 2013, 13:20:05 UTC |
6e0b72c | Kai Muehlbauer | 22 October 2013, 13:09:50 UTC | MIN: Changes in GAMIC hdf reader. Changes include corrections for azimuthal and elevation alignment of data and missing dynamic range attributes | 22 October 2013, 13:09:50 UTC |
9e39fa3 | Kai Mühlbauer | 22 October 2013, 13:02:08 UTC | wradlib/wradlib mit default zusammengeführt | 22 October 2013, 13:02:08 UTC |
3ad6a90 | heisterm | 22 October 2013, 11:30:41 UTC | DOC: Corrected typos with respect to freqencies of typical radar bandwidths. | 22 October 2013, 11:30:41 UTC |
885c336 | Kai Mühlbauer | 22 October 2013, 06:52:25 UTC | wradlib/wradlib mit default zusammengeführt | 22 October 2013, 06:52:25 UTC |
c3d9374 | Kai Muehlbauer | 22 October 2013, 06:42:53 UTC | ENH: Added funtion read_generic_hdf5. As discussed in Ticket #11, adding support for reading all types of hdf5 files. | 22 October 2013, 06:42:53 UTC |
e2f6e5f | heisterm | 21 October 2013, 09:31:22 UTC | MIN: Changes on linear_despeckle and fill_phidp. For linear_despeckle, the first gate is set to nan if the second is nan. For fill_phidp, the computation has been made more robust for the margins of the arrays. | 21 October 2013, 09:31:22 UTC |
05d08bf | heisterm | 17 October 2013, 13:20:39 UTC | FIX: Made sure that the setup_distribute module is packaged for PyPI. | 17 October 2013, 13:20:39 UTC |
84ef25e | heisterm | 16 October 2013, 14:33:15 UTC | MIN: Using the median instead of the mean to compute the value at the margins surronding a NaN region in function fill_phidp. This way, local jumps in PhiDP do not affect the local fill value too much. | 16 October 2013, 14:33:15 UTC |
b0e8b5b | heisterm | 16 October 2013, 11:16:14 UTC | MIN: Removing some redundant ines from fill_phidp. | 16 October 2013, 11:16:14 UTC |
76e80eb | heisterm | 16 October 2013, 11:02:05 UTC | ENH: Changed the way PhiDP values are filled in function fill_phidp in order to avoid Kdp artefacts. Citing from the new function docstring: Contiguous NaN regions are filled by the average of the values of the margins that surround the NaN region. At the left and right margins of the array, these averages are extrapolated to the end. Using the average over the margins takes into account noisy PhiDP. As a consequence, a contiguous region of missing PhiDP will be filled by constant values determined by the edges of that region. Thus, the derivative (i. e. Kdp) in that region will be zero. This bahaviour is more desirable than the behaviour produced by linear interpolation because this will cause arbitrary Kdp values in case of noisy PhiDP profiles with large portions of missing data. | 16 October 2013, 11:02:05 UTC |
aed2e36 | heisterm | 16 October 2013, 08:36:38 UTC | FIX: Removed error in addressing the variable kdp in function kdp2r. | 16 October 2013, 08:36:38 UTC |
6715625 | heisterm | 30 September 2013, 10:07:30 UTC | DOC: Enhancing the documentation of the kdp_from_phidp function. | 30 September 2013, 10:07:30 UTC |
8e20438 | heisterm | 30 September 2013, 09:50:26 UTC | MIN: Making the use of "mean" the default to compute MFB, and "median" optional. Using AdjustMFB was using the median to compute the mean field bias in order to make the correction more robust against outliers. In order to account for the name (MEAN field bias), the behaviour was changed. Now, the field bias is, by default, computed using the mean of all obs/est ratios. By using the argument "biasby" when calling the AdjustMFB object, you can decide to use the median (biasby="median") instead. | 30 September 2013, 09:50:26 UTC |
701a070 | heisterm | 30 September 2013, 09:44:44 UTC | MIN: Considering the possibility that ErrorMetrics does not receive any valid obs/est pair. | 30 September 2013, 09:44:44 UTC |